ARK's Big Ideas 2025: Unlocking Exponential Growth

ARK's Big Ideas 2025: Unlocking Exponential Growth

• 1. FEBRUARY 04, 2025 RESEARCH REPORT FOR INFORMATIONAL PURPOSES ONLY WWW.ARK-INVEST.COM ARK Investment Management LLC. This is not a recommendation in relation to any named particular securities/cryptocurrencies and no warranty or guarantee is provided. Any references to particular securities/cryptocurrencies are for illustrative purposes only. There is no assurance that the Adviser will make any investments with the same or similar characteristics as any investment presented. The reader should not assume that an investment identified was or will be profitable. PAST PERFORMANCE IS NOT INDICATIVE OF FUTURE PERFORMANCE, FUTURE RETURNS ARE NOT GUARANTEED. BIG IDEAS 2025

• 2. Risks Of Investing In Innovation Please note: Companies that ARK believes are capitalizing on disruptive innovation and developing technologies to displace older technologies or create new markets may not in fact do so. ARK aims to educate investors and seeks to size the potential investment opportunity, noting that risks and uncertainties may impact our projections and research models. Investors should use the content presented for informational purposes only, and be aware of market risk, disruptive innovation risk, regulatory risk, and risks related to certain innovation areas. Please read risk disclosure carefully. DISRUPTIVE INNOVATION RAPID PACE OF CHANGE UNCERTAINTY AND UNKNOWNS EXPOSURE ACROSS SECTORS AND MARKET CAP RISK OF INVESTING IN INNOVATION REGULATORY HURDLES COMPETITIVE LANDSCAPE POLITICAL OR LEGAL PRESSURE Sources: ARK Investment Management LLC, 2023. à Aim to understand the regulatory, market, sector, and company risks. (See Disclosure Page) à Aim for a cross-sector understanding of technology and combine top-down and bottom-up research. ARK’s Big Ideas 2025

• 3. ARK’s Big Ideas 2025 Table Of Contents Introduction Convergence AI Agents Bitcoin Stablecoins Scaling Blockchains Robotaxis Autonomous Logistics Energy Robotics Reusable Rockets Multiomics 04 05 15 29 41 54 68 84 93 103 114 125

• 4. ARK’s Big Ideas 2025 BIG IDEAS 2025 4 As we stand on the cusp of a new era of unprecedented growth, ARK Invest’s “Big Ideas 2025” illuminates the complex convergences between and among the five technologically enabled innovation platforms evolving today: Artificial Intelligence, Robotics, Energy Storage, Public Blockchains, and Multiomic Sequencing. These platforms should drive exponential advances across industries and catalyze a step change in global economic growth. Presented in this year’s report are 11 Big Ideas illustrating the massive transformations occurring today. Our research suggests that these Big Ideas are poised to boost productivity dramatically, to revolutionize industries, and to create long-term investment opportunities. Join our journey into the future of innovation and its profound implications for investors, businesses, and society at large. Welcome to ARK’s Big Ideas 2025. Unlocking Exponential Growth Through Disruptive Innovation

• 5. The Second Half Of The Chessboard Convergence Among Technology Platforms Leading To Significant Acceleration In Macroeconomic Growth Brett Winton CHIEF FUTURIST ARK’s Big Ideas 2025 5

• 6. Five Innovation Platforms Are Catalyzing Accelerated Growth Public Blockchains Upon large-scale adoption, all money and contracts will likely migrate onto Public Blockchains that enable and verify digital scarcity and proof of ownership. The financial ecosystem is likely to reconfigure to accommodate the rise of Cryptocurrencies and Smart Contracts. These technologies increase transparency, reduce the influence of capital and regulatory controls, and collapse the costs of contract execution. In such a world, Digital Wallets will become increasingly necessary as more assets become money-like and corporations and consumers adapt to the new financial infrastructure. Corporate structures may be called into question. Multiomics The cost to gather, sequence, and understand digital biological data is falling precipitously. Multiomic Technologies provide research scientists, therapeutic organizations, and health platforms with unprecedented access to DNA, RNA, protein, and digital health data. Cancer care should transform with pan-cancer blood tests. Fed by rich multiomic data and powered by Programmable Biology, AI systems running autonomous labs could collapse the cost of drug discovery, development, and trial, transforming returns in a sector that has stagnated. Biological discoveries should power novel Precision Therapies that target and cure rare diseases and chronic conditions, unlocking profound economics. Over time, the design and synthesis of novel biological constructs will yield advances in agriculture, material science, and even computation. AI Computational systems and software that evolve with data can solve intractable problems, automate knowledge work, and accelerate technology’s integration into every economic sector. The adoption of Neural Networks should prove more momentous than electrification and potentially create tens of trillion dollars of value. At scale, these systems will require unprecedented computational resources, and AI-specific compute hardware should dominate the Next Gen Cloud datacenters that train and operate AI models. The potential for end-users is clear: a constellation of AI-driven Intelligent Devices that permeate people's lives, changing the way that they spend, work, and play. The adoption of artificial intelligence should transform every sector, impact every business, and catalyze every innovation platform. Robotics Catalyzed by artificial intelligence, Humanoid Robots should operate alongside humans and navigate legacy infrastructure, changing the way products are made and sold, and eventually the way we live our lives at home. 3D Printing should contribute to the digitization of manufacturing, increasing not only the performance and precision of end-use parts, but also the resilience of supply chains. Meanwhile, the world’s fastest robots, Reusable Rockets, should continue to reduce the cost of launching satellite constellations and enable uninterruptible connectivity and earth observation. A nascent innovation platform, robotics could collapse the cost of transporting across distance, with hypersonic travel, the cost of manufacturing complexity with 3D printers, and the cost of physical work with AI-guided robots. Energy Storage The declining costs of Advanced Battery Technology should cause an explosion in form factors, enabling Autonomous Mobility systems that collapse the cost of transportation. Electric drivetrain cost declines should unlock micro-mobility and aerial systems, including flying taxis, enabling business models that transform cities. Autonomy should reduce the cost of taxi, delivery, and surveillance by an order of magnitude, enabling frictionless transport that will increase the velocity of ecommerce and make individual car ownership the exception rather than the rule. These innovations combined with large-scale stationary batteries and Distributed Energy Generation, notably solar and small-scale fission, should cause a transformation in energy, substituting electricity for liquid fuel and increasing system-wide resilience, reliability, and production. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Convergence 6

• 7. Row 2X Wheat Grains Value Year Computers Crossed The Same Compounding Threshold 8 128 2 teaspoons of wheat 1975 16 65 thousand 9 loaves of bread 1983 24 17 million 4 year’s sustenance 1998 32 4 billion 7 pounds of gold 2008 40 1 trillion 1 ton of gold 2018 48 300 trillion 20% of India GDP in 600 AD 2023 56 70 quadrillion 50x Global GDP in 600 AD 2027e 64 18 quintillion 9x Global GDP in 2024 2030e By lore, when he agreed to the reward, the Indian ruler thought it quite reasonable. By the 6th row of the chessboard, however, he had exhausted India’s treasury, having paid just ~.001% of the bill. Today, computational advance has completed the equivalent of the 6th row on the chessboard. Entering the AI cycle in 2018, computers had crossed 40 performance doublings and in 2023 surpassed 48. By the end of this decade, thanks to the acceleration in AI, computation could reach the end of the chessboard. In Compounding, The Real Action Takes Place On The Second Half Of The Chessboard I don’t require much, your highness, just a grain of wheat on the first square of the board, doubling consecutively on each square thereafter. Legendarily requesting his reward for inventing chess, the 6th century advisor to the Indian ruler: Note: “GDP”: Gross Domestic Product, aggregate global production for the global datapoints. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, Kurzweil 2006 as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Convergence 7

• 8. 1975 8 performance doublings 1983 16 performance doublings 1998 24 performance doublings 2008 Halfway through the chessboard Google announces the TPU 2023 48 performance doublings 2027 56 performance doublings 2030 End of the chessboard 2018 40 performance doublings 2028 48 performance doublings 1E-02 1E+00 1E+02 1E+04 1E+06 1E+08 1E+10 1E+12 1E+14 1E+16 1E+18 Compute Capability (Per 2024 $) Cumulative Compute Purchased (Unit Of xAI Colossus-Sized Datacenters) Integrated Circuit Performance Improvement Curve 100 Billionth 1 Billionth 100 Millionth 1 Millionth 10 Thousandth 1 Hundredth 1 100 10,000 1,000,000 AI Is Accelerating The Performance Doubling Rate, Closing In On The End Of The Chessboard Enabled primarily by architectural improvements in AI systems, performance per dollar of AI compute is expected to improve >1000x by 2030. At that time, we expect that compute performance will have doubled 64 times since the advent of the integrated circuit. Improvements in AI Architecture Driving Compute Performance Compute Hardware Driving Performance Improvements Each Circle Represents 1 Performance Doubling Hollow Circles Are Forecasts ARK’s Big Ideas 2025: Convergence 8 Note: The “Compute Capability per $2024 Dollar” displayed on the Y axis above is a single metric that measures 1000s of computations per second purchasable for $1. For AI compute performance, computational impact is adjusted upwards by the rate of AI architectural performance per dollar improvements according to ARK’s research. It's a very large number, so we use scientific notation, e.g., "1E+21," meaning 1 followed by 21 zeros. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Kurzweil 2005 and Jurvetson 2024 as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon.

• 9. 2023 2024 0 50 100 150 200 250 0 500 1000 1500 # Of Hypotheses Tested Per FTE Researcher H100 Equivalents Compute In Recursion’s Datacenter Recursion Drug Development Efficiency 10/1/23 4/18/24 11/4/24 5/23/25 Log Error Rate Scale SWE-Bench Verified Best In Class AI Software Engineering AI Advances Should Unlock Massive Market Opportunities 72% 54% As AI continues to accelerate, robotaxis should proliferate, drug development timelines and costs should collapse, and AI agents should solve software engineering challenges autonomously, monitoring and modifying systems around the clock. 4% 22% 38% 45% FSD 11.5 12/31/23 FSD 12.5 10/24/24 FSD 13.4 1Q25* FSD 13.6 2H25* 1 10 100 1000 10000 100000 1000000 - 100,000 200,000 Miles Per Intervention H100 Equivalent Compute In Tesla’s Datacenter Tesla FSD Performance Waymo intervention rate post commercialization Human driver accident rate ARK’s Big Ideas 2025: Convergence 9 *Estimates. Note: “FSD”: Full Self-Driving. SWE-bench is a dataset that tests systems' ability to solve GitHub issues automatically. The dataset collects 2,294 Issue-Pull Request pairs from 12 popular Python repositories. Evaluation is performed by unit test verification using post-PR behavior as the reference solution. See Jiminez et al. 2024. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Tesla, Recursion, and SWEbench.com, and OpenAI as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon.

• 10. Convergence Is Accelerating The Technology Revolution ARK measures the degree to which technologies serve as catalysts between and among innovation platforms. The convergences among them are increasing, with network density up 30% in the past year. AI is proving more critical to unlocking the value of Precision Therapies and Multiomic Technologies. Smart Contract ecosystems are serving as test beds in which autonomous AI agents can be renumerated for sharpening their capabilities. Next Gen Cloud energy demand is pulling forward timelines for Distributed Energy Generation. Catalyzed Increase In Market Opportunity Order Of Magnitude+ Multiples Very Meaningful Meaningful Potential ARK’s Big Ideas 2025: Convergence 10 Note: “Network density” measures the degree of interconnectedness between nodes relative to the maximum potential interconnectedness. In our research, if every technology were expected to catalyze another technology to increase in value by an order of magnitude or more, that would equate to a fully interconnected network. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon.

• 11. The Acceleration In Neural Networks Is Accelerating Every Other Disruptive Technology Cryptocurrencies Smart Contracts Digital Wallets Precision Therapies Multiomic Technology Programmable Biology Neural Networks Next Gen Cloud Intelligent Devices Autonomous Mobility Advanced Battery Technology Reusable Rockets Humanoid Robotics 3D Printing Distributed Energy Generation CryptocurrenciesSmart Contracts Digital Wallets Precision Therapies Multiomic Technology Neural Networks Next Gen Cloud Intelligent Devices Autonomous Mobility Advanced Battery Technology Reusable Rockets Humanoid Robotics 3D Printing Programmable Biology Distributed Energy Generation Technology Catalyzing Technology Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Convergence 11

• 12. Advances In AI, Energy Storage, And Public Blockchains Are Critical To The Pace Of Technological Advance Technology Importance As A Catalyst Advances another technology by 1 order of magnitude Neural Networks Intelligent Devices Next Gen Cloud Advanced Battery Technology Digital Wallets Reusable Rockets 3D Printing Cryptocurrencies Smart Contracts Autonomous Mobility Distributed Energy Generation Multiomic Technologies Precision Therapies Programmable Biology Humanoid Robots Among disruptive technologies, Neural Networks are the most important catalyst. According to our research, advances in neural networks will increase the value of six of the other 14 technologies by at least an order of magnitude, creating massive market expansions for Next Gen Cloud, Intelligent Devices, Autonomous Mobility, Humanoid Robots, Precision Medicine, and Multiomic Technology. The AI-led revolution in technology is likely to lead to dramatic productivity gains and a stepfunction increase in economic growth. Technology Platform AI Energy Storage Multiomics Robotics Public Blockchains SCALE Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Convergence 12

• 13. Technology Inflection Should Lead To GDP Inflection Structural changes in the underlying rate of macroeconomic growth are the historical rule, not the exception. After 100,000 years of economic stagnation leading to year 1, innovations, notably writing, allowed empires to stitch together continents, which quadrupled real growth through 1000 AD. Agricultural innovations enabled denser populations and labor specialization, leading to a doubling of growth through 1500 to 0.3% per annum. During the 400 years through 1900, annual GDP growth doubled again to 0.6% as the Enlightenment and Industrial Revolution swept the globe. The Second Industrial Revolution—marked by electrification, the automobile, and telephony—ushered in the modern age and quintupled growth to 3% on average over the last 125 years. Technological breakthroughs in AI, Autonomous Mobility, and Humanoid Robots are likely to turbocharge productivity again, causing another step function in growth over the next 5 to 10 years. 0.0% 0.1% 1.0% 10.0% 100.0% Compounded Annual Growth (Log Scale) Global Real GDP Growth Log Years Until 2050 100,000 BC 1 1000 1500 1900 2024 .037% 0.14% 0.3% 0.6% 3% IMF Forecast 3.1% ARK Forecast 7.3% 2030 ARK’s Big Ideas 2025: Convergence 13 Note: “Real GDP”: Aggregate Global Production adjusted for changes in price levels. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources including DeLong 1998 and Open Philanthropy as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon.

• 14. - 50 100 150 200 250 300 $ Trillions (Nominal) Innovation Should Drive A Momentous Business Cycle Disruptive Innovation Could Grow To Dominate Markets 2014 2019 2024 2030e Disruptive innovation could command more than two-thirds of the global equity market, compounding at a 38% rate through 2030. Within innovation, market value could broaden out beyond the so-called Mag 6. Non-innovation businesses could begin to shrink as technological deflation associated with innovation threatens margins and competitiveness, even amid rapid macroeconomic growth. GDP Non-disruptive Mag 6** Other Disruptive Innovation* 45 1.7 4.7 73 92 127 220 38% CAGR ARK’s Big Ideas 2025: Convergence 65 5 9 103 16 12 89 -2% CAGR 60 24% CAGR 140 50% CAGR 14 Note: “CAGR”: Compounded Annual Growth Rate. *We include cryptocurrencies and crypto assets in disruptive innovation capitalization. **Microsoft, Nvidia, Meta, Apple, Amazon, and Alphabet. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. Global Market Capitalization

• 15. Redefining Consumer Interactions And Business Workflows ARK’s Big Ideas 2025 AI Agents Nicholas Grous Frank Downing ASSOCIATE PORTFOLIO MANAGER DIRECTOR OF RESEARCH, NEXT GENERATION INTERNET Varshika Prasanna Jozef Soja RESEARCH ASSOCIATE, NEXT GENERATION INTERNET RESEARCH ANALYST, NEXT GENERATION INTERNET 15

• 16. AI Agents: Understand intent through natural language Plan using reasoning and appropriate context Take action using tools to accomplish the intent Improve through iteration and continuous learning What Are AI Agents? Platform-Level Agents Siloed, Single-Purpose Agents Usefulness Level Of Intelligence Universal Agents AI agents are poised to accelerate the adoption of digital applications and create an epochal shift in human-computer interaction. ARK’s Big Ideas 2025: AI Agents Smarter models using more tools to accomplish highervalue tasks Source: ARK Investment Management LLC, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. 16

• 17. 0 1 2 3 4 5 6 7 8 9 10 - 2 4 6 8 10 12 14 Revenue ($B) Years Since Product Launch AI Monetization is Outpacing Social Media OpenAI Instagram Facebook TikTok YouTube WeChat AI Is Accelerating The Adoption of Hardware And Software OpenAI could surpass $10 billion in revenue in 2025, monetizing at a faster rate than social media companies over the past decade. If the adoption of ChatGPT is an indicator, AI should drive rapid demand for a range of new technologies. ARK’s Big Ideas 2025: AI Agents $0 $1 $2 $3 $4 $5 $6 $7 $8 $9 $10 - 5 10 15 Revenue ($ Billions) Years Since Product Launch AI Monetization Is Outpacing Social Media OpenAI Instagram Facebook TikTok YouTube WeChat 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 1974 1983 1992 2001 2010 2019 2028E 2037E Percent Penetration US Consumer Hardware Adoption PC Cellphone Smartphone Smart Home Smartwatch VR Headset AI Hardware Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Hartig 1998, Dediu 2017, and Sidoti et al. 2024, as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 17 Starting in 2025, we believe most hardware sales will be AI-enabled, leading to fast mass adoption.

• 18. Embedded in the operating systems of consumer hardware, AI agents empower consumers to delegate all discovery and research to AI—a massive time-saver. Curated AI results will contextualize digital ad impressions. AI Agents Will Transform Consumer Search And Discovery TEXT / VOICE COMMAN D CHECKOU T AWARENESS AUTONOMOUS DISCOVERY + RESEARCH CONVERSION QUERY PERSONAL IZED AI AGENT CURATED RESULTS SEARCH ENGINE 1 SEARCH ENGINE 2 SEARCH ENGINE 3 MARKETPLACE 1 MARKETPLACE 2 MARKETPLACE 3 = Ad Opportunities Source: ARK Investment Management LLC, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. 18 ARK’s Big Ideas 2025: AI Agents

• 19. AI Agent Revenue Opportunities AI-Mediated Ads Should Take The Lion’s Share Of Digital Ad Revenue by 2030 If search shifts to personal AI agents, AI-mediated ad revenue could surge. By 2030, we believe AI ad revenue could account for more than 54% of the $1.1 trillion digital ad market. ARK’s Big Ideas 2025: AI Agents 0% 1% 2% 3% 4% 5% 6% 7% 8% August-21 August-22 August-23 August-24 Share Of Page Views ChatGPT, Claude, Perplexity, And Bing Share Of Search VS. Google (Based On 7-Day Rolling Average Of Page Views) Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 19 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% $- $200 $400 $600 $800 $1,000 $1,200 2024 2025E 2026E 2027E 2028E 2029E 2030E Gross Spend ($ Billions) Global Advertising Forecast AI Ad Revenue Digital Ad Spend AI Ad Revenue % of Total Digital Spend

• 20. AI-Mediated Shopping Is Likely To Approach 25% Of Addressable Online Sales Globally By 2030 The growing use of AI agents in consumer shopping should streamline product discovery, personalization, and purchasing. ARK research suggests that AI agents could facilitate nearly $9 trillion in global gross online consumption by 2030. Note: ARK measures “total addressable online sales” using internal and external estimates for global e-commerce and addressable services spending out to 2030. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 20 0% 5% 10% 15% 20% 25% 30% $- $1 $2 $3 $4 $5 $6 $7 $8 $9 $10 2023 2024 2025E 2026E 2027E 2028E 2029E 2030E Percent $ Trillions AI-Facilitated Online Spend AI-Facilitated Spend (LHS) % of Total Addressable Online Sales ARK’s Big Ideas 2025: AI Agents

• 21. Digital Wallets Are Positioned For Continued Share Gains In E-Commerce ARK’s research suggests that digital wallets empowered by AI purchasing agents—taking share from payment methods like credit and debit cards—could account for 72% of all ecommerce transactions by 2030. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2018 2019 2020 2021 2022 2023 2024 2025E 2026E 2027E 2028E 2029E 2030E Percent Share Of E-Commerce Purchase Volume Digital Wallet Other Payment Methods 21 Note: Other Payment Methods for E-commerce purchase volume include Credit Card, Debit Card, Bank Transfer, Cash on Delivery, and Buy Now Pay Later (BNPL). “Credit Card” and “Debit Card” refer to stored card credentials, while “Digital Wallet” includes credit/debit cards stored within the digital wallet. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including WorldPay as of December 31, 2024, which may be provided upon request. 2018-2023 figures are documented in the Worldpay Global Payments Reports. 2024-2030 figures are ARK's adoption estimates for e-commerce payment methods. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: AI Agents

• 22. Digital Wallets Are Consolidating Financial Services And E-Commerce Based on their consumer-facing operations, the market is valuing leading digital wallet platforms like Block, Robinhood, and SoFi at $1,800 per user today. $1,800 $250 $1,000 $300 $150 $100 $- $200 $400 $600 $800 $1,000 $1,200 $1,400 $1,600 $1,800 $2,000 Checking & Savings Lending Insurance Investing E-Commerce Total $ Digital Wallet Enterprise Value Per User (US) 22 Note: “Checking & Savings” data include Checking & Savings, Debit Card, and Peer to Peer Payments. “Lending” data include Loans & Mortgages, Credit Card, and Buy Now Pay Later (BNPL). Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources representing the revenues, margins, and multiples of leading digital wallets and established incumbents. as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. ARK’s Big Ideas 2025: AI Agents

• 23. Agentic lead generation should move digital wallets upstream to capture market share in e-commerce and digital spending globally. “One-click checkout” should give way to “one-query purchasing.” Digital Wallet Purchasing Agents Could Become Central To The Shopping Journey Interaction With Digital Wallet Existing System System With Purchasing Agent Consumer Search Engine / 3rd-Party Aggregator Untailored Results Shopping Cart / Bag Checkout Pay With Digital Wallet SEARCH LAYER AGGREGATION LAYER CHECKOUT LAYER SEARCH LAYER AGGREGATION LAYER CHECKOUT LAYER Consumer Marketplaces Personalized Results Automatic Checkout Source: ARK Investment Management LLC, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. 23 ARK’s Big Ideas 2025: AI Agents

• 24. In 2030, AI-powered purchasing agents could add between $50 and $200 per user to the enterprise value (EV) of digital wallets in the US. Purchasing Agents Should Increase The Enterprise Value Of Digital Wallets, Notably In E-Commerce 2030 Forecast Global Gross Spend Powered By AI Agents ~ $9 Trillion Percent Of AI Agent Spend In Digital Wallets 5% 10% 20% Lead Generation Take-Rate For Digital Wallets 2.5% $10B $20B $40B 5% $20B $40B $90B 10% $40B $90B $200B Based on lead-generation take rates, AI purchasing agents could generate global revenue between $40 billion and $200 billion—ARK’s base and bull cases, respectively—for digital wallet platforms in 2030. $100 $50 $200 $- $50 $100 $150 $200 $250 $300 $350 2023 2030E - Base Case 2030E - Bull Case $ AI-Driven Digital Wallet EV Per User In E-Commerce (US) E-Commerce Forecast - Additional Purchasing Agent Opportunity Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 24 ARK’s Big Ideas 2025: AI Agents

• 25. In The Enterprise, Agents Will Increase Productivity Via Software INBOUND REQUESTS AI Agent Platform SOFTWARE INVESTMENT LABOR SAVINGS Traditional Customer Support AI-Enabled Customer Support Companies that deploy agents should be able to increase unit volume with the same workforce and/or optimize their workforce toward higher-value activities. As AI evolves, agents are likely to handle a higher percentage of workloads and to complete highervalue tasks independently. HUMAN AGENTS INBOUND REQUESTS CUSTOMERS CUSTOMERS HUMAN AGENTS Source: ARK Investment Management LLC, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. 25 ARK’s Big Ideas 2025: AI Agents

• 26. AI Cost Declines Should Impact Agent Economics Significantly New products from OpenAI and Salesforce are supplementing human customer service representatives cost-effectively. Even at a fixed cost of $1 per conversation, AI agents could save enterprises significant sums once they can handle 35% of customer service inquiries. AI agents also should lower onboarding and hiring costs, as well as seat-based software costs, while scaling more easily than human labor. Potential Enterprise Customer Service Savings** ($ Billions) Dollar Cost Per Conversation 8x $290 $430 $500 $540 4x $140 $220 $250 $270 2x $70 $110 $130 $130 1x $40 $50 $60 $70 $1.00 $0.50 $0.25 $0.125 $0 $20 $40 $60 $80 $100 $120 $140 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% $ Billions Percent Of Customer Interactions Handled Successfully By AI AI Improvements In Customer Service Should Shift Labor Costs To Software Costs* Human Labor Costs AI Software Costs Savings Current U.S. Human Labor Spend Demand For Customer Service Relative To Today *Assumes fixed cost of $1 per customer interaction. **Assuming 70% of interactions can be handled by AI in all scenarios, corresponding to the 70% bar in the left-hand chart. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including ZipRecruiter and OpenAI, as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 26 ARK’s Big Ideas 2025: AI Agents

• 27. 0 -1 7 11 -7 -10 -10 -5 0 5 10 15 Share Shift (Percentage Points) PP 19% 31% 23% 39% 15% 27% 0% 20% 40% 60% Spend CAGRs Estimated Enterprise Software Stack Evolution Infrastructure Layer Platform Layer Application Layer o3 (OpenAI) Q Developer Agent (Amazon) Claude 3.5 Sonnet + Tools (Anthropic) Genie (Cosine) SWE-agent + GPT 4 (Princeton, OpenAI) Claude 2 + RAG (Anthropic) 0% 10% 20% 30% 40% 50% 60% 70% 80% Sep-23 Dec-23 Mar-24 Jun-24 Sep-24 Dec-24 SWE-Bench Verified Score* Percentage Of Real World Coding Tasks Solved By Coding Agents The coding skills of AI agents are improving rapidly, accelerating the software development lifecycle. As the cost to create software declines, software production should accelerate and sway enterprise “build vs. buy” decisions, displacing traditional software incumbents that are slow to adapt. As customer software proliferates, growth in all layers of the software stack should accelerate, even as share shifts toward the platform layer. AI Is Reshaping The Software Value Chain 2019 To 2024 (Actual) 2024 To 2030 (Forecast) Note: “CAGR”: Compound Annual Growth Rate. *SWE-Bench is a benchmark that measures AI agents’ ability to write code autonomously. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 27 ARK’s Big Ideas 2025: AI Agents

• 28. AI is driving a Cambrian explosion in software. Through 2030, we expect the amount of software deployed per knowledge worker to grow considerably as businesses invest in productivity solutions. Depending on adoption rates, global spend on software could accelerate from an annual rate of 14% over the last ten years to annual rates of 18% to 48%. AI Will Supercharge Knowledge Work Modest Investment Accelerated Investment Rapid Mass Adoption Annual knowledge worker employment growth 6.3% 3.2% 1.3% Percentage of current working time automated by 2030 31% 61% 81% Reduction in productive working hours 0% 8% 20% Productivity surplus created* $22 trillion $57 trillion $117 trillion Value capture of productivity solutions** 10% 10% 10% New software revenue $2.2 trillion $5.7 trillion $11.7 trillion 2030 Software Market Estimate (current size + AI revenue) $3.5 trillion 18% CAGR $7 trillion 33% CAGR $13 trillion 48% CAGR Adoption Scenarios $0 $10 $20 $30 $40 $50 $60 Current Modest Investment Scenario Accelerated Investment Scenario Rapid Mass Adoption Scenario 2030 Enterprise Spending Forecast Knowledge Work Labor Software 28 Note: “CAGR”: Compound Annual Growth Rate. *Traditional production statistics are unlikely to adequately capture the surplus created by AI software. **Value capture rates are likely to vary across scenarios based on competition in the market and other factors. The rate is held constant in these scenarios for simplicity. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: AI Agents

• 29. A Maturing Global Monetary System With Sound Network Fundamentals And Growing Institutional Adoption ARK’s Big Ideas 2025 Bitcoin Yassine Elmandjra David Puell FORMER DIRECTOR OF DIGITAL ASSETS RESEARCH TRADING ANALYST & ASSOCIATE PORTFOLIO MANAGER, DIGITAL ASSETS Lorenzo Valente DIRECTOR OF DIGITAL ASSETS 29

• 30. $30,000 $40,000 $50,000 $60,000 $70,000 $80,000 $90,000 $100,000 $110,000 Jan-24 Feb-24 Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Dec-24 ($) Bitcoin Price Bitcoin Hit An All-Time High In 2024 Bitcoin Undergoes Its Fourth Halving US Launches Spot Bitcoin ETFs Ethereum’s Dencun Upgrade Activates On Mainnet Sam Bankman Fried Sentenced to 25 Years In Prison US Launches Spot Ether ETFs German Government Sells Out Of 50,000 Seized Bitcoin Bitcoin Hash Rate Hits An All-Time High Genesis Settles SEC Lawsuit Over Gemini Earn Hong Kong Debuts Asia’s First Spot Bitcoin And Ether ETFs SEC Issues Robinhood Crypto With Wells Notice Bitcoin Market Cap Dominance Surges Past 55% For The First Time in Three Years Bitcoin Hits A Technical All-Time High ARK’s Big Ideas 2025: Bitcoin As Donald Trump Wins US Presidential Election, Bitcoin Breaks All-Time High SEC Grants Trading And Listing Bitcoin ETF Options Pennsylvania Is First US State To Propose Bitcoin Strategic Reserve David Sacks Appointed US AI And Crypto Czar Source: ARK Investment Management LLC, 2025, based on data from Glassnode and TradingView as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. Mt. Gox Begins $9 Billion Payout To Creditors MicroStrategy Announces $42-Billion Raise Over The Next 3 Years 30

• 31. The Spot Bitcoin ETF Complex Was The Most Successful ETF Launch In History On their first day of trading, the spot bitcoin ETFs attracted over $4 billion of inflows, a record high for ETF launches, surpassing the $1.2 billion that flowed into the gold ETF in its first month in November 2004. Inflows into the spot bitcoin ETFs have dwarfed the firstmonth inflows into each of the ~6,000 ETFs launched over the past 30 years. $0 $20 $40 $60 $80 $100 $120 Jan-24 Feb-24Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Dec-24 $ Billions Spot Bitcoin ETFs1 AUM2 IBIT FBTC ARKB BITB BTC HODL BRRR BTCO EZBC BTCW 0 10 20 30 40 50 60 70 80 90 100 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 $ Billions Number Of Months From Launch Spot Bitcoin ETFs And Gold ETF AUM In Year One and Beyond Bitcoin ETF Cumulative Flows Gold ETF Cumulative Flows [1] GBTC and DEFI are not included. [2] “AUM”: assets under management. Source: ARK Investment Management LLC, 2025, based on data from Glassnode and World Gold Council as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 31 ARK’s Big Ideas 2025: Bitcoin

• 32. Upon Its Fourth Halving, Bitcoin’s Inflation Rate Dropped Below Gold’s Long-Term Supply Growth Growth in the supply of bitcoin “halved” for the fourth time in history, dropping from ~1.8% at an annual rate to ~0.9%. Critical to its design, the halving highlights bitcoin’s predictable monetary policy and its role as a scarce asset. Seven Lines Of Code Created Bitcoin’s Mathematically-Metered Monetary Policy1 0% 20% 40% 60% 80% 100% 0 5 10 15 20 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 Percent BTC, Millions Bitcoin Supply Mined Supply (lhs) Percent to Be Mined (rhs) 0% 1% 10% 100% 1000% 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 Annualized Percent Inflation Rate: Bitcoin VS. Gold BTC Supply Growth Gold Supply Growth [1] Screenshot as of December 31, 2024, sourced from (github.com/bitcoin/bitcoin/blob/master/src/validation.cpp) lines 1939-1949. Source: ARK Investment Management LLC, 2025, based on data from Glassnode and World Gold Council as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 32 ARK’s Big Ideas 2025: Bitcoin

• 33. Bitcoin’s Annual Volatility Fell To An All-Time Low, While Its Risk-Adjusted Returns Remained Superior To Most Major Asset Classes Asset Class1 2024 Return Sharpe Ratio2 Sortino Ratio3 5-Year CAGR4 Bitcoin 122.2% 1.4 4.1 67.2% Gold 26.6% 1.7 4.8 10.4% Equities 19.2% 1.3 2.3 11.7% Emerging Markets 8% 0.2 0.4 2.1% Real Estate 3.9% -0.01 -0.01 1.5% 0% Bonds -3.1% -1.1 -1.2 -2.7% 50% 100% 150% 200% 250% 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 Annualized Percent, 1-Year, Rolling Bitcoin Volatility []1 Data used for calculations are as follows: S&P Bitcoin PR USD for bitcoin; S&P GSCI Gold TR for gold; MSCI World GR USD for equities; MSCI EM GR USD for emerging markets; S&P Global REIT TR USD for real estate; and Bloomberg Gbl Agg Govt TR USD for bonds. The returns shown are of unmanaged indices and therefore do not reflect the deduction of any fees or expenses that would be applicable to investors in these asset classes, which would lower the returns shown. [2] Sharpe ratio is calculated using average returns over 2024 adjusted for the risk-free rate, divided by the standard deviation. [3] Sortino ratio is calculated using average returns over 2024 adjusted for the risk-free rate, divided by the downside deviation. [4] “CAGR”: Compound Annual Growth Rate. Source: ARK Investment Management LLC, 2025, based on data from Glassnode and Morningstar Direct as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 33 ARK’s Big Ideas 2025: Bitcoin

• 34. 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 $0 $10 $20 $30 $40 $50 $60 $70 $80 Jan-22 Jul-22 Jan-23 Jul-23 Jan-24 Jul-24 BTC $ Millions Miner Revenue Miner Revenue ($) Miner Revenue (BTC) Despite A Steep Decline In Miner Revenue After The Halving, Bitcoin’s Hash Rate Hit An All-Time High Even though the halving cut bitcoin miner revenue by 50%, its hash rate—a proxy for network security—hit a record high. In other words, miners’ long-term conviction in bitcoin remains strong. Miner revenue cut in half 0 100,000,000 200,000,000 300,000,000 400,000,000 500,000,000 600,000,000 700,000,000 800,000,000 900,000,000 2017 2018 2019 2020 2021 2022 2023 2024 TH/s Hash Rate1 [1] In terahashes per second, a unit of measurement used to determine the estimated rate of processing power securing the Bitcoin network. It equals 1 trillion hashes per second. Source: ARK Investment Management LLC, 2025, based on data from Glassnode as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 34 ARK’s Big Ideas 2025: Bitcoin

• 35. Transaction Count Soared, Thanks To Runes Bitcoin’s daily transaction count hit a record high, thanks to the launch of the Runes protocol. Runes facilitates the creation of fungible tokens directly on the Bitcoin blockchain. 0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 Change-Adjusted, 14-Day Average Transaction Count 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% May-23 Aug-23 Nov-23 Feb-24 May-24 Aug-24 Nov-24 Distribution (%) Transaction Count Share1 Runes BRC20 Inscriptions Ex BRC20 Regular Transactions [1] “BRC20”: Token standard that enables the minting and transaction of fungible tokens via the Ordinals protocol on the Bitcoin network. Source: ARK Investment Management LLC, 2025, based on data from Dune Analytics and Glassnode as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 35 ARK’s Big Ideas 2025: Bitcoin

• 36. Bitcoin Absorbed Major Selling Pressure In 2024 In January 2024, the German government seized 50,000 bitcoin linked to an online piracy group. Six months later, it sold the bitcoin. The market absorbed the supply successfully, and then the price of bitcoin rallied from $53,000 to $68,000. Also, around mid-year, the long-awaited Mt. Gox1 creditor repayment process distributed more than 109,000 bitcoin, removing its largest overhang. $0 $20,000 $40,000 $60,000 $80,000 $100,000 $120,000 0 50,000 100,000 150,000 200,000 250,000 Jan-24 Apr-24 Jul-24 Oct-24 Bitcoin Balance (BTC) Bitcoin Price ($) Major Bitcoin Distributions VS. Bitcoin Price US Government Balance (BTC) Mt. Gox Balance (BTC) German Government Balance (BTC) Bitcoin Price ($) [1] Mt. Gox was a bitcoin exchange that once handled up to 70% of bitcoin’s total trading volume. It declared bankruptcy in 2014 after it was found to have run a fraudulent fractional reserve, incurring the loss of 850k bitcoin from users’ funds at the time. During 2024, a large portion of reclaimed bitcoin was repaid to creditors, from an estimated 141,686 bitcoin at the start of the year to 27,877 by November 13. Data estimates sourced from Glassnode. Source: ARK Investment Management LLC, 2025, based on data from Glassnode as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 36 ARK’s Big Ideas 2025: Bitcoin

• 37. An Increasing Number Of Public Companies Now Hold Bitcoin Public Company (Top 10) Bitcoin Holdings USD Value (Millions) BTC Holdings As Percentage Of Market Cap Microstrategy (MSTR) 446,400 $41,780 58.7% Marathon Digital (MARA) 44,893 $4,201 73.8% Riot Platforms (RIOT) 17,429 $1,631 46.4% Hut 8 Mining (HUT) 10,096 $944 49.4% Tesla (TSLA) 9,720 $909 0.07% Coinbase (COIN) 9,480 $887 1.4% CleanSpark (CLSK) 9,297 $870 32.3% Block (SQ) 8,363 $782 1.4% Bitcoin Group SE (ADE) 3,589 $335 133.5% Boyaa Interactive (BOYAA/0434.HK) 3,183 $297 10.4% $0 $10 $20 $30 $40 $50 $60 0 100 200 300 400 500 600 700 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec $ Billions BTC, Thousands Bitcoin On Public Company Balance Sheets Bitcoin Balance USD Value Source: ARK Investment Management LLC, 2025, based on data from Bitcoin Treasuries and TradingView as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. Seventy-four public companies now hold bitcoin on their balance sheets. The value on corporate balance sheets quintupled during the past year, from $11 billion in 2023 to $55 billion. 37 ARK’s Big Ideas 2025: Bitcoin

• 38. Bitcoin’s Aggregate Cost Basis Hit A New High In 2024 In 2024, Bitcoin’s realized capitalization—or cost basis—grew by 86%, as its average acquisition cost hit a record high at $40,980 per bitcoin, or $811.7 billion in total. $0 $500 $1,000 $1,500 $2,000 $2,500 2019 2020 2021 2022 2023 2024 $ Billions Bitcoin Cost Basis Market Cap ($) Realized Cap ($) Source: ARK Investment Management LLC, 2025, based on data from Glassnode as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 38 ARK’s Big Ideas 2025: Bitcoin

• 39. Bitcoin’s Transaction Velocity1 And Holding Behavior Underscore Its Role As A Store Of Value Bitcoin’s velocity dropped to a 14-year low in 2024, as supply held for three years or more reached an all-time high. 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 Percent Bitcoin Supply (%) Supply Last Active 3+ Years Ago 1 10 100 1,000 2011 2012 2013 2014 2015 2016 2017 20182019 2020 2021 2022 2023 2024 Change-Adjusted, Annualized, 30-Day Average Bitcoin's Velocity [1] “Velocity”: a measure of how quickly bitcoin units circulate in the network, calculated by dividing the on-chain bitcoin transaction volume by the outstanding supply (annualized). Source: ARK Investment Management LLC, 2025, based on data from Glassnode as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 39 ARK’s Big Ideas 2025: Bitcoin

• 40. $0 $200,000 $400,000 $600,000 $800,000 $1,000,000 $1,200,000 $1,400,000 $1,600,000 2030 Price Target Bear Case 2030 Price Target Base Case 2030 Price Target Bull Case ($) 2030 Bitcoin Price Targets (12/31/24 to 12/31/30 CAGR) Institutional Investment Digital Gold Emerging Market Safe Haven Nation-State Treasury Corporate Treasury Bitcoin On-Chain Financial Services Bitcoin Is On Pace To Meet Our 2030 Price Targets Bear Case Base Case Bull Case Institutional Investment TAM: Global Market Portfolio Ex Gold 1% 2.5% 6.5% Digital Gold TAM: Gold Market Cap 20% 40% 60% Emerging Market Safe Haven TAM: EM M2 Monetary Base 0.5% 2.5% 6% Nation-State Treasury TAM: Global Treasury Reserves Ex Gold 0.5% 2.5% 7% Corporate Treasury TAM: Global Cash & Cash Equivalents 1% 2.5% 10% Bitcoin On-Chain Financial Services TAM: L2s, LN, Sidechains, Restaking & WBTC 20% CAGR 40% CAGR 60% CAGR $300,000 CAGR ~21% $710,000 CAGR ~40% $1.5 million CAGR ~58% Source: ARK Investment Management LLC, 2025. This ARK analysis is based on a range of external sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 40 Price Target Assumptions ARK’s Big Ideas 2025: Bitcoin

• 41. Reshaping The Digital Asset Space Stablecoins Yassine Elmandjra Lorenzo Valente FORMER DIRECTOR OF DIGITAL ASSETS David Puell RESEARCH TRADING ANALYST & ASSOCIATE PORTFOLIO MANAGER, DIGITAL ASSETS DIRECTOR OF DIGITAL ASSETS Frank Downing DIRECTOR OF RESEARCH, NEXT GENERATION INTERNET 41 ARK’s Big Ideas 2025

• 42. One Of The Fastest-Growing Segments In Digital Assets, Stablecoins Overtook Mastercard And Visa In Transaction Value In 2024 Despite a two-year bear market and a 70%+ decline in market capitalization, stablecoin growth has remained uninterrupted. In 2024, the annualized transaction value of stablecoins hit $15.6 trillion—roughly 119% and 200% that of Visa and Mastercard, respectively. The number of transactions hit 110 million monthly, roughly 0.41% and 0.72% of those processed by Visa and Mastercard, respectively. In other words, the stablecoin value per transaction is much higher than that for Visa and Mastercard. -0.50 0.50 1.50 2.50 3.50 4.50 0 20 40 60 80 100 120 140 Jan-20 May-20 Sep-20 Jan-21 May-21 Sep-21 Jan-22 May-22 Sep-22 Jan-23 May-23 Sep-23 Jan-24 May-24 Sep-24 $ Trillions Millions Stablecoin Transaction Count VS. Total Digital Asset Market Capitalization Stablecoin Transaction Count (lhs) Digital Asset Market Capitalization (rhs) $0 $2 $4 $6 $8 $10 $12 $14 $16 $18 2019 2020 2021 2022 2023 2024 $ Trillions Transaction Value: Stablecoins VS. Traditional Payment Processors Visa Mastercard American Express Stablecoins 42 Note: In the lefthand chart, stablecoins data are for calendar year, while credit card data are for fiscal year. Source: ARK Investment Management LLC, 2025. This ARK analysis is based on a range of external data sources, including Wall Street Zen, CoinGecko, and Visa OnChain Analytics, as of December 31, 2024, which may be provided upon request, For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. ARK’s Big Ideas 2025: Stablecoins

• 43. 0 0.5 1 1.5 2 2.5 3 Dec-23 Jan-24 Feb-24 Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Dec-24 $ Trillions Stablecoin Volume By Blockchain Solana Tron Ethereum Base Arbitrum BNB Chain Avalanche Optimism Polygon POS TON Celo Stablecoin Volumes Hit Record Highs In December 2024 Solana, Tron, Ethereum, and Base are the leading blockchains that drove stablecoin volume in 2024. December 2024 set new records with a daily volume of $270 billion and a monthly volume of $2.7 trillion, highlighting the sector's rapid growth. 3.8 x Source: ARK Investment Management LLC, 2025., based on data from Artemis Terminal as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Stablecoins 43

• 44. Ethena Labs Tokenized A Popular Basis Trade And Amassed $6 Billion In 12 Months Innovation in the stablecoin space is thriving, with Ethena Labs as one of the fastest-growing projects. Despite criticism of its novel design, the project has captured significant share of the non-fiatbacked stablecoin market and has become a key competitor in this $200 billion space. Ethena Labs reached $6 billion in total value locked during its first twelve months. By tokenizing a delta-neutral position, Ethena Labs can offer yields as high as 20%-30%, depending on market conditions. The yield could turn negative in a bear market. Having captured 10% of total Ether open interest, Ethena Labs is expanding into Solana and Bitcoin delta-neutral trades. User provides $100 USD worth of ETH Ethena swaps the collateral for $100 worth of yield bearing ETH such as stETH which is deposited at a custodian An Exchange opens a short position worth $100 with the deposit at the custodian as collateral Delta neutral position in which the short position offsets the spot collateral $100 USDE are minted to the user User Note: A delta neutral position in finance is a strategy designed to hedge against price movements in the underlying asset. Source: ARK Investment Management LLC, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 1 2 3 4 5 44 ARK’s Big Ideas 2025: Stablecoins

• 45. UDST (Tether) continues to dominate the stablecoin landscape, followed by USDC (Circle). Together, they account for 90% of the total supply. Stablecoins are “multichain” and have penetrated almost every major Layer 1 blockchain. At $203 billion, the stablecoin supply represents ~0.97% of M2* money supply in the US. During December, active stablecoin addresses hit 23 million, an all-time high. As measured by monthly active addresses, Tron is the leading network, favored in emerging markets for its low transaction fees. After A Drawdown In 2023, The Supply Of Stablecoins And The Number Of Active Stablecoin Addresses Hit All-Time Highs In 2024 ARK’s Big Ideas 2025: Stablecoins *M2 is a measure of the US money stock that includes M1 (currency and coins held by the non-bank public, checkable deposits, and travelers’ checks) plus savings deposits (including money market deposit accounts), small time deposits under $100,000, and shares in retail money market mutual funds. Source: ARK Investment Management LLC, 2025, based on data from rwa.xyz (“Stablecoins”) as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 0 5 10 15 20 25 30 Jan-18 Apr-18 Jul-18 Oct-18 Jan-19Apr-19 Jul-19Oct-19 Jan-20 Apr-20 Jul-20 Oct-20 Jan-21 Apr-21 Jul-21Oct-21Jan-22 Apr-22 Jul-22 Oct-22 Jan-23 Apr-23 Jul-23 Oct-23 Jan-24 Apr-24 Jul-24Oct-24 $ Millions Monthly Active Stablecoin Addresses By Network TRON Binance Smart Chain Solana Ethereum Arbitrum Celo Polygon Optimism Aptos Avalanche NEAR Mantle XDC Base $0 $50 $100 $150 $200 $250 Jan-20 Jan-21 Jan-22 Jan-23 Jan-24 $ Billions Stablecoin Supply USDT USDC USDS USDe FDUSD 45

• 46. Digital Assets are undergoing “stablecoinization” and “dollarization,” while China and Japan have sold record volumes of US Treasuries, Saudi Arabia has ended its 45-year petrodollar agreement, and BRICS* nations have reduced their reliance on US dollar payments by circumventing the SWIFT network. Historically, bitcoin and ether were the bridges into the digital asset ecosystem. During the past two years, however, stablecoins have dethroned them and now account for 35% to 50% of on-chain transaction volumes. Dollar-pegged stablecoins dominate, accounting for 98%+ of the supply, followed by Gold-backed and Euro-backed stablecoins, with ~1% and ~0.5%, respectively. ARK’s research suggests that the market will expand and include Asian currency-backed stablecoins. As Some Countries Shift Away From The Dollar, Digital Assets Are Moving Toward It 0% 20% 40% 60% 80% 100% Sep-19 Dec-19 Mar-20 Jun-20 Sep-20 Dec-20 Mar-21 Jun-21 Sep-21 Dec-21 Mar-22 Jun-22 Sep-22 Dec-22 Mar-23 Jun-23 Sep-23 Dec-23 Mar-24 Jun-24 Sep-24 Dec-24 Share Settlement Volume Across Different Blockchains Stablecoins L1 L2 Bitcoin Ethereum *BRICS refers to five major emerging economies—Brazil, Russia, India, China, and South Africa. Source: ARK Investment Management LLC, 2025, based on data from Artemis Terminal as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 46 ARK’s Big Ideas 2025: Stablecoins

• 47. Stablecoins Are Attracting Retail Interest, Thanks To The Lower Cost And Higher Efficiency Of Layer 2s Retail investors are flocking to Layer 2s for less expensive and more convenient stablecoin transactions, increasing the market share of Blockchains like Arbitrum, Base, and Optimism. Meanwhile, whales* and institutions continue to operate on Ethereum's base layer. Transactions below $100 dominate activity on Base and Optimism, while those above $100 dominate Ethereum’s base layer. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Jan-22 Mar-22 May-22 Jul-22 Sep-22 Nov-22 Jan-23 Mar-23 May-23 Jul-23 Sep-23 Nov-23 Jan-24 Mar-24 May-24 Jul-24 Sep-24 Nov-24 Stablecoin Transaction Count Share By Blockchain Ethereum Arbitrum Base Optimism Celo 0% 20% 40% 60% 80% 100% Celo Optimism Base Arbitrum Ethereum Transaction Size Share By Blockchain In 2024 Less than $100 Between $100 and $1K Between $1K and $10K Between $10k and $100K Between $100k and $1M Between $1M and $10M Over $10M *“Whale”: an individual or entity that holds a large amount of a specific cryptocurrency, usually north of 5-10 million dollars depending on the asset. Source: ARK Investment Management LLC, 2025, based on data from Visa OnChain Analytics as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 47 ARK’s Big Ideas 2025: Stablecoins

• 48. Externally Owned Accounts (EOAs)— standard Ethereum addresses used for peer-to-peer (P2P) transactions and storage—account for 60% of USDC’s usage, while centralized exchanges account for 11%, bridges across Layer 2 solutions for 7%, and decentralized exchanges (DEXs) and money markets each for 1.7%. As DeFi usage surges in the next few years, DEXs, Bridges, and Money Markets are likely to retake market share from P2P. While usage in lending/borrowing markets, DEXs, and Bridges fluctuates with market cycles, P2P transactions and storage are more resilient because of product-market fit beyond trading. Peer-to-Peer Transactions And Personal Wallet Storage Dominate Stablecoin Use Cases Source: ARK Investment Management LLC, 2025, based on data from Dune Analytics as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 48 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Jan-21 May-21 Sep-21 Jan-22 May-22 Sep-22 Jan-23 May-23 Sep-23 Jan-24 May-24 Sep-24 Share Top Addresses Storing USDC EOA CeFi Other Bridge Treasury DEX Lending Yield Farm ARK’s Big Ideas 2025: Stablecoins

• 49. Four Stablecoin Issuers Dominate The Revenue Generated In Digital Assets With fewer than 200 employees, Tether reported $5.2 billion in profits during the first half of 2024, including USDT, the rest of its products and services, and unrealized gains on its Digital Assets—clearly one of the most capitalefficient businesses in history. Tether (USDT) and Circle (USDC) account for 60% of the revenue generated by the top five networks and applications. Collectively, stablecoins USDT, USDC, DAI/USDS, and USDE generated revenues of $3.35 billion, or $6.7 billion at an annual rate, during the second half of 2024. 0 500 1000 1500 2000 2500 3000 Tether (usdt) Tron Circle (usdc) Ethereum Solana Aerodrome Sky (usds/dai) Ethena labs (usde) Lido Finance Aave Uniswap Pancakeswap $ Millions Top 12 Blockchain Networks And Applications By Revenue* Stablecoin Issuers Others *Second Half of 2024. Source: ARK Investment Management LLC, 2025, based on data from Token Terminal of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. Tether (USDT) Circle (USDC) Sky (USDS/DAI) Ethena Labs (USDE) 49 ARK’s Big Ideas 2025: Stablecoins

• 50. Tether’s Financial Performance Is Stunning, Both Absolutely And Relatively Tether employs fewer than 200 people compared to 300,000+ each at JP Morgan and Berkshire Hathaway. The only institutions in the S&P Financial Select Sector Index generating more net income than Tether in the first half of 2024 were Berkshire Hathaway, JP Morgan, Bank of America, and Wells Fargo. 1 10 100 1000 10000 100000 1000000 10000000 100000000 0 5 10 15 20 25 30 35 Tether Visa Mastercard Berkshire Hathaway Goldman Sachs JP Morgan Amercan Express Bank of America Progressive S&P Global Wells Fargo $, Log Scale $ Billions Tether VS. Leading Financial Firms* Net Income (lhs) Net Income Per Employee (rhs) *First Half of 2024. Source: ARK Investment Management LLC, 2025. This ARK analysis is based on a range of external data, sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 50 ARK’s Big Ideas 2025: Stablecoins

• 51. In a world trending toward deglobalization and de-dollarization, stablecoins could drive steady demand for US Treasury securities. Combined, Tether and Circle already ranked as the 20th largest holder of US Treasuries as of December 2024. In populous emerging markets like Brazil, Nigeria, Turkey, Indonesia, and India, individuals and companies are adopting stablecoins as a store of value, means of payment, and crossborder currency. Stablecoins could become one of the most effective ways to export US dollars. Counterbalancing ”De-Dollarization,” Stablecoins Are Increasing The Demand For US Government Debt As Collateral 0 200 400 600 800 1000 1200 Japan China United Kingdom Cayman Islands Luxembourg Canada Belgium France Ireland Switzerland Taiwan Singapore India Brazil Hong Kong Norway Saudi Arabia Korea, South Germany Tether & Circle Mexico $ Billions Top Foreign Holders Of US Treasury Securities As Of September 2024 Source: ARK Investment Management LLC, 2025. This ARK analysis is based on a range of external data sources, including Tether, Circle, and TicData Treasury, as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 51 ARK’s Big Ideas 2025: Stablecoins

• 52. Circle and Tether have been generating billions in revenue from Treasury bills and other securities serving as collateral for their stablecoins. In 2024, however, in response to competition and demand, stablecoins operating outside of the US began to pass a significant portion of their interest income to users. Circle and Tether are unlikely to follow the trend until absolutely necessary. Though still small, yield-bearing stablecoins are the fastest-growing category in the stablecoin market. Stablecoin Issuers Are Transferring Risk-Free Rates To Users Source: ARK Investment Management LLC, 2025, based on data from Dune Analytics as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 52 0 1 2 3 4 5 6 7 8 Mar-23 May-23 Jul-23 Sep-23 Nov-23 Jan-24 Mar-24 May-24 Jul-24 Sep-24 Nov-24 $ Billions Size Of The Yield Bearing Stablecoin Market sUSDe USD0 sDAI USDY deUSD USDB sFRAX USDM USDL rUSD 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Jan-23 Apr-23 Jul-23 Oct-23 Jan-24 Apr-24 Jul-24 Oct-24 Share Of Interest-Bearing VS. Non Interest Bearing Stablecoins Yield Bearing Non Yield-Bearing ARK’s Big Ideas 2025: Stablecoins

• 53. 203 1409 0.00% 0.10% 0.20% 0.30% 0.40% 0.50% 0.60% 0.70% 0.80% 0.90% 1.00% $0 $200 $400 $600 $800 $1,000 $1,200 $1,400 $1,600 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 $ Billions Stablecoin Supply VS. Fiat Global M2 2019 To 2030 (2025-2030 Forecast) Stablecoin Supply (lhs) Stablecoins as Percentage of Global M2 Supply (rhs) Stablecoins Could Increase From 0.17% to 0.9% Of Fiat* Global M2 Supply By 2030 Currently $203 billion, or 0.17% of global M2** supply, stablecoins could grow to $1.4 trillion and 0.9%, respectively, by 2030. If so, stablecoins would be the 13th largest currency in circulation, behind Spain and ahead of Netherlands. 38% CAGR 53 Note: “CAGR”: Compound Annual Growth Rate. *Fiat money is a type of currency that is issued by a government and has value because the government declares it to be legal tender. **M2 is a measure of the US money stock that includes M1 (currency and coins held by the non-bank public, checkable deposits, and travelers’ checks) plus savings deposits (including money market deposit accounts), small time deposits under $100,000, and shares in retail money market mutual funds. Source: ARK Investment Management LLC, 2025, based on data from rwa.xyz (“Stablecoins”) as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. ARK’s Big Ideas 2025: Stablecoins

• 54. Enabling Steep Cost Declines And New Use-Cases At The Application Layer Scaling Blockchains Yassine Elmandjra Lorenzo Valente FORMER DIRECTOR OF DIGITAL ASSETS David Puell RESEARCH TRADING ANALYST & ASSOCIATE PORTFOLIO MANAGER, DIGITAL ASSETS DIRECTOR OF DIGITAL ASSETS 54 Frank Downing DIRECTOR OF RESEARCH, NEXT GENERATION INTERNET ARK’s Big Ideas 2025

• 55. Navigating The Smart Contract Landscape As the digital asset space becomes increasingly complex, smart contracts are driving innovation across a growing number of sectors. The ecosystem is evolving rapidly to meet diverse and dynamic needs—from user-focused applications like gaming and SocialFi, to advanced financial tools like derivatives and structured products, to decentralized infrastructure networks powering wireless connectivity and energy storage. Consumer Applications Gaming Ticketing Smart Devices Trading Apps Decision Markets SocialFi Music Messaging Prediction Markets Metaverse Community And Culture Memecoins NFTs Media And Entertainment Community Accelerators Conference and Events DAOs Generative Art Clubs And Collectibles DeFi RWA Derivatives And Perpetuals NFT Marketplaces and Tooling Trading Bots Launchpads DEX/CEX/HEX Restaking Liquid Staking Asset Creators/ Issuers NFT Lending And Fractionalization Stablecoins Money Markets OTC/ RFQ Derivatives And Spot Aggregators YieldStructured Products Indexes Insurance Risk Management DePIN Mapping Wireless Storage Environment Compute Data Management Energy Services Crypto x AI Compute Marketplaces AI Agents Data Availability Gaming Privacy–ZK-FHE Coordination Networks CoProcessors Model TrainingInference Model Creation Infrastructure And Developer Tooling Wallets DAO Tooling Oracles Developer Tooling And Standards Clients And Nodes Data And Analytics Bridges And CrossChain Messaging Layer 1s RaaS-Shared Sequencers Web 3 Identity Payment And Blink Tools MEV Clients And Tooling Privacy Testing and Frameworks Custodial Services Layer 2s, Layer 3s, and RaaS Onramp Tools Auditing and Bug Bounties Roll-Up Frameworks PoW And PoS Services Source: ARK Investment Management LLC, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. ARK’s Big Ideas 2025: Scaling Blockchains 55

• 56. Applications Seeking Lower Fees And Greater Efficiency Deploy Either On Solana’s High Throughput—Layer 1—Or On Ethereum’s Layer 2s Layer 1 Blockchain Layer 2 Blockchain Application Layer User Aggregation Layer Source: ARK Investment Management LLC, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. 56 Ledger Metamask Coinbase Wallet Arbitrum Base Polygon Solana Ethereum Pendle Interest rate Marketplace GMX Perpetual Exchange Polymarket Prediction Market Aave Money Market Morpho Borrow/Lend Infrastructure Jupiter Derivatives Exchange Jito Payment For Order Flow Orca Spot Exchange ARK’s Big Ideas 2025: Scaling Blockchains

• 57. 0 0.5 1 1.5 2 2.5 Dec-23 Jan-24 Feb-24 Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Dec-24 $ Ethereum Layer 2s Median Fee Arbitrum Optimism Base zkSync Scroll Blast Linea One of Ethereum's most significant technical upgrades to date, EIP 4844, impacted Layer 2 networks by enabling faster, less-expensive transactions. Ethereum’s EIP 4844 Technical Upgrade Cut Transaction Costs By 10x, Spurring Use Prior to the EIP 4844 upgrade, the average transaction fee on Ethereum Layer 2s was ~$0.50. Now, users pay ~$0.05. EIP 4844 was the first milestone in a roadmap that should enable a 250- to 1250- fold increase in transactions per second (TPS) from ~400 today to 100,000-500,000. More than 200 Layer 2 projects have launched, with others on the way. EIP 4844 Upgrade Source: ARK Investment Management LLC, 2025, based on data from Dune Analytics as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 57 ARK’s Big Ideas 2025: Scaling Blockchains

• 58. Plummeting Transaction Costs Have Led To A Boom In Layer 2 Activity, Pulling Users Away From Ethereum’s Base Layer Of the Daily Active Addresses transacting on the Ethereum ecosystem, Layer 2s have captured 85% share. The activity on Layer 2s enabled Ethereum to scale daily transactions by 400% from 3 million to 15 million in 2024. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Dec-21 Mar-22 Jun-22 Sep-22 Dec-22 Mar-23 Jun-23 Sep-23 Dec-23 Mar-24 Jun-24 Sep-24 Dec-24 Share Daily Active Addresses: Ethereum VS. Layer 2s Ethereum Top 10 Ethereum L2s 0 5 10 15 20 25 30 Dec-21 Feb-22 Apr-22 Jun-22 Aug-22 Oct-22 Dec-22 Feb-23 Apr-23 Jun-23 Aug-23 Oct-23 Dec-23 Feb-24 Apr-24 Jun-24 Aug-24 Oct-24 Number (Millions) Daily Transactions: Ethereum VS. Layer 2s Ethereum Top 10 Ethereum L2s Combined Source: ARK Investment Management LLC, 2025, based on data from Artemis Terminal as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 58 ARK’s Big Ideas 2025: Scaling Blockchains

• 59. Within one year of launch, Base outpaced all other Ethereum Layer 2 solutions in growth and market share. In 2024, Base captured 46% of active users and generated 63% of fees among Ethereum Layer 2s. With $15 billion in total value locked and more than 300 applications deployed, Base has been contributing significantly to Coinbase’s cash-flows. Base Is The Fastest Growing Ethereum Layer 2 Blockchain Source: ARK Investment Management LLC, 2025, based on data from Artemis Terminal as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 59 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Dec-23 Jan-24 Feb-24 Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Share Layer 2 Daily Active Addresses Base Celo Arbitrum Mantle OP Mainnet Linea zkSync Era Blast Scroll StarkNet 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Dec-23 Jan-24 Feb-24 Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Share Layer 2 Fees Base Arbitrum OP Mainnet Linea Mantle StarkNet zkSync Era Blast Scroll Celo ARK’s Big Ideas 2025: Scaling Blockchains

• 60. Institutions, high-value users, and whales are settling their transactions primarily on the Ethereum base layer. The unit economics of Ethereum’s base layer, as measured by Total Value Locked (TVL) and Decentralized Exchange (DEX) volume per user, are unparalleled. Despite Migration To Layer 2s, Ethereum's Base Layer Still Dominates High-Value Storage And Settlement Note: “Whale”: an individual or entity that holds a large amount of a specific cryptocurrency, usually north of 10 million dollars depending on the asset. Source: ARK Investment Management LLC, 2025, based on data from Artemis Terminal as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 60 $0 $20 $40 $60 $80 $100 $120 $140 $160 $180 $200 Dec-23 Jan-24 Feb-24Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 $ Thousands Total Value Locked Per Daily Active Address Ethereum StarkNet Blast Scroll Celo Linea Mantle OP Mainnet zkSync Era ARK’s Big Ideas 2025: Scaling Blockchains

• 61. When Binance settled with the U.S. Securities and Exchange Commission (SEC) for more than $4 billion and its CEO stepped down in 2024, centralized exchanges (CEXs) lost some share to decentralized exchanges (DEXs). From January to their peaks, spot and derivatives DeFi volume nearly doubled their market share from 8% and 3%, respectively, to 14% and 8%, reaching all-time highs. During the same period, Binance’s market share among centralized exchanges dropped from 62% to 35%. As Decentralized Finance (DeFi) Surged To All-Time Highs, Decentralized Exchanges Challenged Centralized Venues In Both Spot And Derivatives Trading 0% 2% 4% 6% 8% 10% 12% 14% 16% Jan-20 Apr-20 Jul-20 Oct-20 Jan-21 Apr-21 Jul-21 Oct-21Jan-22 Apr-22 Jul-22 Oct-22 Jan-23 Apr-23 Jul-23 Oct-23 Jan-24 Apr-24 Jul-24 Oct-24 DEX Market Share By Trading Volume Spot Trading Volume Derivatives Trading Volume 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Jan-23 Mar-23 May-23 Jul-23 Sep-23 Nov-23 Jan-24 Mar-24 May-24 Jul-24 Sep-24 Nov-24 Share Centralized Exchange Market Share By Volume Binance Others Source: ARK Investment Management LLC, 2025, based on data from The Block as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 61 ARK’s Big Ideas 2025: Scaling Blockchains

• 62. Decentralized exchanges (DEXs) like Uniswap, Aerodrome, and Raydium leverage the efficiency of small, agile teams to develop and maintain core protocol infrastructure. With massive efficiency advantages over centralized venues, DEXs have one-tenth the headcount. With ~9,000 employees, Binance leads the centralized exchange space headcount. Smart Contract-Powered Exchanges Are 5-10 Times More Efficient Than Their Centralized Counterparts *The Spot Volume per Employee data are from the month of November 2024. Source: ARK Investment Management LLC, 2025, based on data from Token Terminal, Coin Gecko, and Pitchbook as of November30, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 0 500 1000 1500 2000 2500 OKX Coinbase Crypto.com Binance Bybit Orca Uniswap Aerodrome Pancakeswap Raydium $ Millions Monthly Spot Volume Per Employee* 62 Centralized Exchanges Decentralized Exchanges ARK’s Big Ideas 2025: Scaling Blockchains

• 63. Liquid staking and restaking have become the preferred methods for generating yield on Ether. Thanks to their yield-bearing properties, liquidity, and accessibility, they now account for 40% of ETH staking. Demand for rehypothecating staked ETH to increase yield has created restaking platforms that now account for ~5.5 million Ether, or 17% of staked ETH. Liquid Staking And Restaking Protocols Are The Preferred Solutions For Staking And Rehypothecating ETH 0 1 2 3 4 5 6 Jun-23 Jul-23 Aug-23 Sep-23 Oct-23 Nov-23 Dec-23 Jan-24 Feb-24 Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Dec-24 Millions ETH Staked In Leading Restaking Platforms Eigenlayer Symbiotic Karak Note: Staking involves locking up a cryptocurrency in a wallet to support the operations of a blockchain network, typically for Proof-of-Stake (PoS) or similar consensus mechanisms. Liquid staking is an enhancement of traditional staking where users can stake their tokens while still having access to liquidity. Restaking refers to using the same staked assets or their derivatives for multiple purposes or layers of staking. Rehypothecation occurs when assets (such as staked tokens or derivatives) are used as collateral multiple times by various entities. Source: ARK Investment Management LLC, 2025, based on data from Dune Analytics as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 63 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Jan-21 Mar-21 May-21 Jul-21 Sep-21 Nov-21 Jan-22 Mar-22 May-22 Jul-22 Sep-22 Nov-22 Jan-23Mar-23 May-23 Jul-23 Sep-23 Nov-23 Jan-24 Mar-24 May-24 Jul-24 Sep-24 Nov-24 Share Ethereum Staking Methods Liquid Staking CEXs Unidentified Staking Pools Liquid Restaking Solo Stakers ARK’s Big Ideas 2025: Scaling Blockchains

• 64. Led By Polymarket, Prediction Markets Became The Breakout Consumer Application In 2024 During 2024, Polymarket surpassed three million cumulative users and $1.2 billion in monthly volume, with elections and politics driving 70% of its activity. After a brief post-election pull-back, the number of unique daily users has recovered to 50,000 and volume is trending back up with Sports being the most important category. 0 0.5 1 1.5 2 2.5 3 3.5 0 10 20 30 40 50 60 Jan-24 Feb-24Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Dec-24 Millions Thousands Number Of Unique Addresses Betting On Polymarket Unique Addresses (lhs) Cumulative Unique Addresses (rhs) $0 $200 $400 $600 $800 $1,000 $1,200 20-May-24 3-Jun-24 17-Jun-24 1-Jul-24 15-Jul-24 29-Jul-24 12-Aug-24 26-Aug-24 9-Sep-24 23-Sep-24 7-Oct-24 21-Oct-24 4-Nov-24 18-Nov-24 2-Dec-24 16-Dec-24 $ Millions Polymarket Volume By Category Elections Politics Unspecified Sports Science Pop Culture Crypto Business Source: ARK Investment Management LLC, 2025, based on data from Dune Analytics as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 64 ARK’s Big Ideas 2025: Scaling Blockchains

• 65. 0 100 200 300 400 500 600 700 800 900 Oct-23 Nov-23 Dec-23 Jan-24 Feb-24Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Dec-24 Number Transactions Per Second Solana Ethereum + L2s Thanks to Layer 2s and the EIP 4844 upgrade, Ethereum’s throughput has more than doubled to ~200 transactions per second (TPS). Nonetheless, Solana continues to outperform, with an average throughput of ~800TPS. Solana’s success is based on a set of trade-offs, including more expensive hardware requirements and parallel transaction processing. Solana’s new Firedancer client could increase its throughput to hundreds of thousands of TPS. Solana’s Throughput Is Higher Than Ethereum’s, Thanks To High-Performance Design And Strategic Trade-Offs EIP 4844 Upgrade Source: ARK Investment Management LLC, 2025, based on data from Artemis Terminal as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 65 ARK’s Big Ideas 2025: Scaling Blockchains

• 66. After hitting bear market lows at $8 in 2023, Solana has turned around dramatically relative to other Layer 1s. Daily active users, revenue, transaction count, and total value locked (TVL) reached all-time highs or grew by an order of magnitude. Solana is the only Layer 1 that competes with Ethereum and Bitcoin on metrics like daily active addresses and revenue. Thanks To Retail Adoption, Solana Has Gained Share Based On Several Metrics 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Oct-23 Dec-23 Feb-24 Apr-24 Jun-24 Aug-24 Oct-24 Dec-24 Share Daily Active Addresses Solana Bitcoin Ethereum + L2s 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Sep-23 Oct-23 Nov-23 Dec-23 Jan-24 Feb-24Mar-24 Apr-24 May-24 Jun-24 Jul-24 Aug-24 Sep-24 Oct-24 Nov-24 Share Fees Generated Solana Bitcoin Ethereum + L2s Source: ARK Investment Management LLC, 2025, based on data from Artemis Terminal as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 66 ARK’s Big Ideas 2025: Scaling Blockchains

• 67. Of the 39,139 new crypto developers in 2024, Solana led the way with 7,625, surpassing Ethereum Mainnet. With 4,287 developers in total, Base was sixth overall, surpassing both Arbitrum and Starknet as the leading Layer 2 solution on Ethereum. Solana And Base Are Leading The Way In Developer Adoption And Mindshare 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Solana Ethereum Other Internet Computer Aptos Base Bitcoin Sui Near Polkadot Polygon Starknet Arbitrum Number Of Net New Crypto Developers In 2024 0 2000 4000 6000 8000 10000 12000 14000 16000 Ethereum Mainnet Base Arbitrum Starknet Optimism Scroll Zksync Polygon Zkevm Number Of Crypto Developers In The Ethereum Ecosystem In 2024 Source: ARK Investment Management LLC, 2025, based on data from Electric Capital 2024 as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security or cryptocurrency. Past performance is not indicative of future results. 67 ARK’s Big Ideas 2025: Scaling Blockchains

• 68. Robotaxis Transforming Personal Mobility While Lowering Costs And Enhancing Safety Tasha Keeney, CFA Daniel Maguire, ACA DIRECTOR OF INVESTMENT ANALYSIS & INSTITUTIONAL STRATEGIES RESEARCH ANALYST, AUTONOMOUS TECHNOLOGY & ROBOTICS 68 ARK’s Big Ideas 2025

• 69. Electric Vehicles Continue To Take Share From Internal Combustion Engine Vehicles 4% 2% -2% -5% -16% -1% -10% 4% -1% 57% 67% 70% 18% 33% 113% 59% 30% 12% -40% -20% 0% 20% 40% 60% 80% 100% 120% 2016 2017 2018 2019 2020 2021 2022 2023 2024e Year-Over-Year Percent Change Global Vehicle Sales Growth Internal Combustion Battery Electric 10 20 30 40 50 60 70 80 90 100 2016 2017 2018 2019 2020 2021 2022 2023 2024e Units (Millions) Global Vehicle Sales Internal Combustion Battery Electric Slide by Sam Korus, Director and Akaash TK, Associate, Autonomous Technology & Robotics. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including EVVolumes as of January 3, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 69 ARK’s Big Ideas 2025: Robotaxis

• 70. The Future Is Autonomous Electric Vehicles Despite the recent slowdown in the growth of electric vehicles, the trend is clear: the internal combustion engine is on its way out. Slide by Sam Korus, Director and Akaash TK, Associate, Autonomous Technology & Robotics. Note: Percentages may not total 100% due to rounding. Source: ARK Investment Management LLC, 2025, based on data from EV-Volumes as of January 3, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. 70 92% 87% 81% 77% 72% 4% 5% 6% 7% 8% 1% 2% 4% 5% 7% 3% 6% 10% 12% 13% 50 100 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024e Unit (Millions) Global Vehicle Sales And Share By Segment Internal Combustion Hybrid Plug-in Hybrid Battery ARK’s Big Ideas 2025: Robotaxis

• 71. After More Than A Decade, Most Automakers Have Dropped Out Of The Autonomy Race. Tesla And Waymo Remain. ARK’s 2014 Estimates Of Autonomous Commercial Launch In 2014, many automakers expected to debut an autonomous vehicle by 2020. Only Waymo delivered, launching its first commercial autonomous rides in 2018. Tesla plans to launch in 2025. Waymo’s launch and Tesla’s internal testing of robotaxis suggest that 2025 could be the standout year in which consumers and businesses agree that the future of transportation is autonomous. Outcome Tesla Volvo Google GM Audi Mercedes Nissan BMW Daimler Continental 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2025 LAUNCH PLANNED LAUNCHED IN 2018 NO EXPECTATION OF LAUNCH NO EXPECTATION OF LAUNCH NO EXPECTATION OF LAUNCH NO EXPECTATION OF LAUNCH NO EXPECTATION OF LAUNCH NO EXPECTATION OF LAUNCH NO EXPECTATION OF LAUNCH NO EXPECTATION OF LAUNCH Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 71 ARK’s Big Ideas 2025: Robotaxis

• 72. $2.10 $1.10 $1.10 $1.10 $0.25 1871 1934 1950 2016 2021 Cost Per Mile Of A Personally Owned Vehicle (2024 $) Autonomous Ride-Hail Should Increase Access To Convenient Point-To-Point Transportation Adjusted for inflation, the cost of owning and operating a personal car has not changed since the Model T rolled off the first assembly line more than 100 years ago. ARK estimates that autonomous taxis at scale could cost consumers as little as $0.25 per mile, spurring widespread adoption. Note: Figures are rounded. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Ulvog 2012, Model T Ford Forum 2010, American Automobile Association 1950, and American Automobile Association 2024, as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 72 2024 2035 ARK’s Big Ideas 2025: Robotaxis

• 73. Autonomous Ride-Hail Will Lower The Cost Of Point-To-Point Transportation In early days, robotaxi companies are likely to price rides close to those for human-driven ride-hail. With scale and increased utilization, autonomous technology is likely to drive the cost per mile below that of personal vehicle travel. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 73 $2.00 $1.10 $0.35 $0.25 Human Driven Ride-Hail US, 2024 Personal Car US, 2024 Human Driven Ride-Hail China, 2024 Robotaxi Global, 2035 Cost Per Mile (2024 $) Human-Driven Ride-Hail US, 2024 ARK’s Big Ideas 2025: Robotaxis

• 74. The US And China Are Blazing The Robotaxi Trail, Followed By The Middle East And Europe 74 United States • Tesla • Waymo • Zoox • Wayve • Pony.ai • WeRide China Saudi Arabia & UAE United Kingdom • Wayve • Baidu Apollo Go • Tesla • Pony.ai • WeRide • Pony.ai • Baidu Apollo Go • WeRide Price competition in China could continue to push local players into more lucrative partnerships abroad. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. ARK’s Big Ideas 2025: Robotaxis

• 75. Robotaxis Already Have Hit 12 Million Rides At An Annual Rate Waymo and Baidu are responsible for most of the autonomous rides in the US and China, respectively, as shown on the left below. With the largest data lake of all players globally, its proprietary data advantage should give Tesla a competitive edge if it launches ride-hail operations as planned in 2025. 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 Waymo Baidu Apollo Go Pony.ai Tesla Zoox Million Rides (Annualized) Autonomous Commercial Rides Run Rate 75 ~ 37 ~ 16 ~1 ~ 3,500 * ~0.1 - 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 Waymo Baidu Apollo Go Pony.ai Tesla Zoox Million Miles (Annualized) Autonomous Miles Run Rate Baidu Apollo Go *Tesla Full Self-Driving (FSD) currently is supervised and requires a human behind the wheel. Waymo opened driverless operations to the pubic in October 2020 and now completes ~175,000 weekly rides. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Krafcik 2020, as of January 29, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Robotaxis

• 76. Theoretically, Tesla Alone Could Hit ARK’s Robotaxi Industry Forecast In 2030 By 2030, ARK’s research suggests that the number of robotaxis will have scaled to roughly 50 million globally, with Tesla at ~50% market share. Competitors could face barriers to scale, given the intense price competition in China and the lack of alignment between technology providers and traditional automakers. Separately, if it were to allocate 100% of its future vehicle production to robotaxi service—instead of the ~30-40% assumed—Tesla alone could hit ARK’s robotaxi industry forecast in 2030. - 10 20 30 40 50 60 2024 2025 2026 2027 2028 2029 2030 Autonomous Vehicle Fleet (Millions) Global Autonomous Vehicle Adoption Forecast Tesla Robotaxis Other (Waymo, Baidu Apollo Go, etc.) 76 Note: Autonomous adoption curve overestimates figures in initial years. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Robotaxis

• 77. Autonomous Electric Vehicles Could Rapidly Shift Vehicle Miles Traveled (VMT) To Electric Autonomous electric vehicles could exceed 50% utilization, pushing a disproportionate number of miles traveled from gas-powered vehicles to electric drivetrains. 0% 5% 10% 15% 20% 25% 30% 35% 40% 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 Share Global Autonomous Electric Miles As A Share Of Total Miles Electric Vehicle Miles Traveled Without Autonomous Forecasted Electric Vehicle Miles Traveled Without Autonomous Autonomous Electric Vehicle Miles Traveled Forecasted Autonomous Electric Vehicle Miles Traveled Slide by Sam Korus, Director, Daniel Maguire, Analyst, and Akaash TK, Associate, Autonomous Technology & Robotics. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 27, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 77 ARK’s Big Ideas 2025: Robotaxis

• 78. Wright’s Law Suggests That A $15,000 Robotaxi Is Possible $1 $10 $100 $1,000 $10,000 100,000 1,000,000 10,000,000 100,000,000 1,000,000,000 10,000,000,000 $/kWh Cumulative kWh Battery Cost Decline Nickel Cells Nickel Forecast LFP Cells LFP Forecast Battery Cost ($/kWh) $100 $75 $50 Miles/kWh 4 $6,250 $4,688 $3,125 4.5 $5,556 $4,167 $2,778 5 $5,000 $3,750 $2,500 5.5 $4,545 $3,409 $2,273 Assumes a 250-mile range vehicle Tesla has suggested that the Cybercab will have an efficiency of 5.5 miles/kWh. Paired with the battery cost decline anticipated by Wright’s Law, that efficiency suggests that the battery could cost as little as $2,300. Typically, the drivetrain of a vehicle is ~20% of the total vehicle cost, which suggests a Cybercab could be profitable at a price of $15,000 or below. Slide by Sam Korus, Director of Research, Autonomous Technology & Robotics Research. Note: Wright’s Law states that for every cumulative doubling of units produced, costs will fall by a constant percentage. See Winton 2019. “kWh”: Kilowatt-hour, a unit of energy, representing the use or generation of 1 kilowatt of power for 1 hour. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 78 ARK’s Big Ideas 2025: Robotaxis

• 79. Tesla’s Cost Per Mile Is ~30-40% Lower Than Waymo’s $- $0.20 $0.40 $0.60 $0.80 $1.00 $1.20 Waymo 5th Gen (Early Launch) Model 3 (Early Launch) Waymo 6th Gen (At Scale) Cybercab (At Scale) Robotaxi Incremental Cost Per Mile Depreciation Operations Support Insurance Energy Maintenance & Tires Parking Tax Its higher cost structure is likely to make Waymo less competitive or less profitable than a Cybercab robotaxi service. Waymo’s dependence on LIDAR and higher cost auto manufacturers are two drags on its competitive positioning. - 50,000 100,000 150,000 200,000 250,000 300,000 350,000 Waymo Hyundai Ioniqs Tesla Cybercabs Number Of Vehicles Vehicles Per $5 Billion In Capex Note: “Capex”/“Capital Expenditure” is a one-off upfront cost required to build or deploy an asset. “Early launch” assumes ten cars per remote operator with utilization rates roughly equivalent to ride-hail today. “At scale” assumes 100 cars per remote operator with an improved autonomous operation utilization rate. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 79 ARK’s Big Ideas 2025: Robotaxis

• 80. $- $0.05 $0.10 $0.15 $0.20 $0.25 Robotaxi Forecast Cost Per Mile Fees Robotaxi Platform Take Depreciation Parking Energy Maintenance & Tires Deadhead Cost Insurance $- $0.20 $0.40 $0.60 $0.80 $1.00 $1.20 $1.40 $1.60 $1.80 $2.00 Human Ride-Hail Cost Per Mile Uber Take Booking Fee (Ex Insurance) Driver Take Deadhead Cost Maintenance & Tires Depreciation Fuel Insurance Financing Parking Autonomous Ride-Hail Should Reduce Costs, Including Deadhead Miles Autonomous electric travel should reduce costs, including fuel, insurance, and deadhead miles, thanks to higher utilization, increased safety and efficiency, and electric drivetrains. Deadhead miles are the number of miles driven without passengers, which account for 45% of ride-hail miles today. Autonomous vehicles could reduce deadhead miles by removing the human-in-the-loop and harnessing AI tools to predict demand and traffic patterns. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 80 Fees Robotaxi Platform Take ARK’s Big Ideas 2025: Robotaxis

• 81. Tesla Expects Its Autonomous Software To Outperform Human Drivers On Safety Soon, Tesla expects more than 10,000 miles between critical interventions, or roughly one year’s worth of human driving before the average person would need to touch the wheel in Full Self-Driving (FSD) mode. Later in 2025, Tesla plans to outperform the national accident rate—or ~700,000 miles between critical interventions—paving the way for conversations with regulators. In our view, Tesla could launch a robotaxi service with remote operators before hitting that milestone. Waymo’s commercial robotaxi service has surpassed 400,000 miles between police-reported collisions—impressive performance for a fleet of its size. 10 1,000 10,000 National Average Human Accident Rate (700,000) 700,000 One Year Of Personal Driving (10,000) Waymo Police Reported Collisions (400,000) 1 10 100 1,000 10,000 100,000 1,000,000 16,214 66,736 120,000 190,000 (Log Scale) Compute In H100 Equivalent GPUs Tesla Miles Between Critical Intervention Jan 24 11.4.8.1 Oct 24 12.5.4.1 100x Mar 25 13.4.5 10x Oct 25 13.? 70X Date FSD Version Improvement 81 Note: “Miles Between Critical Intervention” measures the distance traveled before a human driver must take control to avoid an accident. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Robotaxis

• 82. Ride-Hail Could Create A ~$10 Trillion Market $0.00 $0.50 $1.00 $1.50 $2.00 $2.50 $3.00 $3.50 $4.00 $4.50 Price Per Mile Ride-Hail Addressable Market* $4B 4.00 3.00 2.00 1.10 0.60 0.50 0.25 $30B $100B $1T: Addressable ridership in Western markets at ~$1 $2.4T: Non-commuting miles in higher income countries priced at ~$0.60 per mile $1.92T: Long tail of demand priced at human driven ridehail prices of $0.35 per mile in lower income countries $5T: Low cost, accessible autonomous travel at $0.25 per mile $134B: Existing addressable market for ride-hail companies charging $2-$4 per mile in Western Markets 1 5 10 0 30 Price Points ($): 0.35 At $0.25 cents per mile, autonomous transportation could serve a much larger population than human-driven ride-hail does today. Meanwhile, based on how consumers value their time, demand could be significant even at higher price points. Miles (Trillions) 82 *$10 trillion is the addressable market, not the revenue we expect in 2030, as we do not expect autonomy to penetrate all addressable miles. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Robotaxis

• 83. 500 40 500 1,400 200 1,000 2,600 2,200 33,600 $0 $5,000 $10,000 $15,000 $20,000 $25,000 $30,000 $35,000 $40,000 Revenue (Net) EBIT Enterprise Value Billions Forecasted Revenue, Earnings, And Enterprise Value In 2030 Autonomous Electric Auto Manufacturers Fleet Owners Autonomous Platform Providers Robotaxis Could Generate ~$34 Trillion In Enterprise Value By 2030 Multiples: 13X EBIT 5X EBIT 15X EBIT 83 Note: Numbers are rounded. “EBIT”: Earnings Before Interest and Tax. This slide depicts net revenue in 2030, the portion of gross revenue that the platform operator receives, while the previous slide depicts the total addressable market. Source: ARK Investment Management LLC, 2025This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Robotaxis

• 84. Cutting Costs, Revolutionizing Supply Chains, And Reshaping Consumer Behavior Autonomous Logistics Tasha Keeney, CFA Daniel Maguire, ACA DIRECTOR OF INVESTMENT ANALYSIS & INSTITUTIONAL STRATEGIES RESEARCH ANALYST, AUTONOMOUS TECHNOLOGY & ROBOTICS 84 ARK’s Big Ideas 2025

• 85. $5.40 $0.35 Human-Driven App Delivery Forecasted Drone Delivery Local Small Item Delivery Cost (Per Trip) Autonomous Vehicles That Roll And Fly Could Lower Supply Chain Costs Dramatically $2.40 $0.40 Human-Driven Delivery Rolling Integrated Traffic Robot Delivery Local Batch Delivery Cost (Per Trip) Autonomous vehicles should operate at higher utilization rates than human-in-the-loop systems, creating more costeffective last-mile delivery ecosystems. -83% -94% $0.07 $0.03 $0.00 $0.01 $0.02 $0.03 $0.04 $0.05 $0.06 $0.07 $0.08 Human-Driven Diesel Truck Forecasted Autonomous Electric Truck Truckload Delivery Cost (Per Ton-Mile) -57% Human-Driven Vehicle Forecasted Robot Delivery ARK’s Big Ideas 2025: Autonomous Logistics $6.00 $5.00 $4.00 $3.00 $2.00 $1.00 $0.00 85 Note: We have updated our drone price per mile with our latest assumptions for replacement costs, launching and charging infrastructure, insurance, and labor costs. Fees for drone and robot delivery are shown net of infrastructure costs (outside of charging and launch/land), which we believe either could be borne by the drone or robot delivery operators or shared with logistics or retail partners. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon.

• 86. Robotrucks And Drones Face More Hurdles To Commercialization Than Robotaxis Robotaxis Robotrucks Drones Technical Hurdles Regulatory Hurdles Customer Partnerships Manufacturing Partnerships Data Collection/ Testing/Validation Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 86 ARK’s Big Ideas 2025: Autonomous Logistics

• 87. While Tesla has generated billions of miles of customer data for training, Waymo and Baidu have access only to millions of miles. In trucking, autonomous players have taken a route-by-route approach, building their data libraries as they add more routes. For drones, while many flights have occurred in Africa, companies continue to test with partners in the US and Europe, with limited services offered, thus far. In the robotaxi industry, Tesla is vertically integrated while companies like Waymo and Baidu have partnered with Geely and Hyundai/BAIC, respectively. In the trucking space, players like Aurora have partnered both with OEMs* like Paccar and Volvo and with integrators like Continental. Meanwhile, many drone providers manufacture in-house but have not hit mass manufacturing scale. Robotaxis sell rides to end consumers, while robotrucking and drone companies need to forge partnerships with retailers, food delivery providers, and other logistics operators. Robotaxis were first to commercialize, but most robotrucks still require a human operator, with a number of companies pushing to remove the driver this year. Meanwhile, drones have been able to fly autonomously for a number of years: the difficulty lies more in logistics than on the autonomous flight technology. Waymo and Baidu have paved the regulatory path for robotaxis in the US and China, respectively. In the US, state regulators could give way to federal legislation in 2025. Trucking is influenced by lobbyists, while drones are regulated by the FAA, which historically has missed many of its own timelines for developing a framework for drone delivery. Manufacturing Partnerships: Data/Testing/Validation: Technical: Regulation: Customer Partnerships: Hurdles To Commercialization * ”OEM”: Original Equipment Manufacturer. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. 87 ARK’s Big Ideas 2025: Autonomous Logistics

• 88. Company Region Of Operations Customer US Regulatory Approval* Cumulative Number Of Commercial Flights Zipline US Africa Hospital & Consumer Part 135 (Paid) Wing US Australia EU UK Hospital & Consumer Part 135 (Paid) Meituan China UAE Consumer N/A Manna US EU Hospital & Consumer Part 107 (Unpaid) FlyTrex US Consumer Part 107 (Unpaid)^ Matternet US EU UAE Hospital & Consumer Part 107 (Unpaid)^ Amazon US Consumer Part 135 (Paid) 5,000 e 50,000 100,000 280,000 e 400,000 400,000 1,320,000 Logistics Drone Companies Are Beginning To Break Through Regulatory Barriers • Thus far, most drone deliveries have taken place in rural areas outside the US to serve unmet medical needs. • Companies with more real-world data and regulatory approvals to fly beyond-visual-line-of-sight without observers will pull away from the competition. 88 *In the US, paid drone delivery requires Federal Aviation Administration (FAA) Part 135 certification; Part 107 allows operation but prohibits compensation. That said, the FAA is working on updating drone regulation to standardize beyond visual line of sight operations for commercial drones in the US under Part 108. ^FlyTrex and Matternet have Part 135 partners. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 17, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. ARK’s Big Ideas 2025: Autonomous Logistics

• 89. 1,520,000 2,200,000 3,100,000 3,500,000 125,000,000 Gatik Aurora Pony.ai Kodiak Cumulative Number Of Miles Traveled • In the US, commercial autonomous trucking requires a safety driver today. • China leads in generating real-world driving data, but companies in the US should advance considerably once regulators approve the removal of safety drivers in 2025. Autonomous Trucking Companies With The Most Real-World Miles Should Be Best Equipped For Scale Inceptio Technology Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. 89 ARK’s Big Ideas 2025: Autonomous Logistics

• 90. Company Countries Sidewalk Or Road? College Campus Or Beyond? Number Of Deliveries Completed Starship Technologies US GB DE FI EE Sidewalk Mostly Campus Neolix CN Road Beyond Meituan CN Road Beyond Coco* US FI Sidewalk Beyond Avride KR US UAE Sidewalk Beyond Cartken JP US Sidewalk/ Indoor Mostly Campus Serve Robotics US Sidewalk Beyond Robot Delivery Companies With Broad-Based Experience Should Be In A Superior Competitive Position • In the US, most sidewalk deliveries are on college campuses, which do not seem to be productive training ground for urban and residential areas. • In China, road robots are commonplace. 50,000 200,000 e 200,000 500,000 4,000,000 6,000,000 7,000,000 90 *Coco has evolved from human-driven to hybrid robots and is more reliant on humans than are its competitors. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 17, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. ARK’s Big Ideas 2025: Autonomous Logistics

• 91. Autonomous Delivery Revenue Could Approach ~$900 Billion Globally In 2030 280 160 420 Robots & Drones Last Mile Trucks Middle Mile & Over-The-Road Forecasted Autonomous Delivery Revenue ($ Billions, 2030) Parcels Food All Goods Technology-enabled delivery is likely to reshape consumer buying habits, as the cost of transporting goods falls. Our research suggests that robot/drone delivery fees and autonomous trucking revenues could reach ~$440 billion and ~$420 billion, respectively, in 2030. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 17, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 91 ARK’s Big Ideas 2025: Autonomous Logistics

• 92. The US Needs Low-Cost, AI-Powered Aircraft • The current state of warfare has proven that high-cost aircraft are unsustainable against low-cost internationally made drones. • Meanwhile, AI is improving decision-making and efficiency, reducing the need for human troops. “Smaller And Cheaper AI-controlled Unmanned Jets Are The Way Ahead.” US F-16 ~$30-60 million Chinese DJI ~$2,000-20,000 US SM-2 Missile ~$2 million Ukraine Red Sea: Iranian Shahed 136 ~$20,000 Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. “Replicator will galvanize progress in the too-slow shift of U.S. military innovation to leverage platforms that are small, smart, cheap, and many.” “We need them potentially at very large scale and very quickly.” “manned fighter jets are obsolete in the age of drones...” “I want to turn the Taiwan Strait into an unmanned hellscape…” “If we're shooting down a $50,000 one-way drone with a $3 million missile, that's not a good cost equation." Bill LaPlante, Under Secretary of Defense for Acquisition and Sustainment Frank Kendall, 26th Secretary of the Air Force Doug Bush, Army Acquisition Chief, on buying drones in large quantities Kathleen Hicks, Deputy Secretary of Defense Samuel Paparo, US Indo-Pacific Command Chief Navy Admiral Elon Musk 92 ARK’s Big Ideas 2025: Autonomous Logistics

• 93. Powering The Artificial Intelligence Revolution Energy Sam Korus Daniel Maguire, ACA DIRECTOR OF RESEARCH AUTONOMOUS TECHNOLOGY & ROBOTICS RESEARCH ANALYST AUTONOMOUS TECHNOLOGY & ROBOTICS Akaash TK RESEARCH ASSOCIATE AUTONOMOUS TECHNOLOGY & ROBOTICS 93 ARK’s Big Ideas 2025

• 94. 50 100 150 200 250 300 350 400 ARK Forecast: Incremental Power Necessary For AI Hardware 2030 Power Capacity Added In China 2023 GW Incremental Power Necessary For AI Hardware In 2030 VS. Power Capacity That China Added In 2023 Energy Production Is Important To Economic Growth According to our research, China added more power capacity in 2023 than would be necessary to meet our estimate of incremental power demand from AI data centers through 2030. In other words, the buildout rate necessary to accommodate AI power demand is achievable. 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 20052007 2009 2011 2013 2015 2017 2019 2021 2023 2025e 2027e 2029e Billion kWh Global Electricity Generation Global Electricity Generation Forecast Of Global Electricity Generation Extrapolated From 5 Year Trailing CAGR Forecast Of Global Electricity Generation Including That Necessary For AI Hardware Demand Note: “CAGR”: Compound Annual Growth Rate. “kWh”: Kilowatt-hour, a unit of energy, representing the use or generation of 1 kilowatt of power for 1 hour. “GW”: Gigawatt, a unit of power equal to one billion watts. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. Power Capacity Added In China 2023 Incremental Power Necessary For AI Hardware In 2030e 94 ARK’s Big Ideas 2025: Energy

• 95. Nuclear Energy Regulation Ended The Declines In Electricity Prices 1893 1944 1974 2021 $0.10 $1.00 $10.00 1 10 100 1,000 10,000 100,000 1,000,000 10,000,000 $/kWh 2023 Dollars Cumulative Megawatts US Electricity Prices Informed by Wright’s Law,* ARK’s research indicates that, apart from WWII, US electricity prices fell consistently from the late 1800s until 1974. The U.S. Energy Reorganization Act, enacted in 1974, reversed the drop in nuclear construction costs. $1 $10 $100 $1,000 $10,000 $100,000 1 10 100 1,000 10,000 100,000 1,000,000 Overnight Construction Cost (2023 $/kW) Cumulative Power Deployed (MW) United States & Japan Nuclear Construction Cost Trends (Log Log) U.S. Before Significant Regulatory Change (1953 - 1974) U.S. Post Significant Regulatory Change (1975-2023) Japan Before Significant Regulatory Change (1961-1977) Japan Post Significant Regulatory Change (1978-2009) Japan Nuclear Reactor Regulation Revision (1978) U.S. Energy Reorganization Act (1974) Note: “MW”: Megawatt, a unit of power equal to one million watts. “kW”: Kilowatt, a unit of power equal to one thousand watts. “kWh”: Kilowatt-hour, a unit of energy, representing the use or generation of 1 kilowatt of power for 1 hour. *Wright’s Law states that for every cumulative doubling of units produced, costs will fall by a constant percentage. See Winton 2019. Source: ARK Investment Management LLC, 2025, based on data from Smil 2000 and Cleveland 2023 (left chart) and Lovering et al. 2016 (right chart). For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 95 ARK’s Big Ideas 2025: Energy

• 96. 20 40 60 80 100 120 140 160 2006200720082009201020112012201320142015201620172018201920202021202220232024 Transcripts Mentioning "Nuclear Energy" "Nuclear Energy" Mentions The Nuclear Tide Is Rising Interest in nuclear energy is surging as companies seek a clean, reliable power source to sustain AI initiatives and meet ambitious net-zero carbon goals. Microsoft in deal for Three Mile Island nuclear power to meet AI demand - Financial Times, Sept 20, 2024 Google orders small modular nuclear reactors for its data centers - Financial Times, Oct 14, 2024 Amazon buys stake in nuclear energy developer in push to power data centers - Financial Times, Oct 16, 2024 Fukushima Nuclear Accident Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Capital IQ Pro, as of December 18, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 96 ARK’s Big Ideas 2025: Energy

• 97. Supported by clear regulations, nuclear fission could be a cost-competitive, timely solution to increasing power demands globally. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% $2,000 $4,000 $6,000 $8,000 $10,000 $12,000 $14,000 Utility Solar Onshore Wind Natural Gas Combined Cycle Nuclear Capacity Factor $/kW Overnight Capital Cost 2024 Prices (lhs) 1974 Price (2024 $ lhs) Capacity Factor* (rhs) 1 2 3 4 5 Small Hydro Biomass Wind OnshoreWind Offshore Solar PV Nuclear (5MW) Vallecitos (1957) Years Average Commissioning Time For Small Scale Renewable Energy Projects (1-10 MW)^ Wind Onshore Wind Offshore 97 Nuclear Fission Projects Could Deliver Economics Superior To Other Renewable Energy Projects Note: In 1957, the Vallecitos Boiling Water Reactor (VBWR) in Sunol, California, was the first nuclear power plant privately owned and operated to supply electricity to a public utility grid. “MW”: Megawatt, a unit of power equal to one million watts. “kW”: Kilowatt, a unit of power equal to one thousand watts. ^Average across 48 countries between 2005-2022. *Capacity Factor is defined as the ratio of the electrical energy produced by a generating unit for the period of time considered to the electrical energy that could have been produced at continuous full power operation during the same period. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including World Nuclear Association 2024, Gumber et al. 2023, and Lovering et al. 2016, as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Energy

• 98. While Most Small Modular Reactor (SMR) Startups Will Fail, The Survivors Should Thrive SMRs are early stage and face significant execution risks. To triple nuclear capacity by 2050, the US must build large reactors, SMRs, and microreactors. Shorter construction timelines, modular design, and scalability could enable smaller reactors to meet near-term demand. Selecting from more than 30 US designs will be important to reigniting Wright’s Law and lowering electricity costs. 0 1 10 100 1,000 10,000 Net Power In Megawatts Electric (MWe) Size Of US Advanced Reactors (Log Scale) NRC Approved Designs Designs Under Development SMR Micro Gen III Gen III+ Gen IV Large Advanced Note: “NRC”: U.S. Nuclear Regulatory Commission. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Small Modular Reactor (SMR) Global Tracker as of November 14, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 98 Global SMR Projects Early Stage Under Construction Operable Number 0 10 20 30 40 50 60 70 80 90 100 Global SMR Projects Global SMR Projects Under Construction ARK’s Big Ideas 2025: Energy

• 99. Nuclear Fusion Faces Significant Hurdles To Commercialization JET 1997 Forecasted SPARC 2027 Forecasted ITER 2039 1E+12 1E+13 1E+14 1E+15 1E+16 1E+17 1E+18 1E+19 1E+20 1E+21 1E+22 1950 1970 1990 2010 2030 Triple Product (Log Scale) [Density x Time x Temperature] Fusion Project Progress Toward Scientific Breakeven* Scientific Breakeven Q=1 (D-T Fuel) Nuclear Fission Milestones Demonstration Of SelfSustaining Nuclear Fission 1942 First Nuclear Power Plant To Produce Usable Electricity 1951 15 Years Time From Key Milestone To Commercialization First Commercial Nuclear Power Plant In The US 1946 1957 Creation Of The U.S. Atomic Energy Commission Despite ~75 years of research, nuclear fusion has not generated net energy in a meaningful way. While private fusion companies are promising breakthroughs as early as 2027, commercialization could take another ~15 years, if fission history is any guide. Commercial Breakeven The fusion reaction generates surplus power to cover the cost of building and operating the plant. Engineering Breakeven The fusion reaction generates the same power as the reactor consumes. Scientific Breakeven The fusion reaction generates the same power as injected into the fuel. 99 Note: The ‘Triple Product” displayed on the Y axis above is a single metric that measures nuclear fusion progress by considering density, confinement time, and temperature. It's a very large number, so we use scientific notation, e.g., "1E+21," meaning 1 followed by 21 zeros. “SPARC”: Smallest Possible Affordable Robust Compact reactor. SPARC is a Tokamak design. “ITER”: International Thermonuclear Experimental Reactor. *Excluding Laser-Based Inertial Confinement. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Energy

• 100. Solar, wind, and stationary energy storage are just getting started. One solution will not meet all demands. Interestingly, Tesla accounts for ~19% of global energy storage today. Nuclear And Other Renewables Can Serve As Viable Solutions 20 40 60 80 100 120 140 160 180 2018 2019 2020 2021 2022 2023 2024 Gigawatt-Hours (GWh) Global Stationary Energy Storage Tesla Energy Storage Non-Tesla Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. 4% 8% 12% 16% 1,000 2,000 3,000 4,000 5,000 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024e Share (%) Terawatt-Hours (TWh) Global Renewable Energy Generation Wind (lhs) Solar (lhs) Percent of Total Generation (rhs) 100 Percent Of Total Generation (rhs) ARK’s Big Ideas 2025: Energy

• 101. $0.05 $0.10 $0.15 $0.20 $0.25 $0.30 30% 50% 70% 90% $/kWh Uptime Capacity Lowest Levelized Cost Of Energy For Desired Uptime Current Cost For Solar And Battery ARK's Estimated Future Cost For Solar And Battery As Battery Costs Decline, Intermittent Energy Sources Become Less Problematic Declining battery costs should make intermittent energy systems economically attractive with 100% uptime. ARK’s modeling suggests that energy-intensive industries will accelerate the pairing of renewables with energy storage. Bitcoin mining is a prime example, showcasing how energy-intensive industries with business models orthogonal to power generation can accelerate the shift to 100% renewable grid integration. Note: “kWh”: Kilowatt-hour, a unit of energy, representing the use or generation of 1 kilowatt of power for 1 hour. “MWh”: Megawatt-hour, a unit of energy, equal to 1,000 kilowatt-hours, representing the use or generation of 1 megawatt of power for 1 hour. Source: ARK Investment Management LLC, 2025, based on data from National Renewable Energy Laboratory as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. 101 Solar Installation Power (Megawatts) Battery System (MWh) Solar, Battery, Bitcoin Systems Required To Meet Grid Dispatch Targets At Constant Cost Of Electricity (Size Of Bubble = Bitcoin Mining Power) ARK’s Big Ideas 2025: Energy

• 102. 0 20 40 60 80 100 120 140 160 180 Gigawatt-Hours (GWh) Status Of US Energy Generation Capacity In Interconnection Queue (2009 VS. 2023) Solar Wind Gas Between 2009 and 2023, US interconnection approval wait times increased ~80%, from 31 months to 56 months. Quicker approvals could lower project costs and accelerate the scaling of energy-intensive projects. Regulatory Reform Should Boost All Renewables 20 40 60 80 2009 2011 2013 20152017 2019 2021 2023 Median Duration (Months) US Interconnection Wait Times For New Energy Generation Solar Wind Gas Combined In Queue/ Active Complete/ Withdrawn Operational In Queue/ Active Withdrawn Complete/ Operational 2009 2023 102 *An interconnection queue is a system used by grid operators to manage and prioritize requests from energy developers to connect new power generation projects to the electricity grid, involving assessments of each project's impact on grid reliability, planning for necessary infrastructure upgrades, and determining cost responsibilities for integrating new plants. Source: ARK Investment Management LLC, 2025, based on data from Lawrence Berkeley National Laboratory as of December 31, 2024. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. ARK’s Big Ideas 2025: Energy

• 103. 103 Decoupling Physical Labor From Output Robotics Sam Korus DIRECTOR OF RESEARCH AUTONOMOUS TECHNOLOGY & ROBOTICS Daniel Maguire, ACA RESEARCH ANALYST AUTONOMOUS TECHNOLOGY & ROBOTICS Akaash TK RESEARCH ASSOCIATE AUTONOMOUS TECHNOLOGY & ROBOTICS ARK’s Big Ideas 2025

• 104. Robots Should Boost Productivity And Transform Industries 104 2 4 6 8 10 12 14 16 Before Washing Machines After Washing Machines Hours Time To Do Laundry 0 2 4 6 8 10 12 14 Before Assembly Line After Assembly Line Hours Time To Manufacture A Car 10 20 30 40 50 60 70 80 Before Kiva Robots After Kiva Robots Minutes Time From Click To Ship At An Amazon Warehouse -87% -88% -78% Automation’s impact on productivity has transformed industries, with specialized robots as simple as washing machines reducing dramatically the time required for tasks. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. ARK’s Big Ideas 2025: Robotics

• 105. Generalizable Robotics Represent A $26+ Trillion Global Revenue Opportunity 105 ~2.3 Hours Of Unpaid Work Per Day ~2.8 Billion Working Age Population ~$12 Weighted Average Hourly Wage ~$13+ Trillion Opportunity ½ Value Attributed To Free Time VS. Paid Time Household Robotics Manufacturing Robotics ~$13+ Trillion Opportunity (Average Of The Green Cells) ARK Forecasts Global Manufacturing GDP At ~$32 Trillion In 2030 × = Revenue Opportunity* ($ Billions) = × × Productivity Uplift 10% 25% 50% 100% 200% 400% Take Rate 10% 320 800 1,600 3,200 6,400 12,800 20% 640 1,600 3,200 6,400 12,800 25,600 50% 1,600 4,000 8,000 16,000 32,000 64,000 Note: The $12 hourly wage is an estimate of average global wage, not US wage + benefits. *The cells highlighted in green represent what ARK believes to be a reasonable or likely outcome. ^We define “Take Rate” as the percentage of a transactions value that the business keeps. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Robotics

• 106. Humanoid Robots Are Debuting Around The World Source: ARK Investment Management LLC, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. 106 Why the human form factor? Key is that a humanoid robot is generalizable. While a wrench can tighten nuts better than a human hand can, it is not a generalizable tool. The human hand is generalizable, particularly in an environment built by and designed for humans. Tesla Optimus Sanctuary AI Phoenix Figure AI Figure 02 Unitree G1 Boston Dynamics Atlas Agility Digit Fourier GR-1 1x Technologies NEO Apptronik Apollo ARK’s Big Ideas 2025: Robotics

• 107. $- $250,000 $500,000 $750,000 $1,000,000 $1,250,000 0% 50% 100% 150% 200% Upfront Robot Cost Productivity Uplift Net Present Value (NPV) Of A Robot Operating The Same Number Of Hours As A Human Worker Over Ten Years Robot Price For NPV=0 Tesla Optimus Price Expectation At Scale 107 The Adoption Of Humanoid Robots Should Increase As Their Costs Fall And Their Productivity Increases A 100% productivity uplift is equivalent to replacing a human working one shift over ten years. Human Labor Costs On Average In The US Hourly Annually $92,420 $46, including benefits $924,200 ~20,000 10-Year Cumulative Total Hours ~$550,000, not including turnover Net Present Value Of Costs Not Worth Investing In A Humanoid Robot Worth Investing In A Humanoid Robot Note: Per hour salary based on Bureau of Labor Statistics Employer Costs For Employee Compensation press release on September 10, 2024: Average employer costs for all civilian workers = $46.21 per hour; Wages average = $31.80; average benefit costs = $14.41. We assume a 40-hour work week and a 50-week work year. A positive net present value in this calculation suggests that it would be worthwhile to invest in a humanoid robot at that upfront cost and productivity uplift. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Robotics

• 108. Jobs Are Bundles Of Tasks 108 What does a professional dishwasher do? $8 $14 $246 $50 $100 $150 $200 $250 $300 Dishwashing Machines Professional Dishwasher Pay Value of Unpaid Time Spent Washing Dishes At Home $ Billions Annual US Dishwashing Costs In a restaurant with a dishwashing machine, washing dishes is one of thirteen tasks required of a “dishwasher.” The home is not much different. The dishwasher is one of the most common “robots” in a house. According to our research, automating activities at home like loading and unloading dishes and cleaning countertops represents a $250 billion revenue opportunity in the US alone. Professional Pay For Dishwasher Value Of Unpaid Time Spent Washing Dishes At Home • Wash dishes, glassware, flatware, pots, and pans, using dishwashers or by hand. • Place clean dishes, utensils, and cooking equipment in storage areas. • Sort and remove trash, placing it in designated pickup areas. • Sweep and scrub floors. • Maintain kitchen work areas, equipment, and utensils in clean and orderly condition. • Clean garbage cans with water or steam. • Receive and store supplies. • Stock supplies like food and utensils in serving stations, cupboards, refrigerators, and salad bars. • Transfer supplies and equipment between storage and work areas, by hand or using hand trucks. • Clean and prepare various foods for cooking or serving. • Prepare and package individual place settings. • Load and unload trucks that deliver or pick up food and supplies. • Set up banquet tables. Tasks: Note: For the value of unpaid time spent washing dishes, we assume, based on a study, that people value their free time at half their wage. From the Bureau of Labor Statistics, we take a median wage of ~$35. The Bureau of Labor Statistics reports that people spend 0.65 hours per day on food prep and cleanup. We assume .22 hours is spent on cleaning. We then multiply by the working age population. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Robotics

• 109. Small Businesses Should Benefit Disproportionately From Humanoid Robots Large manufacturing firms are organized by specialized and automated tasks. Specialization and automation give large firms the wherewithal to scale significantly and, in turn, lower labor costs as a share of revenue. Consequently, and somewhat counter-intuitively, large companies typically pay higher wages than small firms because automating specific tasks boosts productivity in large firms more than in small firms. Because generalized automation solutions—those for multiple tasks—have not evolved as quickly as automation solutions for specific tasks, small firms typically have a disproportionate number of automatable-but-not-yetautomated tasks that would benefit from generalizable solutions like humanoid robots. Moreover, a high percentage of manufacturing employees in the US work in small firms. 1 2 3 4 5 6 <500 500-749 750-999 1,000-1,499 1,500-1,999 2,000-2,499 2,500-4,999 5,000-9,999 10,000-19,999 20,000+ Millions Firm Size By Number Of Employees US Manufacturing Employment Sample Small Firm Sample Large Firm Revenue $1,000 $5,000 Labor Share Of Revenue 40% 20% Number Of Employees 2 4 Wage Per Employee $200 ($1,000 x 40%/2) $250 ($5,000 x 20%/4) Tasks Per Employee 4 2 Benefit Of Automating A Single Task $50 ($200 x (1/4 tasks)) $125 ($250 x (1/2 tasks)) Valued Automation Solution Generalizable Single Task Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including U.S. Census Bureau 2023, as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 109 ARK’s Big Ideas 2025: Robotics

• 110. Automation Turns Non-Market Activity Into Revenue-Generating Activity And Gross Domestic Product (GDP) 110 Family Farmworkers Hired Farmworkers 1 2 3 4 5 6 7 8 Total Breakdown Number (Millions) Farm Worker Jobs “Lost” 1950-2000 Of the 82% who “lost" their jobs between 1950 and 2000, most farm workers were unpaid family members. 2 4 6 8 10 12 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 Number (Millions) Farm Workers Paid Workers Unpaid Family Workers Source: ARK Investment Management LLC, 2025, based on data from U.S. Department of Agriculture 2025 as of January 10, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. ARK’s Big Ideas 2025: Robotics

• 111. Consumers Continue To “Contribute” Unpaid Labor To The US Economy 111 If US consumers were to value their time at half their hourly wages, food preparation and cleanup would add ~$1 trillion to GDP annually. Food Cost Prep And Cleanup Cost $2 $4 $6 $8 $10 $12 $14 Home Cooked Fast Casual Cost Of A Meal Food Away From Home Food At Home Food Prep And Cleanup 500 1,000 1,500 2,000 2,500 2023 Addressable Market $ Billions Total US Food Cost ~$1 Trillion 200 400 600 800 1,000 1,200 1,400 1,600 1997 2000 2003 2006 2009 20122015 2018 2021 $ Billions US Food Expenditure Food At Home Food Away From Home Food Away From Home Less Delivery Home-Cooked Note: The Food Prep and Cleanup Cost is based on the following: The Bureau of Labor Statistics (BLS) reports that time spent on food prep and cleanup is .65 hours per day. Based on a study, we assume people value their free time at half their hourly wage; for this, we use BLS data of ~$35. We use the population above 18 years old of ~260 million. Source: ARK Investment Management LLC, 2025, based on data from U.S. Department of Agriculture 2025 and Franchisee Resource Center 2021 as of January 10, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Robotics

• 112. 3D Printing Revenues Could Grow ~40% At An Annual Rate To $180 Billion By 2030 20 40 60 80 100 120 140 160 180 200 2024 2025 2026 2027 2028 2029 2030 $ Billions 3D Printing Revenue Forecast • The 3D printing industry consolidated significantly in 2024, highlighted by Nano Dimension’s acquisition of Desktop Metal and Markforged. • If manufacturers decide to bring the technology in house, as did General Electric, consolidation could continue. • The Trump administration’s efforts to bolster US manufacturing could benefit 3D Printing. • Industries like drone manufacturing and nuclear power, which should scale rapidly, could use 3D printing to accelerate time to market. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. Slide by Tasha Keeney, Director of Investment Analysis & Institutional Strategies 112 ARK’s Big Ideas 2025: Robotics

• 113. 3D Printing Makes Parts For Nuclear Applications More Efficiently • Westinghouse uses 3D printing for nuclear debris filters. • Those components improved on the measure of resistance to debris by 30%, thanks to the design flexibility offered by 3D printing. • The US Navy engaged Australian additive manufacturing company AML3D to 3D-print tailpiece components for its nuclear submarine program. • AML3D produced those parts in fewer than five weeks, ~15X more quickly than traditional manufacturers. Source: ARK Investment Management LLC, 2025, based on data from Shaikhnag 2024 and Tyrer-Jones 2025 as of January 10, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. Slide by Tasha Keeney, Director of Investment Analysis & Institutional Strategies 113 ARK’s Big Ideas 2025: Robotics

• 114. Reusable Rockets Sam Korus DIRECTOR OF RESEARCH AUTONOMOUS TECHNOLOGY & ROBOTICS Akaash TK Daniel Maguire, ACA RESEARCH ANALYST AUTONOMOUS TECHNOLOGY & ROBOTICS RESEARCH ASSOCIATE AUTONOMOUS TECHNOLOGY & ROBOTICS 114 ARK’s Big Ideas 2025

• 115. SpaceX Has A Ten-Year Head Start On Reusable Orbital Rockets “Past experience with launch vehicle reusability has been mixed at best in terms of achieving sustainable cost reductions. So I am a skeptic with regard to many of the claims that have been made for cost reductions.” –David W. Thompson, CEO, Orbital ATK, 2016 “We don’t have any plans to look at reusability at all right now…The reality is that, for the next ten years, other than for Falcon, I don’t think any of us will look at this.” -Phil Slack, President of International Launch Services, 2014 “We have not really changed our assessment over the last couple of years because we have yet to see the other forms of reusability—flyback or propulsive return to Earth—demonstrate economic sustainability on a recurring basis.” –Tory Bruno, CEO, United Launch Alliance, 2020 2015 SpaceX Lands First Falcon 9 2017 SpaceX Launches 1st Flight Proven Booster 2024 No Orbital Reusable Rocket Besides SpaceX 2025+ Blue Origin Rocket Lab China Stoke Space Firefly Relativity 20 40 60 80 100 120 140 160 180 200 2010 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025e Number Falcon Launch Rate New Boosters Reused Boosters Forecast 115 Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Klotz 2020, de Selding 2016, and Faust 2014, as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Reusable Rockets

• 116. SpaceX Is Refurbishing Rockets In Record Time ARK’s Big Ideas 2025: Reusable Rockets When the Space Shuttle cost ~$1.5 billion per launch, industry experts assumed that an economic reusable rocket would be impossible. SpaceX then flipped the script. According to ARK’s research, the first stage of the Falcon 9 cost <$1 million to refurbish. Now, rocket turnaround time should be proportional to the cost required to refurbish a rocket booster, the key metric in tracking launch cost declines. 50 100 150 200 250 2016 2018 2020 2022 2024 Days Between Reuse Falcon 9 Average Reuse Days Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. 252 54 356 27 21 25 14 50 100 150 200 250 300 350 400 Space Shuttle: Average 1972- 2011 Space Shuttle: Fastest 1985 First SpaceX Turnaround Time SpaceX 2021 Fastest Turnaround Time SpaceX 2022 Fastest Turnaround Time SpaceX 2023 Fastest Turnaround Time SpaceX 2024 Fastest Turnaround Time Days Between Reuse Rocket Turnaround Time 116

• 117. Rocket Reuse And Size Are Driving Rocket Launch Costs Down On A $/kg Basis ARK’s Big Ideas 2025: Reusable Rockets Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. $1 $10 $100 $1,000 $10,000 $100,000 0 1 10 100 1,000 10,000 100,000 1,000,000 $/kg To Low Earth Orbit (2023 $) Cumulative Upmass (Metric Tons) SpaceX Launch Cost Decline Historical 2023 ARK Calculation Forecast 0.1 117

• 118. SpaceX Continues To Drive Satellite Bandwidth Costs Down In Line With Wright’s Law Anik-F2 (2004) WildBlue-1 (2006) ViaSat-1 (2011) ViaSat-2 (2017) ViaSat-3 (2020) Starlink (2021) (estimated) Starlink (2024) (estimated) Starlink Forecast Starship 2025/2026 Starlink Forecast Rapidly Reusable Starship 1 10 100 1,000 10,000 100,000 1,000,000 10,000,000 100,000,000 1,000,000,000 1 10 100 1,000 10,000 100,000 1,000,000 10,000,000 100,000,000 $/Gbps In Orbit (2024 $) Cumulative Gbps Satellite Bandwidth Cost Decline According to Wright’s Law, satellite bandwidth costs should decline roughly ~45% for every cumulative doubling in gigabits per second (Gbps) in orbit. Since 2004, the cost of satellite bandwidth has dropped 15,000-fold from $300,000,000 to ~$20,000/Gbps. According to ARK’s research, 1 Gbps can serve 200 customers. At a capital cost of ~$1,000/Gbps, SpaceX could recoup its Starship and satellite investment with a one-time charge of $5 per customer. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 118 ARK’s Big Ideas 2025: Reusable Rockets

• 119. Lower Satellite Launch Costs Should Enable Continuous Global Coverage With Low Latency ARK’s Big Ideas 2025: Reusable Rockets While latency precluded geostationary orbit (GEO) satellites from offering a compelling broadband internet solution, now thousands of low-cost, low earth orbit (LEO) satellites can provide service with low latency, continuous global coverage, and direct-to-mobile device connectivity. GEO ~22,000 miles 700 ms latency LEO ~300 miles <40 ms latency Debris will fall back to earth within ~5 years Debris will fall back to earth within 1,000+ years Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. 119

• 120. Starship Will Help The Starlink Constellation Achieve Its Potential ARK’s Big Ideas 2025: Reusable Rockets 500 1,000 1,500 2,000 2,500 2023 2024e Metric Tons Spacecraft Upmass SpaceX Other Providers Starship’s payload capacity to LEO is ~5x that of the Falcon 9. While impressive, given the five-year life of its satellites, Starship still will have to fly every 2.3 days to maintain its target constellation of 42,000 Starlink satellites. As of January 2025, SpaceX has a constellation of over 7,000 satellites. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 18,000 3.8 (2023 Cadence) 2.7 (2024 Cadence) 2.1 (2025 Goal) 1.0 (Modeled) 0.5 (Modeled) Metric Tons Falcon 9 Annual Upmass Capability By Launch Cadence Upmass necessary to maintain a 42,000 satellite constellation 2.3 (Modeled) Starship Annual Upmass Capability By Launch Cadence Days Between Launches 120

• 121. Despite A Proliferation In Company Creation, Small Launch Providers May Not Be The Winners ARK’s Big Ideas 2025: Reusable Rockets After capital spending booms, industries tend to consolidate. In the space industry, while launch capability is critical, the larger opportunity could be in the services enabled by low launch costs. Today, only 17 of the 194 small launch providers created since 1996 are operational. Source: ARK Investment Management LLC, 2025, based on data from NewSpace Index as of January 10, 2025. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 5 10 15 20 25 30 19961997 1998 1999 2001 2002 2003 2004 20052006 2008 200920102011 20122013 20142015 20162017 2018 2019 20202021 20222023 2024 Number Of New Small Launch Providers Founded 10 27 10 47 56 27 17 50 100 150 200 Stage Of Small Launch Providers Created Since 1996 Operational Development <5 years Development >5 years Dormant Concept Cancelled Retired 121

• 122. SpaceX Has Converted The Addressable Market Into Revenue With Increased Starlink Capacity ARK’s Big Ideas 2025: Reusable Rockets Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including SpaceX 2024, as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ~ 20 24 96 1,000 200 400 600 800 1,000 1,200 V1 V1.5 V2 Mini V3 Gbps SpaceX Bandwidth Per Satellite 100 200 300 400 1 2 3 4 5 2021 2022 2023 2024 2025 Terabits Per Second (Tbps) Millions Starlink Subscribers And Bandwidth Capacity Starlink Subscribers (lhs) Cumulative Bandwidth 122

• 123. Thanks to declining rocket launch costs, telecom operators could provide 100% of cellular subscribers satellite connectivity anywhere in the world. Satellite Connectivity Revenues Could Exceed $130 Billion Per Year 0.001% 0.010% 0.100% 1.000% 10.000% 100.000% 1997 2007 2017 2027 Log Scale Growth Of Satellite Subscribers Relative To Cellular Subscriber RVs 11 Million $1,620 ~$18 Billion 25 Thousand $225,000 ~$6 Billion Commercial Aircraft Fleet 100 Thousand $60,000 ~$6 Billion Cruise Ships, Warships, Commercial Ships Recreational Boats 8.5 Million $1,620 ~$14 Billion Global Households Without Access To Broadband 600 Million $60 ~$40 Billion Addressable Subscribers* Annual Revenue* Direct-To-Device 8 Billion $6 ~$48 Billion Annual Addressable Market* Total: ~$132 Billion *Forecasts. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. SpaceX Direct To Device With T-Mobile, At Scale 123 ARK’s Big Ideas 2025: Reusable Rockets

• 124. By 2030, Hypersonic Flight Could Be A ~$35 Billion Market, Ready To Scale To ~$350 Billion Longer Term ARK’s Big Ideas 2025: Reusable Rockets According to the US Department of Transportation, leisure travelers are willing to spend 60%-90% of their estimated hourly household income to save one hour.* Compared to conventional flights that can take 28 hours roundtrip, ARK estimates that hypersonic flights could take just 6 hours, saving each traveler ~22 hours. Given the typical cost and potential time savings, ARK’s research suggests that a firstclass passenger should be willing to spend $44,000 roundtrip for a hypersonic flight. If launch costs decline in line with ARK’s expectations, early adopters of hypersonic flight could generate $35 billion revenue by 2030. 110kg $200/kg to LEO $44,000 Starship Price 2 Roundtrip Total Number Of Airline Passengers Worldwide: 6.7 Billion 5% Of Flights Are Long-Haul 5% Of Passengers Are First-Class Number Of Passengers On Long-Haul Flights: ~335 Million Number Of Passengers Flying First-Class: ~16 Million 50% Adoption At Maturity Number Of Passengers Flying Hypersonic: ~8 Million $44,000 Roundtrip Ticket Annual Addressable Market: ~$350 Billion Building Blocks Of Addressable Market Forecast *Forecasts. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of January 10, 2025, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 124

• 125. Multiomics Operationalizing Data With AI To Transform Diagnostics, Drug Discovery, And Therapies Nemo Despot Marjanovic, PhD Rong Guo, PhD Brett Winton ANALYST, MULTIOMICS RESEARCH ASSOCIATE, MULTIOMICS CHIEF FUTURIST 125 ARK’s Big Ideas 2025

• 126. Multiomics Performance Should Improve By Orders Of Magnitude By 2030* 10x increase in cancer monitoring market 20x increase in screening productivity 1.6x faster commercialization 4x decrease in cost to develop 5x higher return on R&D 1,000x more performant at same cost 20x more valuable than standard of care drugs 2.4x more valuable than best-in-class precision drugs Generate Data Run Tests Infer And Inform Decisions Run Experiments Infers And Informs Development Address Small Disease Populations Economically Identify Patients And Diseases More Precisely • Intellia Therapeutics • CRISPR Therapeutics • Beam Therapeutics • Prime Medicine • Natera • Guardant Health • Recursion Pharmaceuticals • Absci • Isomorphic Labs Biological Insights Generates Molecular Diagnostics • Adaptive Biotechnologies • Veracyte AI Developed Drugs Cures AI Multiomic Tools 100x decrease in cost to read DNA et al. 1,000x decrease in cost to write DNA et al. • Twist Biosciences • Tempus AI • 10X Genomics • Illumina 126 *The multiomics performance statistics provided on this slide represent ARK’s research-based forecasts for 2030, which may not be realized. The companies listed represent companies that are currently working toward achieving the forecasted results, but the list does not include all companies that may be pursuing the same goals, and which may do so more successfully. The companies listed may or may not be held in ARK portfolios. The information provided should not be used as the basis for any investment decision, and it should not be assumed that an investment in any of the companies listed was or will be profitable. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 127. Biological Information Is Encoded In Three Layers Sequence Layer The static genetic blueprint within DNA, RNA, and proteins. STATIC Illumina Oxford Nanopore PacBio QuantumSi Bruker DeepMind Isomorphic Labs ThermoFisher 10X Genomics Akoya ScaleBio Vizgen Biological Information Measuring Tools* DYNAMIC The linear arrangement of nucleotides in DNA or amino acids in proteins Examples: • DNA sequences store genetic information, determining an organism's primary traits. • Protein sequences define structure and function in cellular processes. Transcription Translation DNA mRNA Protein 3' Tyr Ser Gly Ser 5' 5' 3' 3' Codons 5' Amino acids G A C T A G T C T G Sequence The three-dimensional organization of biomolecules like proteins and nucleic acids Examples • Protein structures determine binding sites and catalytic activity. • Nucleic acid structures affect stability, replication, and transcription. Agglomeration of molecules Examples • Gene regulation: mRNA and protein abundance vary, controlling gene expression. Structure Systems t-SNE2 -20 -10 0 10 20 -30 -30 -20 -10 0 10 20 30 t-SNE1 30 Label Label Label Label Label Label Label Label Single Cell Genomics AlphaFold AI DNA/Protein Sequencing Advances in sequencing, AI, and single-cell technologies enable us to read, model, and predict complexity. Biological information is organized by three interconnected layers that drive life processes. Systems Layer The dynamic abundance of biomolecules responding to cellular and environmental cues. Structure Layer The 3D folding of biomolecules, defining their function. 127 *The companies listed are examples of companies that employ the measuring tools described, but the list does not include all companies that may be utilizing those tools, and which may do so more successfully. The information provided should not be used as the basis for any investment decision, and it should not be assumed that an investment any of the companies listed was or will be profitable. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 128. $1 $1,000 $1,000,000 $1,000 M $1,000 B 1990 1995 2000 2005 2010 2015 2020 2025 2030 Cost Per 1B Base Pairs (Log Scale) DNA Sequencing Cost Per Whole Human Genome Actual Projected Moore's Law Projection $0 $1 $10 $100 $1,000 $10,000 $100,000 $1 M $10 M $100 M $1 B $10 B 1990 2000 2010 2020 2030 DNA Synthesis Cost Per 1B Base Pairs Actual Projected Moore's Law Projection The Costs Of DNA Sequencing And Synthesis Are Falling Precipitously Illumina’s introduction of high-density flow cell DNA sequencing and synthesis costs are declining faster than Moore's Law* predicted. The cost of reading biology through multiomics tools like DNA sequencing has plummeted by 10 billion-fold (10¹⁰) in its 35-year history. DNA synthesis costs have dropped 100,000-fold (10⁵) and could continue to fall 10 million-fold (10⁷) by 2030. Twist Bioscience’s silicon-based DNA synthesis, combined with miniaturization and parallelization, has reduced costs by increasing efficiency, reducing reagent usage, and scaling production. Pressure From Competitors Illumina's acquisition of Solexa Twist chip miniaturization for DNA storage $1 M $1 K $100 K $10 K $1 K 128 *Moore’s Law predicts that the number of transistors on a chip doubles every 18 months to two years. See Winton 2019. Note: The total price of sequencing the first human genome was estimated at ~2.7B, which included all the costs of developing and improving sequencing tools, actual sequencing, and all analysis. In the above graph, we are calculating only actual sequencing cost per base pair. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including BusinessWire 2024 and Carlson 2022 as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 129. The Productivity Of Genomics Analysis Has Skyrocketed Since 2001, the compute time required to analyze a human genome has plummeted from 180 days to 10 minutes. Today we can analyze ~1.4 billion genomes for the cost of analyzing a single genome in 2001. Alongside the plummeting costs of sequencing and synthesis, as noted on the previous slide, genome analysis is commoditizing, driving a revolution in biological research and applications, from personalized medicine to global-scale genomics. ~180 Days ~70 Days ~15 Days ~15 Hours ~60 Min ~34 Min <10 Min 1 10 100 1,000 10,000 100,000 1,000,000 2001 2006 2010 2012 2015 2020 2024 Minutes Compute Time For Analysis Of One Human Genome 1 ~1.4 Billion 1 10 100 1,000 10,000 100,000 1,000,000 10,000,000 100,000,000 1,000,000,000 10,000,000,000 2001 2024 Number Genomes Analyzed For $450,000 Budget 129 Note: 2001 marks the publication of the first draft of the human genome, covering approximately 90% of the genome. In 2003, a near-complete sequence (99.99%) was published, while the fully complete genome was finalized in 2022. Although sequencing the first human genome took about 13 years in total, this slide focuses solely on the time required for the computational analysis, excluding the experimental phase, which was a significantly larger bottleneck. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Saunders et al. 2021, Miller et al. 2015, Gorzynski et al. 2022, and Clifford 2024 as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 130. 15 408 2,880 112,000 800,000 1 10 100 1,000 10,000 100,000 1,000,000 Used to Train LLaMa 405b Available in UK biobank Largest publicly available whole genome datasource Generated Annually if we sequenced every newborn in the US Generated annually if we sequenced every newborn globally Devote 1% of global healthcare budget to sequence a billion people* Trillion Tokens (Log) Biological Data Is Big Data LLaMa 405b LLaMa 405b UK Biobank US Newborns Sequenced Per Year Forecast Global Newborns Sequenced Per Year Forecast A Billion People Sequenced Forecast Multiomics tools create volumes of data. The largest publicly-available genomics database, the UK Biobank, includes a half-million patients with 27x more data than the 15 trillion tokens of text powering large AI language models. If newborn sequencing were to become the standard globally, as we expect, data volumes would explode by 1,000x or more. Data Volumes Should Explode As Measurement Costs Decline Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources , including UK Biobank as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 130 ARK’s Big Ideas 2025: Multiomics

• 131. 0.001 0.01 0.1 1 10 100 1000 $ (Log Scale) Number Of Cells Profiled (Approximations) Cost Per Single Cell 1 2 100 1K 3K 23K 123K 33M 277M 1B 2.5B 4.2B 7B 13B 20B 27B 2.6T 9T 11T 10^5 10^6 10^7 10^8 10^9 2024 2028 $ (Log Scale) Cellular Foundation Model Training $240 Million $2 Million Share of Cost: Compute Sequencing Single Cell Virtual Cell Foundation Models Are Collapsing The Cost Of Single-Cell Genomics Single-cell genomics costs are dropping exponentially, in line with Wright’s Law.* According to ARK’s research, trillion-cell experiments are feasible now that single-cell sequencing costs have dropped to $0.01 per cell. Virtual cell foundation models are likely to become the killer AI application in drug discovery and precision medicine. By 2028, virtual cell costs could drop ~120x. Companies like 10x Genomics and Scale Bio are leading the way. 2024 ~$0.01 2024 ~$0.05 2019 ~$0.2 2014 ~$9 2009 ~$400 120x cost reduction 42% 42% 16% 58% 20% 22% 131 Note: The companies mentioned are examples of companies developing virtual cell models, but the list does not include all companies that may be developing these models, and which may do so more successfully. The information provided should not be used as the basis for any investment decision, and it should not be assumed that an investment any of the companies mentioned was or will be profitable. *Wright’s Law states that for every cumulative doubling of units produced, costs will fall by a constant percentage. See Winton 2019. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Tang et al. 2009, Jovic et al. 2022, Ding et al. 2020, and Rood et al. 2024 as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 132. AlphaFold Is Redefining The Prediction Of Molecular Structures AlphaFold has transformed the prediction of protein structures, surpassing traditional methods like X-ray crystallography and Cryo-EM in both productivity* and scope.** Each version of AlphaFold has delivered exponential gains, not only improving efficiency, but also unlocking entirely new structural predictions. AlphaFold1, for example, predicted single-chain proteins (∼60M structures), but AlphaFold3 can predict more than 1 billion structures, including multi-chain complexes, protein-ligand, protein-DNA/RNA interactions, and post-translational modifications. 1.1 289 508 0 100 200 300 400 500 600 Productivity Gains: Alphafold VS. Traditional Methods 60 140 1000 0 200 400 600 800 1000 1200 Millions Structures Predicted AlphaFold 2018 AlphaFold2 2020 AlphaFold3 2024 AlphaFold 2018 AlphaFold2 2020 AlphaFold3 2024 Unit Of Productivity Gains 132 *Productivity gains are defined as the ratio of AlphaFold's Efficiency Metric—accuracy per cost per speed—to the average efficiency of traditional methods. **Scope is calculated as the Efficiency Metric × Number of Structures Predicted. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Jumper et al. 2021, CASP13 2019, and Liu et al. 2024, as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 133. Minimal residual disease (MRD) testing can detect the recurrence of cancer up to 20 months earlier than traditional imaging. Yet, more than eight million suitable cancer patients in the US currently are not reimbursed for MRD testing. Now, at only 10% penetration, MRD testing should become the standard of care for every cancer patient, with multiple tests per year and follow-ups for five years. ARK’s research suggests that MRD testing could produce 700x more data than the largest genomics project to date, the UK Biobank, enabling advanced predictive and prescriptive analytics that could revolutionize how we understand, monitor, and manage cancer. MRD Testing Is Catalyzing The Generation Of Genomics Data 0.3 0.7 8.1 Number (Millions) 0 1 2 3 4 5 6 7 8 9 10 MRD Patient Population 30 4320 21600 0 5000 10000 15000 20000 25000 PetaBytes MRD Testing Data MRD Tested Patients 2024 Current Patient TAM Future Patient TAM UK Biobank MRD TAM Annual MRD TAM Five-Year Forecast 5-fold 144-fold 133 Note: “TAM”: Total addressable market. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Wan et al. 2017, Luskin et al. 2017, and Borfitz 2024 as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 134. 100 240 $0 $50 $100 $150 $200 $250 $300 Traditional Cancer Screening Universal MultiCancer Screening Market Potential For Screening Methods ($ Billions) 0.82 3.15 0 0.5 1 1.5 2 2.5 3 3.5 Lives Saved (Millions) $1,500 $300 $0 $200 $400 $600 $800 $1,000 $1,200 $1,400 $1,600 Cost Efficiency Of Screening Methods ($ Billions) Multi-Cancer Screening Could Save Lives And Cut Costs Powered by liquid biopsy technology, multi-cancer screening is a revolutionary approach to early cancer detection. Unlike traditional methods, a single blood test can screen for multiple cancers simultaneously, streamlining screening and expanding its scope. The FDA’s recent approval of Guardant Health’s “Shield” colorectal cancer screening test suggests that the technology is proving its potential. Pending resolution of regulatory and reimbursement paths, universal multi-cancer screening could more than double the Total Addressable Market (TAM) in the US. ~2x increase ~5x Reduction ~4x Increase Traditional Cancer Screening Universal MultiCancer Screening Traditional Cancer Screening Universal MultiCancer Screening 134 Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Philipson et al. 2023, Goddard et al. 2024, and Mannucci and Goel 2024 as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 135. Target Hypothesis TargetTo-Hit Hit-ToLead Lead Optimization Preclinical Phase I Phase II Phase III Submission To Launch ~1 year ~1.5 years ~1.5 years ~1 year ~1.5 years ~2.5 years ~2.5 years DISCOVERY DEVELOPMENT ~1.5 years The “virtual cell” is a cutting-edge achievement that integrates artificial intelligence and single-cell genomics. By leveraging multimodal, multi-scale data, virtual cells use foundation models to simulate cellular functions and predict responses to perturbations across various biological states. A transformative leap in drug discovery—particularly target identification and validation—virtual cells enable in-silico experiments to test hypotheses, simulate perturbations, and identify viable targets efficiently and cost-effectively. They also predict how specific mutations or drugs influence cellular pathways, potentially accelerating the pipeline from target discovery to validation and reducing reliance on costly and time-consuming wet-lab experiments. We expect the same timeline reductions as AlphaFold achieved in the prediction of molecular structure. Companies like 10X Genomics, Illumina’s FluentBio, ScaleBio, and ParseBio are providing technologies that build Virtual Cells. Single-Cell Genomics And AI Are Creating Virtual Cells And Transforming Drug Discovery Patient Cell Single Cell Genomics +AI Virtual Cell Academic Studies That Can Take Decades 135 Note: The companies mentioned are examples of companies developing virtual cell models, but those examples do not include all companies that may be developing these models, and which may do so more successfully. The information provided should not be used as the basis for any investment decision, and it should not be assumed that an investment any of the companies mentioned was or will be profitable. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Bunne et al. 2024 and Rood et al. 2024 as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics Lead Candidate IND Hit

• 136. A self-driving lab (SDL) combines not only multiomics tools that read biology at scale, but also automation that enables high-throughput experimentation and large language models (LLMs) that analyze data and design the next set of experiments. The SDL’s integrated approach should enable the transition from automated to autonomous drug discovery and research, potentially accelerating the cycle of discovery and reducing costs/time while increasing efficiency. According to our research, SDLs could save up to two years and hundreds of millions of dollars in the discovery process. Currently achieving ~200x productivity gains, Recursion Pharmaceuticals is pioneering this approach. Self-Driving Labs Are Using Scalable Biology And AI To Revolutionize Drug Discovery Self-Driving Lab Target Hypothesis Lead Optimization Preclinical Phase I Phase II Phase III Submission To Launch ~1 year ~1.5 years ~1.5 years ~1 year ~1.5 years ~2.5 years ~2.5 years DISCOVERY DEVELOPMENT ~1.5 years Lead Candidate IND Hit 136 Academic Studies That Can Take Decades TargetTo-HitHit-ToLead Note: Recursion Pharmaceuticals is the company currently pioneering the use of SDLs, but there may be other companies that utilize SDLs and which may ultimately do so more successfully. The information provided should not be used as the basis for any investment decision, and it should not be assumed that an investment in the company mentioned was or will be profitable. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Recursion 2024, Abolhasani and Kumacheva 2023, and Martin et al. 2023 as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 137. Often called the "valley of death" in drug discovery, preclinical studies are prone to high failure rates that stem from predictive models like 2D cell systems and animal testing. Thanks to advances in microfluidics, 3D bioprinting, and organoids, next-generation preclinical models like organs-on-chips are evolving. The new systems are more physiological, scalable, and capable of supporting higher-throughput testing. Integrating Gen 2 models—grown from patient-derived biopsies— with AI-powered analyses of vast datasets could lead to precise, patient-specific predictions that improve drug discovery outcomes dramatically. Large Language Models are beginning to impact and accelerate clinical trials by optimizing regulatory filings, patient selection, trial designs, and data analysis. Organs-On-Chips And AI Are Rescuing Drug Discovery From The Valley Of Death Target Hypothesis Lead Optimization Preclinical Phase I Phase II Phase III Submission To Launch ~1 year ~1.5 years ~1.5 years ~1 year ~1.5 years ~2.5 years ~2.5 years DISCOVERY ~1.5 years DEVELOPMENT Lead Candidate IND Hit Percentage responding(%) 25 50 75 100 ED5 0 TD5 0 Therapeutic effect Toxic effect Therapeutic index Gen 1 2D Human Systems Animal Models Organs On Chip 3D Human Systems Human Biopsy Gen 2 Gen 3 3D Printing/ Microfluidics AI Valley of Drug Discovery Death Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources, including Zhou et al. 2023, Tong et al. 2024, and Thng et al. 2024 as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 137 Academic Studies That Can Take Decades TargetTo-HitHit-ToLead ARK’s Big Ideas 2025: Multiomics

• 138. Biotech valuations include little-to-nothing for the preclinical and Phase 1 stages of development, but increasingly they are likely to incorporate the higher probability of longer patent lives. AI-driven drug development could reduce time-to-market nearly 40%, from 13 years to 8 years, while reducing total drug costs 4-fold, from $2.4 billion to $0.6 billion. 0 100 200 300 400 500 600 13 12 11 10 9 8 7 6 5 4 3 2 1 11 10 9 8 7 6 5 4 3 2 1 8 7 6 5 4 3 2 1 $ Millions Per Phase Years To Commercialization Drug Development Average CostHit-to-Lead Lead Optimization Pre-clinical Safety Phase I Efficacy Phase II Phase III Confirmation Target-to-Lead AI Is Transforming The Economics Of Drug Development Time Cost Of Money Failed Candidates Successful Candidates Licensing INDUSTRY AVERAGE 2024 • $2.4 billion total • $1.1 billion pre-clinical • 7.5 human trial failures • 13 years time-to-market INITIAL AI-DRIVEN DRUG EFFORTS • $1.5 billion total • $0.6 billion pre-clinical • 4.3 human trial failures • 11 years time-to-market FUTURE AI-DRIVEN DRUG DESIGN • $0.6 billion total • $0.13 billion pre-clinical • 1.4 human trial failures • 8 years time-to-market ARK’s Big Ideas 2025: Multiomics 138 Note: 10% discount rate, 2024 dollars. For initial AI-driven drug efforts, assumes 70+% failure reduction in phase I that moderates to 20% in phase II and phase III. For future AI-driven drug design, assumes demonstrated 70+% reduction in phase I failure rates that moderates to a 50% failure reduction in phase II and 25% failure reduction in phase III. Assumes that pre-clinical efficiency is similar to what Absci has indicated is achievable. Also assumes that licensing timeframe reduces to 12 months from 18 months for future drugs. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon.

• 139. Reducing Development Timelines Should Increase The Value Of Patents And Commercialization Industry Average Current AI methods accelerate time-to-market by 2 to 3 years, increasing the value of intellectual property by 30% to 50% Emerging AI methods could accelerate time-to-market by 4 to 5 years, increasing the value of intellectual property by 70% to 80% 0.5x 1.0x 1.5x 2.0x 3 4 5 6 7 8 9 10 Patent Life Years Remaining At Commercialization Value Of A Drug At Commercialization By Patent Life Remaining Relative To Industry Average Faster commercialization extends patent-protected revenue, allowing companies to generate higher returns for longer. Current AI efforts shorten development timelines by 2-3 years, increasing a drug's lifetime value by 30-50% compared to the industry average of five years. Emerging AI designs could accelerate time-tomarket by 4-5 years and boost the value by 70-80%. Note: 10% discount rate. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 139 ARK’s Big Ideas 2025: Multiomics

• 140. Industry Average Current AI methods accelerate time to market by 2 to 3 years, increasing IP value by 30% to 50% Emerging AI methods could accelerate time to market by 4 to 5 years, increasing IP value by 70% to 80% 0.5x 1.0x 1.5x 2.0x 3 4 5 6 7 8 9 10 Patent Life Years Remaining At Commercialization Value Of A Drug At Commercialization by Patent Life Remaining Relative To Industry Average -$1,500 -$500 $500 $1,500 $2,500 $3,500 $4,500 Years From Start Of Development Drug Development Cumulative Cashflow ($ Millions) Drug Development Cost Efficiencies Could Boost Cash Flow Significantly AI Designed Drugs Forecast The value of AI-driven drug development should compound thanks to lower costs, an acceleration in time-to-market, and a longer timeline of patent-protected revenue. Over a 30-year period, the cumulative cashflow for the average AI-designed drug could reach $4 billion, compared to less than $1 billion for traditional drug models. Even before a traditional drug breaks even, an AI-developed drug could generate $2.5 billion in cashflow. Traditional Drug 5 10 15 20 25 30 140 Note: 10% discount rate, 2024 dollars. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 141. Cures for diseases can command much higher prices than traditional treatments. The average price of a cure today exceeds $1 million, nearly 15x the the lifetime prescription cost that would be necessary to manage the disease. Capturing revenue upfront and addressing much of the patient population well before patent expiration suggests that cures could become more valuable—potentially 20x more valuable—than a typical drug and 2.4x more valuable than a chronic prescription that abates the disease. Biological Cures Could Become Quite Valuable $8,000 $75,000 $390,000 $620,000 $1,100,000 $0 $200,000 $400,000 $600,000 $800,000 $1,000,000 $1,200,000 Price Of A Cure For Rare Disease (Per Patient, On Average) Typical Prescription Cost Per Year Lifetime Prescription Cost Lifetime Non-Drug Health Service Costs Saved Life Years Gained Value Of Cure $0 $500 $1,000 $1,500 $2,000 $2,500 $3,000 $3,500 Value Of A Cure For Rare Disease ($ Millions, On Average) Cures avoid competition, with lifetime value captured on the first dose Cures frontload cash in a drug’s commercial lifecycle Typical Prescription Medicine Drug That Displaces Health Service Costs Chronic Prescription That Abates The Disease One Time Cure Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 141 ARK’s Big Ideas 2025: Multiomics

• 142. 0% 15% 30% 45% Drug Development Return On Invested Capital Cures And Accelerated Development Should Transform The Economics Of Drug Development At inception Preclinical Phase I Phase II Phase III Commercial (300) (200) 0 200 500 1,000 200 600 1,000 1,400 2,000 3,000 300 500 600 800 1,200 1,800 600 700 900 1,000 1,300 1,800 1,500 1,800 2,300 2,500 3,400 4,400 1,800 2,000 2,500 2,700 3,500 4,400 Value Of Assets at Different Clinical Phases ($ Millions) The combination of AI-accelerated drug development and the value of cures for disease could boost returns on research and development (R&D) and improve the value of pre-clinical pipeline assets materially. According to our research, an AI-accelerated cure in the first phase of human testing could be worth more than $2 billion. Traditional assets typically pay back only the cost of capital for Phase 1 pipelines. Representative Companies* Industry Average Beam, Crispr Tx, Editas, Intellia, Prime Abcellera , Absci, Generate Bio, Isomorphic Lab, Lantern Pharma, Recursion, Relation, Roivant Sciences, Xtalpi In Development Traditional Drug, Traditional Development Cure, Traditional Development Traditional Drug, AI Speedup, Traditional Fail Rate Traditional Drug, AI Speedup And AI Success Rate Cure, AI Speedup, Traditional Fail Rate Cure, AI Speedup And AI Success Rate 142 Note: 10% discount rate. *Companies pursuing this strategy for drug development, though this does not imply that every asset the company has at that particular stage is worth that amount or that company return on R&D dollars will meet the modeled value. The information provided should not be used as the basis for any investment decision, and it should not be assumed that an investment any of the companies mentioned was or will be profitable. Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. ARK’s Big Ideas 2025: Multiomics

• 143. AI-Driven Drug Development And Cures For Disease Could Reverse A Multi-Decade Decay In Pharma/Biotech Returns -10% 0% 10% 20% 30% 40% 50% 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 As AI and other technologies Return On Drug Development R&D impact the drug development pipeline, industry-wide returns should turn around. Meanwhile, many development projects in place today could become sunk costs as more nimble AI drug development platforms beat them to market with cures that eliminate disease. A bifurcation could grow between the returns generated by AI driven, cure-seeking firms and those traditional pharma/biotech companies that do not adapt. 14% 22% 30% 47% Cure, Traditional Development Traditional Drug, AI Speedup, Traditional Fail Rate Traditional Drug, AI Speedup and AI Success Rate Cure, AI Speedup, Traditional Fail Rate Cure AI Speedup And AI Success Rate Source: ARK Investment Management LLC, 2025. This ARK analysis draws on a range of external data sources as of December 31, 2024, which may be provided upon request. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security. Past performance is not indicative of future results. Forecasts are inherently limited and cannot be relied upon. 143 ARK’s Big Ideas 2025: Multiomics

• 144. Works Cited 144 ARK’s Big Ideas 2025

• 145. Abolhasani, M. and E. Kumacheva. 2023. “The Rise of Self-driving Labs in Chemical and Materials Sciences.” Nature Synthesis. American Automobile Association. 2050. “Your Driving Costs.” American Automobile Association. 2024. “How Much Does it Really Cost to Own a New Car?” Artemis Terminal. 2024. Bitcoin Treasuries. 2024. Borfitz, D. 2024. “UK Biobank Pivoting to Platform-Only Model For Big Data Sharing.” Bio IT World. Bunne, C. et al. 2024. “How to Build the Virtual Cell with Artificial Intelligence: Priorities and opportunities.” Cell. BusinessWire. 2024. “Ultima Announces UG 100™ and Reveals Disruptive Cost and Accuracy Profile to Enable the Era of the $100 Genome and Beyond.” Calma, J. 2024. “Google Orders Small Nuclear Reactors for its Data Centers.” Financial Times. Carlson, R. 2022. “DNA Cost and Productivity Data, aka ‘Carlson Curves.’” Synthesis. Capital IQ Pro. 2024. CASP13. 2019. “13th Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction.” Circle. 2024. Cleveland, C. 2023. “United States Electricity History in Four Charts.” Boston University Institute for Global sustainability. Clifford, J. 2024. “Boost Multi-Omics Analysis with GPU-Acceleration and Generative AI.” Nvidia Developer. CoinGecko. 2024. Dediu, H. 2017. “Share of United States Households Using Specific Technologies.” Our World in Data. DeLong, B. 1998. "Estimating World GDP, One Million B.C. — Present." Brad De Long’s homepage. Department of Economics, University of California, Berkeley. de Selding, P.B. “Orbital ATK Believes in Satellite Servicing, but Not Rocket Reusability.” Space News. Ding, J. et al. 2020. “Systematic Comparison of Single-cell and Single-nucleus RNA-sequencing methods.” Nature Biotechnology. Dune Analytics. 2024. Electric Capital. 2024. “Developer Report 2023.” EV Volumes. 2025. Foust, J. 2014. “Reusability and Other Issues Facing the Launch Industry.” The Space Review. Franchisee Resource Center. 2021. “Fast Casual Industry Analysis 2020 - Cost & Trends.” Franchise Help. Glassnode. 2024. Goddard, K. et al. 2024. “Estimation of Cancer Deaths Averted From Prevention, Screening, and Treatment Efforts, 1975-2020.” JAMA Oncology. Goddard, T. 2024. “AlphaFold Performance: Molecule Size, Speed, Memory, and GPU.” UCSF Resource for Biocomputing, Visualization, and Informatics. Gorzynski, J. et al. 2022. “Ultrarapid Nanopore Genome Sequencing in a Critical Care Setting.” The New England Journal of Medicine. Gumber, A. et al. 2024. “A Global Analysis of Renewable Energy Project Commissioning Timelines.” Applied Energy. Hartig, K. 1998. “TUNING IN: Communications Technologies Historically Have Had Broad Appeal for Consumers,” The Wall Street Journal. Jiminez, C. et al. 2024. “SWE-bench: Can Language Models Resolve Real-World GitHub Issues?” arXiv. Jovic, D. et al. 2022. “Single-cell RNA Sequencing Technologies and Applications: A Brief Overview.” Clinical and Translational Medicine. Jumper, J. et al. 2021. “Highly Accurate Protein Structure Prediction with AlphaFold.” Nature. Jurvetson, Steve 2024. “The Moore’s Law Update.” X. Klotz, I. 2020. “Economics Of Rocket Reuse Still Up In The Air.” Aviation Week. Kurzweil, R. 2006. “The Singularity Is Near: When Humans Transcend Biology.” Duckworth. Lawrence Berkeley National Laboratory. 2024. “Energy Markets & Policy.” Liu, L. et al. 2024. “Technical Report of HelixFold3 for Biomolecular Structure Prediction.” arXiv. Lovering, J. et al. 2016. “Historical Construction Costs of Global Nuclear Power Reactors.” Energy Policy. Luskin, M. et al. 2018. “Targeting Minimal Residual Disease: A Path to Cure?” Nature Reviews Cancer. Mannucci, A. and A. Goel. 2024. “Stool and Blood Biomarkers for Colorectal Cancer Management: An Update on Screening and Disease monitoring.” Molecular Cancer. ARK’s Big Ideas 2025: Works Cited

• 146. Martin, H.G. et al. 2023. “Perspectives for Self-driving Labs in Synthetic Biology.” Current Opinion in Biotechnology. McCormick, M. and J. Smith. 2025. “Microsoft in Deal for Three Mile Island Nuclear Power to Meet AI Demand.” Financial Times. Miller, N.A. et al. 2015. “A 26-hour System of Highly Sensitive Whole Genome Sequencing for Emergency Management of Genetic Diseases.” Genome Medicine. Model T Ford Forum. 2010. “How Many Miles Can a T Drive Down, Before it Needs a Rebuild?” Morningstar Direct. 2024. National Renewable Energy Laboratory (NREL). 2024. “Solar Resource Data.” NewSpace Index. 2025. “Small Satellite Launchers.” OpenAI. 2024. OpenPhilanthropy. 2025. Philipson, T. et al. 2023. “The Aggregate Value of Cancer Screenings in the United States: Full Potential Value and Value Considering Adherence.” BMC Health Services Research. Pitchbook. 2024. Recursion. 2024. “Recursion Unveils LOWE Drug Discovery Software at the J.P. Morgan Healthcare Conference.” Rood, J. et al. 2024. “Toward a Foundation Model of Causal Cell and Tissue Biology with a Perturbation Cell and Tissue Atlas.” Cell. Rwa.xyz. 2024 Saunders, C.J. et al. 2021. “Rapid Whole-genome Sequencing for Genetic Disease Diagnosis in Neonatal Intensive Care Units.” Science and Translational Medicine. Shaikhnag, A. 2024. “Westinghouse Facilitates Improved Nuclear Power with Its 3D Printed Nozzles.” 3D printing Industry. Sidoti, O. et al. 2024. “Mobile Fact Sheet.” Pew Research Center. Small Modular Reactor (SMR) Global Tracker. 2024. Smil, V. 2000. “Energy in the Twentieth Century: Resources, Conversions, Costs, Uses, and Consequences.” Annual Review of Energy Environment. Smyth, J. 2024 “Amazon Buys Stake in Nuclear Energy Developer in Push to Power Data Centres.” Financial Times. SpaceX. 2024. “STARSHIPS ARE MEANT TO FLY—2024 PROGRESS REPORT.” Tang, F. et al. 2009. “mRNA-Seq Whole-transcriptome Analysis of a Single Cell.” Nature Methods. Tether. 2024. The Block. 2024. Thng, D. et al. 2024. “A Functional Personalised Oncology Approach Against Metastatic Colorectal Cancer in Matched Patient Derived Organoids.” NPJ Precision Oncology. TicData Treasury. 2024. “Table 5: Major Foreign Holders of Treasury Securities.” Token Terminal. 2024. Tong, L. et al. 2024. “Patient-derived Organoids in Precision Cancer Medicine.” MED. TradingView. 2024. Tyrer-Jones, A. 2025. “3D Printer Nuclear Submarine Components from AML3D Delivered to US Navy.” 3D PRINTING INDUSTRY. Ulvog, J. 2012. “Travel Cost by Stagecoach in 1870s – Part One.” Outrun Change. UK Biobank. 2024. U.S. Department of Agriculture. 2025. “Food Expenditure Series.” U.S. Department of Agriculture. 2025. “Farm Labor.” U.S. CENSUS BUREAU. 2023. “2021 SUSB Annual Data Tables by Establishment Industry.” Visa OnChain Analytics. 2024. Wallstreet Zen. 2024. Wan, J. et al. 2017. “Liquid Biopsies Come of Age: Towards Implementation of Circulating Tumour DNA.” Nature Reviews Cancer. Winton, B. 2019. “Moore’s Law Isn’t Dead: It’s Wrong—Long Live Wright’s Law.” ARK Investment Management LLC. World Nuclear Association. 2024. “Small Modular Reactor Global Tracker.” Worldpay Inc. 2018. “2018 Global Payments Report.” Worldpay Inc. 2019. “2019 Global Payments Report.” Worldpay Inc. 2020. “2020 Global Payments Report.” Worldpay Inc. 2021. “2021 Global Payments Report.” Worldpay Inc. 2022. “2022 Global Payments Report.” ARK’s Big Ideas 2025: Works Cited

• 147. Worldpay Inc. 2023. “2023 Global Payments Report.” World Gold Council. 2024. “Gold ETFs, holdings and flows.” World Nuclear Association. 2024. “United State of America Vallecitos Permanent Shutdown.” Zhou, Z. et al. 2023. “Harnessing 3D In Vitro Systems to Model Immune Responses to Solid Tumours: A Step Towards Improving and Creating Personalized Immunotherapies.” Nature Reviews Immunology. ZipRecruiter. 2024. ARK’s Big Ideas 2025: Works Cited

• 148. ARK’s Big Ideas 2025: Disclosure For more research on disruptive innovation visit www.ark-invest.com ©2021-2026, ARK Investment Management LLC. No part of this material may be reproduced in any form, or referred to in any other publication, without the express written permission of ARK Investment Management LLC (“ARK”). Please note, companies that ARK believes are capitalizing on disruptive innovation and developing technologies to displace older technologies or create new markets may not in fact do so and/or may face political or legal attacks from competitors, industry groups, or local and national governments. ARK aims to educate investors and to size the potential opportunity of Disruptive Innovation, noting that risks and uncertainties may impact our projections and research models. Investors should use the content presented for informational purposes only, and be aware of market risk, disruptive innovation risk, regulatory risk, and risks related to Deep Learning, Digital Wallets, Battery Technology, Autonomous Technologies, Drones, DNA Sequencing, CRISPR, Robotics, 3D Printing, Bitcoin, Blockchain Technology, etc. Cryptocurrency Risk. Cryptocurrencies (also referred to as “virtual currencies” and “digital currencies”) are digital assets designed to act as a medium of exchange. Cryptocurrency is an emerging asset class. There are thousands of cryptocurrencies, the most well-known of which is bitcoin. Cryptocurrency generally operates without central authority (such as a bank) and is not backed by any government. Cryptocurrency is not legal tender. Federal, state and/or foreign governments may restrict the use and exchange of cryptocurrency, and regulation in the U.S. is still developing. The market price of bitcoin and other cryptocurrencies have been subject to extreme fluctuations. Similar to fiat currencies (i.e., a currency that is backed by a central bank or a national, supra-national or quasi-national organization), cryptocurrencies are susceptible to theft, loss and destruction. Cryptocurrency exchanges and other trading venues on which cryptocurrencies trade are relatively new and, in most cases, largely unregulated and may therefore be more exposed to fraud and failure than established, regulated exchanges for securities, derivatives and other currencies. Cryptocurrency exchanges may stop operating or permanently shut down due to fraud, technical glitches, hackers or malware, which may also affect the price of cryptocurrencies. Cryptocurrency Tax Risk. Many significant aspects of the U.S. federal income tax treatment of investments in bitcoin and other cryptocurrencies are uncertain and still evolving. The content of this presentation is for informational purposes only and is subject to change without notice. This presentation does not constitute, either explicitly or implicitly, any provision of services or products by ARK and investors are encouraged to consult counsel and/or other investment professionals as to whether a particular investment management service is suitable for their investment needs. All statements made regarding companies or securities are strictly beliefs and points of view held by ARK and are not endorsements by ARK of any company or security or recommendations by ARK to buy, sell or hold any security. Historical results are not indications of future results. Certain of the statements contained in this presentation may be statements of future expectations and other forward-looking statements that are based on ARK's current views and assumptions and involve known and unknown risks and uncertainties that could cause actual results, performance or events to differ materially from those expressed or implied in such statements. The matters discussed in this presentation may also involve risks and uncertainties described from time to time in ARK's filings with the U.S. Securities and Exchange Commission. ARK assumes no obligation to update any forward-looking information contained in this presentation. Certain information was obtained from sources that ARK believes to be reliable; however, ARK does not guarantee the accuracy or completeness of any information obtained from any third party. ARK and its clients as well as its related persons may (but do not necessarily) have financial interests in securities or issuers that are discussed. ARK Investment Management LLC