Cathie Wood issues a major warning: Five major innovation platforms are quietly resonating, and a wealth reconstruction comparable to the railway era has already begun

Cathie Wood, known as "Wood Sister," has brought her 104-page annual report interpretation, stating that this era is called the "Great Acceleration," with AI as the main engine. Investors who miss out on innovative assets may find themselves stagnant in this decade. Meanwhile, she has provided the latest "betting" list. On March 13, Cathie Wood, the head of ARK Invest, along with her research team, conducted an in-depth interpretation of the 104-page heavyweight annual report "Big Ideas 2026" through an online video, responding to over 800 market questions previously collected. In her opening remarks, Wood defined this report as a return to a research paradigm: “It's like the research done by investment bankers in the 1980s and 1990s when the PC era was just emerging, trying to glimpse the future of technology. The seeds of that revolution were sown back then, and now we are in the midst of a full-blown technological revolution.” The report focuses on five major innovation platforms: artificial intelligence, multi-omics, public blockchain, robotics, and reusable rockets. The core judgment of the ARK team is that these five technologies are simultaneously entering critical inflection points, collectively driving an investment wave that transcends traditional cycles. Notably, ARK's Chief Futurist Brett Winton provided an exciting macro forecast: driven by investments in data centers and the accelerated deployment of AI agents, the global real GDP compound annual growth rate is expected to exceed 7% by the end of this decade, far surpassing the market expectation of 3%. He stated, “Significant technological transformations will inevitably lead to structural leaps—just as 75% of the stock market value in the U.S. was concentrated in railroads at the end of the 1870s, the current five major innovation platforms are at a similar investment cycle inflection point.” Key points from Wood's interview: The five major innovation platforms centered around AI as the core engine, collaboratively accelerating the evolution and integration of multi-omics, robotics, blockchain, and reusable rockets. ARK predicts that by the end of this decade, the global real GDP compound annual growth rate will exceed 7% (market expectation 3%), with over 60% of global market value concentrated in disruptive innovation platforms, analogous to the railroad cycle of the 1870s. AI is driving the third revolution in human-computer interaction from keyboard to natural language, with a proliferation speed twice that of the internet. The reliable task duration of AI agents has jumped from 5 minutes to over 55 minutes, leading to an increase in corporate willingness to pay. The neutral forecast for the enterprise language model software market alone reaches $7 trillion, sufficient to support over $1 trillion in data center investments. In the multi-omics field, the cost of gene sequencing has dropped from $3 billion to $100, and is expected to reach $10 by 2030, with data volumes comparable to large language models. AI can shorten the time to market for new drugs by 40% and reduce costs by four times, with the value of curative therapies reaching 20 times that of traditional drugs, and high-priced therapies have already been recognized by health insurance In terms of autonomous driving, the cost per mile can be reduced to $0.25 after scaling, far lower than human-driven ride-hailing and private cars. The technology is ready, with a market space of $34 trillion, primarily benefiting platform operators that master core technologies, with the US and China leading the way. The development of AI is accelerating the demand for space computing power. In the field of reusable rockets, SpaceX has reduced costs by 95%, and once the Starship is fully reusable, costs are expected to drop below $100/kg, giving rise to trillion-dollar markets such as orbital data centers and global satellite internet. By the end of this decade, global economic growth will exceed 7% ARK's Chief Futurist Brett Winton provided the most "wild" figure of the event: "We believe that by the end of this decade, global economic growth will exceed 7%." Winton explained that this prediction has historical precedent. "Every major technological transformation triggers structural changes in the underlying equilibrium growth rate." He cited the railroad era of the 1870s—when railroads accounted for 75% of the market value of US stocks—arguing that the rise of the current five major innovation platforms will lead to a similar reconstruction of corporate value. The five major platforms include: AI, multiomics, public blockchain, robotics, and reusable rockets, which are highly interconnected, with AI serving as the "central engine." In terms of asset allocation, ARK has made similarly aggressive judgments—“over 60% of the global stock market value will belong to disruptive innovation platforms.” Brett warned that core portfolios not exposed to innovation “could actually shrink over this decade, while the value of these innovative companies will become immense.” AI is not a bubble; every GPU is being utilized In response to external doubts about over-investment in AI infrastructure, the ARK team directly addressed one of the sharpest questions in the market. ARK AI Research Director Frank Downing stated: "In the 90s internet era, we laid a lot of fiber optics, which remained dark for many years, and only now are we truly using it. Today, every GPU is being used and is in a state of shortage. This is a fundamental difference." Wood took over the conversation: "Even older generations of GPUs are being utilized." Frank provided more specific data: GCP (Google Cloud) achieved a 48% annual growth rate on a revenue scale of $70 billion. The time span for AI agents to reliably complete tasks has extended from 5 minutes in early 2025 to now over 55 minutes. Cursor (an AI programming tool) achieved double the revenue that Twilio took six years to reach, with half the people and half the time Frank further puts forward a core prediction: the language model aimed solely at enterprise knowledge workers will create a $7 trillion AI software market. "This is just the tip of the iceberg. Fields like multi-omics and embodied robotics will require trillions of dollars in computational power." AI + Biology Costs Plummet, Multi-Omics Revolution Opens Up a $28 Trillion Market ARK multi-omics analyst Shea describes the integration of AI and biology as "the most profound AI application scenario." The cost of human genome sequencing has dropped from $3 billion and 13 years to $100, and it is expected to approach $10 by 2030. As costs decrease, the volume of testing will double, and the scale of biological data will expand tenfold by 2030. Cathie Wood adds, "The human body has 35 trillion cells, and the data explosion brought by single-cell sequencing will make previous computational eras look insignificant." AI is reshaping drug development: ARK expects it to shorten the time to market by 40% and reduce development costs to a quarter. The return on investment in drug development has fallen from 30% in the 80s and 90s to single digits, but Cathie Wood asserts, "We will return to that golden age." In the face of payment concerns regarding high-priced gene therapies, Shea cites the example of CRISPR's sickle cell disease treatment Casgevy—priced over $2 million, but 90% of American patients are reimbursed. The reason lies in the immense value created for the healthcare system by "one-time cures" compared to lifelong treatment costs. ARK's model shows that AI-driven curative drugs are valued at over $2 billion, 20 times that of traditional drugs. Just 7,000 hereditary angioedema patients in the U.S. could save the system $52 billion. The market potential is even more astonishing: gene editing treatments for high-risk cardiovascular disease patients have a potential market size of $28 trillion. "Capturing just one-twelfth of that is equivalent to the total sales of Lipitor over 20 years." Robotaxi: $34 Trillion Market Accelerating Explosion ARK's Director of Investment Analysis Tasha Keeney defines robotaxi as "the first large-scale physical AI landing scenario that consumers will see." ARK predicts that once Robotaxi platforms are scaled, the cost per mile could be as low as $0.25, less than one-tenth of the cost of hailing a ride in the West, and less than half the cost of driving a car. The cost advantage comes from hardware iteration: the Tesla Model Y has a per-mile incremental cost that is already over 30% lower than Waymo's fifth-generation model; the upcoming Cybercab is expected to expand the cost advantage to 50% compared to Waymo's sixth generation In terms of market size, ARK estimates that by the end of this decade, Robotaxi will generate $34 trillion in enterprise value opportunities, with a potential annual revenue scale exceeding $10 trillion and annual profits of about $2 trillion. Currently, the cumulative mileage of Robotaxi globally has approached 1 million miles. Safety data has vindicated the technology: ARK's early calculations showed that the accident rate for autonomous driving is about 80% lower than that of human drivers, and real operational data has verified this. Chief Futurist Brett Winton issued a stern warning about the lag in European regulation: “Certain regulatory actions are tantamount to active murder.” Winton also revealed a macro-level transformative logic: the implicit labor cost of manual driving in the U.S. reaches $4 trillion annually, accounting for over 13% of U.S. GDP. Transforming this unmeasured activity into tradable economic output is itself a profound economic transformation. Reusable Rockets: SpaceX has reduced launch costs by 95% ARK's in-house analyst Dan McGuire pointed out that SpaceX has reduced rocket launch costs by about 95% since 2008, with over 9,000 active Starlink satellites currently in orbit, accounting for more than two-thirds of all orbital satellites. The key research framework is ARK's proprietary "Wright's Law": for every doubling of cumulative launch mass, launch costs decrease by about 17%. Currently, the launch cost of Falcon 9 is about $1,000 per kilogram, and once Starship becomes fully reusable, ARK expects the cost to drop below $100 per kilogram. “Once this ratio is achieved, AI space computing will be cost-competitive with ground computing,” Brett said, “orbital data centers will be 20 to 25% cheaper than ground computing.” He also mentioned that from Starship to lunar bases, the cost of launching could even reach $10 per kilogram, “but that requires building a lot of infrastructure on the moon, which itself is another brand new infrastructure investment opportunity.” Cathie Wood's Conclusion: AI will not disappear, but will create a whole new world At the end of her speech, Cathie Wood responded optimistically to concerns about AI's impact on employment: “Many people worry that AI and automation will consume job opportunities, but a whole new world is opening up — space is one, and blockchain with immutable digital property rights is another. We are very excited about the AI era; it will be a pure job creator.” She concluded from the perspective of an entrepreneur: “Nowadays, everyone can write code using natural language—go start a business, create your own company.” Here is the full transcript: Cathie Wood (CEO and Chief Investment Officer of Ark Invest): Hello everyone. I am Cathie Wood, the CEO and Chief Investment Officer of Ark Invest. Today, I am here with our research and portfolio team to introduce the "Grand Vision for 2026." This research report is 104 pages long, and today we will distill its essence for you. We believe this research is similar to the kind of research investment bankers conducted in the 1980s and 1990s when personal computers emerged, trying to explore the future of technology. It sowed the seeds of what we believe to be a technological revolution. The seeds of this technological revolution were sown back in the 80s and 90s and have been germinating ever since. Now, we are in the midst of a full-blown technological revolution that requires original research to touch the future and understand what this revolution will trigger. Therefore, we are here today to share this original research with you. I am very pleased to introduce two new members of our team, OID and Shay, who are responsible for our "multi-omics" theme—I believe this is referred to as "genomics" in Europe. We believe this is the deepest application of artificial intelligence. From a revenue perspective, autonomous driving may be the largest application. However, the genomics revolution and the multi-omics revolution, we believe, will be the deepest applications of artificial intelligence. Let me introduce OID, who will host today's event. OID comes from RTW, where he focused on healthcare, particularly medical devices, diagnostics, tools, and so on. In terms of educational background, OID holds a Ph.D. in medical engineering and medical physics from Harvard University and the Massachusetts Institute of Technology. His qualifications are very well-suited for this field as the world of healthcare and technology continues to converge. So, OID, welcome. OID: Thank you, Cathie. I am very happy to be here with everyone today, good afternoon. I will be the host of today's meeting, and we are very excited to share the "Grand Vision for 2026" and respond to the over 800 questions we received. I am glad we can have some dialogue today and respond to the ideas we received. We have distilled the 800 questions down to about a dozen. I look forward to addressing each question. First, I would like to introduce the entire team present today. Starting from my right, we have Brett Winton, our Chief Futurist; Frank, our Director of Artificial Intelligence; Shay, as we mentioned, who is also an analyst on the multi-omics team; Tasha Keeney, our Director of Investment Analysis; and Dan McQuade, an analyst on our autonomous driving team. So, I would like to ask Brett to speak first. Could you talk about the various technologies we see converging at the highest level and how they are leading us toward what is called the "big acceleration"? Brett Winton: Of course. We believe that there are currently five major innovation platforms entering the market. Artificial intelligence is the core driving force, accelerating all other innovation platforms. Multi-omics, which Shay and OID will delve into. Public blockchains, including stablecoins and Bitcoin. Robotics, including humanoid robots and specialized robots. And reusable rockets, which Dan will discuss today. Let me think, have I covered everything? There's also energy storage, including opportunities in robot taxis and autonomous driving. Alright, I think I've mentioned everything. These five major innovation platforms are all reaching critical thresholds and giving rise to an unprecedented investment cycle. If you look at the acceleration of our investments in the underlying infrastructure required for technology, you have to go back to the railroad era to see a similar proportion of technological investment relative to GDP. This not only impacts macroeconomic growth in the short term—data centers are driving marginal GDP activity—but the accelerated investment in AI agents is changing the business landscape. They are also laying the foundation for sustained economic growth, as we will see positive investment returns from this technological infrastructure. Overall, we believe that by the end of this decade, this will drive a compound growth rate of over 7% in global real economic growth. This may sound a bit outrageous, but it actually aligns with economic history; when significant technological transformations occur, they bring about shifts in potential equilibrium growth rates. While the world generally believes we will grow at 3%, we see strong evidence that, driven by these AI-accelerated technologies, the macroeconomy will experience a growth inflection point. The market will follow the macroeconomy. Therefore, we believe that over 60% of the global total stock market value will belong to disruptive innovation platforms. This is the core message I want everyone to understand: you need to embrace innovation. Imagine a scenario where, in an unprecedented macroeconomic cycle, the growth rate shifts, and your core stock portfolio is unrelated to innovation; then, in the remaining years of this decade, as the value of these innovative companies skyrockets, the value of your portfolio may actually decline slightly. We believe this is entirely possible. Again, this is consistent with history. Believe it or not, in the late 1870s, 75% of the stock market value in the U.S. belonged to railroads. Well, we are in a similar investment cycle, and we believe these five major innovation platforms will generate similar corporate value accumulation. This is what we call the "Great Acceleration." Alright, OID, ask me a question. OID: Okay, Brett. Our first question, feel free to ask any of the many colleagues present if you think it's necessary. Question one: Are we overbuilding AI infrastructure relative to energy availability? What does that mean? Brett Winton: I think people are concerned about more than just energy availability. People might think, "Wow, so much money is pouring into AI, how can this possibly be a wise move?" The way we measure AI performance is how much value you and I, as knowledge workers, can derive from the AI software we use. Currently, we believe that if you fully utilize AI, you can achieve a 50% improvement in return on investment, meaning that for every unit of work you put in, you can get 1.5 units of output. Therefore, I should be willing to exchange half of my salary for that AI software Of course, companies will not pay half of the salary; they may only pay 10% of it, but this is still an astonishing income and a very cost-effective deal for companies. So, in this context, as AI continues to improve, we believe that AI software spending will reach $7 trillion, which is enough to support over $1 trillion in data center infrastructure construction. Energy may be a constraint in specific scenarios; for example, when building a data center in central Ohio, you need to find a way to bring energy in and build the data center. Many excellent NeoCloud companies are trying to achieve this. We do not believe this is a global constraint, but it is certainly a friction point that needs to be overcome. Cathie Wood: I would just add one thing. The root of this issue lies in the tech and telecom bubble and crash of the late 1990s. Think about the context Brett just mentioned; we laid a lot of fiber optics for the internet era in the 1990s, and that fiber sat idle for many years until recently when it was truly utilized. Today, every GPU is being used, and there is still a shortage. So, this is a very big difference. Brett Winton: Exactly, even older models of GPUs are still in use. The $7 trillion I just mentioned is only for text-based language models, primarily to assist corporate employees. There is another huge area, such as multi-omics, where there is massive data that we cannot process effectively; and there are physical robots, including autonomous vehicles and humanoid robots, for which we need more computing power. We need trillions of dollars in computing power, so I think you will see this trend develop in this decade. OID: The next question touches on one of the biggest examples in the fusion field. To what extent are space-based data centers feasible and economical? What technological and commercial milestones need to be achieved for them to compete with ground alternatives? Brett Winton: Well, Dan is the expert in this area, but let me try to say something, and you can point out where I go wrong. (Laughter) What I mean is, based on existing launch platforms, it is not economical. SpaceX's Falcon 9 can land, but it has limited tonnage it can send into orbit. Once they have Starship, the next-generation reusable rocket, we believe the cost per ton to get into orbit could drop to hundreds of dollars. Once we cross that threshold, then space-based AI computing power can compete with ground computing power in terms of cost. This is significant for reusable rockets because the required launch volume may increase by 60 times compared to before. This is a great example of how the acceleration of AI drives demand for another technology. This also relates to the previous question; if people in Ohio are vehemently opposed to building data centers, you have another place to put the data centers. This should allow AI computing power to scale and avoid some local restrictions that may be imposed by political factors. Cathie Wood: I would just add one thing. Elon Musk says this is an engineering problem. We understand that when Elon focuses on an engineering problem, the results are often different. Most people thought it was a bad idea to power cars with phone batteries, and no one did it But he was right, he succeeded. So, we are very excited when he is now truly focused on the space sector. (Laughter) OID: Great. Thank you, Brett. Thank you, Casey. Now, Frank Downing, if you could tell us how we view AI, and then we have a few more questions. Frank Downing: Okay. I will elaborate on our AI research. We believe this is a generational platform shift. In the past, we transitioned from the personal computer era to the smartphone era. Now, we believe AI is driving another generational shift in how we interact with technology. The new user interface is not from keyboard to touchscreen, but rather shifting to natural language. We will interact with computers in a fundamentally new way, making it easier and more powerful to control computers and technology. We will see a range of entirely new hardware forms to achieve this. I am wearing my Meta Ray-Ban glasses today, which have an AI assistant built in. Comparing this shift to the last platform shift through the internet era (including smartphones and cloud computing), we see that the adoption speed this time is twice as fast, if not faster. In just three years, we have reached 20% coverage of the relevant population, while the internet took seven years. So, everything is happening very quickly, and we can see this from the data. Part of the reason is the significant decrease in AI costs, with the costs of training and inferring AI models dropping sharply. This allows technology to spread through the economy at an extremely fast pace. We believe this creates opportunities for consumers, businesses, and physical robot forms (which we will discuss later). Just taking the consumer side as an example, we see personal AI agents becoming the preferred entry point for people to access products, services, and information on the internet. People are increasingly trusting ChatGPT or Claude rather than traditional Google searches (by the way, Gemini is also worth watching, as Google's adoption rate of consumer AI is second only to OpenAI). We believe this creates new monetization opportunities, including the subscriptions we see now, e-commerce where AI agents can handle transactions for us (making shopping more convenient), and as attention shifts to AI systems, advertising revenue will also flow to the new assistants we are interacting with. I will quickly give an example of how this is coming together in the e-commerce space. For instance, you can now initiate an app experience in ChatGPT, one of which is Instacart. I have long wanted fresh produce delivery, but entering various groceries one by one in the app is cumbersome; I would rather go to the store, as it is easy for me, and it is hard to change habits I have had for a lifetime. But with ChatGPT, I tried it a few weeks ago, you can take a photo of a recipe book, and through the Instacart integration, just say "place my order," and it can achieve 90% accuracy. You only need to change a few items. As the head of the AI lab likes to say, this is the dumbest the model has ever been. So this experience will only get better, creating revenue opportunities for Instacart that did not exist before, because that was work I used to do myself, and now I am happy to spend money to save time If I continue to talk about the opportunities in knowledge work, specifically the application of AI in enterprises rather than personal life. You may have heard the term "agent" has become very popular over the past year, especially since last December, there has been a lot of discussion about agent-based coding, with products like Claude Code significantly enhancing developer productivity. We believe that since the release of ChatGPT, we have seen this, AI excels at writing software. But the real fundamental turning point occurred in November and December of last year, when models were able to complete tasks over much longer time spans, which means they are more useful and have a stronger leverage on human time. There is no longer a need to constantly monitor them, answering questions every minute or every five minutes; AI agents can now reliably work for over 30 minutes. You can see this in various research reports (which we have displayed on our screen), where the average time that agents can reliably complete tasks increased from 5 minutes to 30 minutes during the period leading up to 2025. In fact, the latest data points have reached over 55 minutes. So, we have seen this huge inflection point, which is actually showing exponential growth, with the capabilities of agents becoming stronger, increasing the willingness of enterprises to pay for it. We looked at the cost of ChatGPT subscriptions, with the basic enterprise version costing between $20 to $40 per month. Based on the time saved by knowledge workers reported by companies, the cost of paying for a monthly subscription can be recouped in less than a workday's worth of time. Therefore, we believe there is still a lot of room for AI monetization. Cathie Wood: I just want to add one point, which responds to the first question. The reasoning ability you just mentioned really took off, and the reasoning ability for long-term tasks took off in November, which greatly stimulated the demand for GPUs. Absolutely. OID: Great. I think the next two questions touch on these two points. First, where has AI created real net new revenue, rather than just efficiency gains that compress margins? How does Ark distinguish between signal and noise? Frank Dowling: That's a good question. I think the example I gave with Instacart is a good example, as it represents revenue that did not exist before, new revenue creation. At the public company level, we see, for example, a surge in demand for computing power, which we have seen from chip companies. All the cloud service providers, AWS, Azure, GCP, have experienced accelerated revenue growth over the past few years. GCP is growing the fastest, with a 48% year-over-year growth for a $70 billion business meaning a lot of new revenue. But I think this question may not only be about the enablers but also about the ultimate beneficiaries and where they are seeing revenue growth. I think companies like Palantir have played an important role in demonstrating how this is happening, which is also why their business is growing so rapidly. I’ll give an example from the insurance industry, where they work with clients like AIG, who receive more insurance applications than they can actually process, thousands more, and humans simply do not have the time to review them. They have some prioritization methods, but still, revenue is left on the table Now, with the help of Palantir, they are using AI agents to assess and underwrite contracts that were previously uninsurable, bringing new revenue to the business. I believe this situation exists in every different industry in the economy, where there are things we can do, but today we lack the time and resources to do them. Overall, this creates a scenario where AI not only reduces operational costs but also massively expands the market. OID: As companies like AIG and other insurance firms continue to adopt AI, what do you see as the biggest bottleneck for AI expansion in the next three years? Is it power, computing power, data quality, regulation, or talent? Frank Downing: Another good question, and the market has been discussing what the current bottlenecks are. I think the trend we are seeing is that the ultimate bottleneck is power and computing power. I mention them together because I believe they are part of the same issue. Think about it: if OpenAI wants to launch a new product, or if Claude Code is scaling, or if Anthropic wants to onboard new users, they need GPUs, data centers, grid power, or to build their own power plants like xAI. All of these need to come together. I think this is the main bottleneck, compared to data or talent, because of the latest models and research trends we see in AI labs: models are increasingly generating their own training data. Human seed ideas and human thinking are important, but then scaling with synthetic data generation, and the models themselves are also involved in finding new algorithmic advancements to improve their own performance. As OpenAI mentioned when discussing their latest coding model, this is the first model where the previous generation of models played a significant role in training the next generation. This also alleviates some of the talent bottlenecks, although talent is obviously very important, which is why so much attention is focused on the flow of talent among the four major AI labs. So, my ranking is that computing power comes first, if you have data centers and power to kickstart the chips. Brett Winton: Yes. And you can also make trade-offs. Interestingly, in the past, people used to say we would run out of data, but the emergence of thinking chains has made us realize that we can actually leverage additional computing power to generate more data based on existing data. So, if you encounter a bottleneck in one area, you can exchange it with another resource to enhance AI capabilities. OID: Great. Thank you, Frank, thank you, Brett. We will move on to the next section, where we will explore multi-omics with Shay. Shay, in your section, you predicted the questions well, so I think we will use your slides to discuss some issues. I will jump straight to the first question: as AI accelerates drug discovery and diagnostics, where do you think the biggest value capture lies in the multi-omics stack? Is it data generation, model development, or therapeutic commercialization? Shay: Perfect. Thank you. That's a great question, and I can understand the intuition to want to pick one layer from the stack, but in reality, we see these components are mutually reinforcing. AI truly integrates here and acts as a key pivot, driving the flywheel of biological innovation What I mean is that better data, more data input into better models, leads to better diagnostics, better treatment interventions, and better tools, which in turn generates more and richer data. In fact, this is a virtuous cycle. We organize this cycle around four key areas: first, multi-omics tools, which obtain better data at a lower cost; second, molecular diagnostics, which can detect diseases earlier and more accurately; third, AI-developed drugs, which leverage these biological insights to develop better drug candidates and bring them to market faster and cheaper; and finally, curative therapies, which are one-time treatments targeting the root causes of diseases. Therefore, they do not operate in isolation but reinforce each other like a flywheel. This is why we do not view it as a single-dimensional aspect. What truly drives the acceleration of this flywheel is the significant decrease in costs. Looking back a few decades, the first human genome sequencing, part of the Human Genome Project, took about 13 years and cost nearly $3 billion, including all the infrastructure needed to complete it. Today, you can sequence an entire human genome for just $100. Looking ahead to 2030, we expect this cost to drop by another order of magnitude to around $10. This cost curve truly changes the paradigm of who gets tested, how often they get tested, and the volume of data generated. Therefore, it is conceivable that as costs decrease, the volume of testing will increase. This is a very important shift, and by 2030, we expect the volume of testing to double. Notably, the total number of tokens or data we have generated is already comparable to the number of tokens used to train large frontier language models. We expect this scale to increase tenfold by 2030. Here, I want to emphasize that overall, biology is becoming one of the largest data-generating engines on Earth, driving real transformation across the healthcare sector. Cathie Wood: If I may add, as a portfolio manager, it is a privilege to learn about these. But regarding data generation, I once asked the team, how many cells do we have in our bodies? The answer is 35 to 40 trillion. Now we have single-cell sequencing technology. This gives you a sense of the scale of the data explosion, which will dwarf anything we have seen in the computing era. OID: Great. Shay, one point you mentioned is that part of the flywheel is the impact on drug discovery and drug development. The question is: Can AI materially reduce the costs, duration, and failure rates of clinical trials? What does this mean for the capital efficiency of biotechnology? Shay: Indeed, this gets to the heart of the matter. It goes back to the idea of richer data input into better models. One of the clearest impacts we might see is on the economics of drug development. If I frame today’s issue, drug development can take over a decade and cost billions of dollars, with 9 out of 10 candidates entering clinical development ultimately failing. There is clearly a problem here that requires higher efficiency. The dynamic created by AI is that you can bring drugs to market faster, thereby generating more revenue from patent protection and reducing costs. This has a real compounding effect From our own modeling, we can see that this could reduce the time to market by 40% and lower the actual cost of drug development by four times. This is a very significant shift. The question touches on capital efficiency, implying that there is a clear issue here: historically, the return on investment in drug development has now fallen to single digits. However, when you consider the compounding effects of faster time to market, lower costs, and higher success rates, this paradigm shifts. If you take it a step further and apply it to curative therapies, this shift becomes even more pronounced. Historically, early-stage assets had almost no economic value, but in contrast, AI-driven curative drugs could each be worth over $2 billion. So, to directly answer the question, yes, we are reflecting a significant impact on biotechnology capital efficiency in our models. Cathie Wood: Let me add another perspective. The golden age of healthcare was in the 1980s and 1990s when the return on R&D spending reached 30%, and now it has dropped to low single digits. We believe that returns will return to the golden age, and considering the current market's psychological expectations for healthcare, this will be quite unexpected. OID: Great. I have a bit of a preference for Shay's part here, so I want to squeeze in one more question: As these technologies scale, how should investors consider the regulatory, insurance reimbursement risks, and the path forward for gene therapy? Shay: Perfect. (Laughter) Let me put it this way, this is a multi-layered question. Let's break it down and look at it layer by layer. In terms of regulation, especially over the past year, we have seen a lot of changes. In the U.S., the agency responsible for drug approval is the FDA. They have recognized how difficult it is to bring drugs to market, as I just described. So, their thinking is to modernize their agency and hope to work with drug developers to streamline the clinical development process so that we can truly address this issue. We have already seen some new frameworks emerging from this, particularly for rare diseases and therapies targeting root cause biology. That’s one layer. The other part of the question is reimbursement and insurance barriers. Of course, I can understand that a price tag of $2 million for a curative drug might be shocking and lead one to question, "How can this be reimbursed? I foresee potential insurance barriers." Let's return to a real-world example. Take CRISPR Therapeutics' Casgevy, which is an approved gene-editing therapy for treating sickle cell disease and transfusion-dependent β-thalassemia. Its price is slightly above $2 million; however, 90% of U.S. patients have already gained access to reimbursement. Why? The reason is that you need to compare the price of this drug with the chronic treatments patients need to undergo and the hospitalizations they need to experience; its value to the healthcare system justifies the price. This is the key to understanding the reimbursement issue. What it really highlights is that the economics of curative drugs are fundamentally different from traditional drugs. Here, curative drugs are one-time treatments, and you receive all the value upfront. It brings cash flow forward, allowing for more revenue protected by patents and potentially avoiding competitive overlap This means that curative drugs may be more valuable than traditional drugs, with a potential value in our model of up to 20 times. To be more specific, I will quickly illustrate with a case study on hereditary angioedema. HAE is a rare disease that causes painful and potentially life-threatening swelling attacks. Currently, patients require chronic treatment throughout their lives to help control these attacks, with lifetime costs potentially ranging from $10 million to $20 million. For example, Antellia Therapeutics is developing a gene-editing therapy that has shown promising clinical data. We estimate its real-world price could be $3 million, but the value-based price could be three to four times that number, ultimately depending on efficacy and durability data. If applied to all 7,000 HAE patients in the U.S. today, this would save the healthcare system $52 billion. The point emphasized here is that even with a high upfront price, you can actually deliver better outcomes for patients, relieving them of the burden of lifelong symptom management while also saving the system a significant amount of money. Finally, regarding the scaling aspect of the issue, as this is very important. There is a significant shift happening, which I will briefly highlight, as we have blog posts published on the Ark Invest website. One major shift occurring is that gene-editing therapies are beginning to evolve towards "in vivo" editing, meaning editing within the body. This helps facilitate the transition from rare diseases to common diseases, including the world's number one killer—cardiovascular disease. We show here that the value-based price could be $165,000; note that this is completely different from the sometimes million-dollar pricing of rare disease therapies I just described, but you still have a huge potential market opportunity. If you only consider the highest-risk patients in the U.S. and multiply by this value-based price, you will arrive at a $2.8 trillion market opportunity. Simply put, Lipitor has been the best-selling drug for years, and capturing just one-twelfth of this market would match Lipitor's cumulative sales over the past 20 years. So, what I want to convey from this multi-omics part is that the true integration of AI and biology is driving huge opportunities and significant transformations in the healthcare sector. I hope everyone feels as excited about this as we do at Ark. Cathie Wood: Well, the stock market hasn't fully reflected this yet, but it will catch up. What I want to say is that the biggest surprise for me is that insurance companies have no problem with the $2 million price tag, and I don't think the market has even noticed this. OID: Great. Thank you, Cathie. Thank you, Shay. Alright, Tasha, it's your turn now. Please tell us about autonomous vehicles. Tasha Keeney: Okay, about autonomous vehicles. Frank mentioned physical artificial intelligence. We believe this will be the first large-scale implementation of physical artificial intelligence that consumers will see, and it is already happening today. We already have cars on the road driving without anyone behind the wheel or in the passenger seat; they are picking up passengers and navigating around. We think a very important point among the existing participants is the underlying cost of the cars This is particularly important in the early stages of commercialization, as when your fleet is small and you are trying to scale up, the cost of the vehicle is indeed significant, but so is the cost per mile. The cost per mile is precisely what we believe will drive the demand for this technology and innovation. For example, look at Tesla versus Waymo; we have already seen that on a per-mile incremental cost basis, the Model Y has reduced costs by over 30% compared to the fifth-generation Waymo vehicle. We expect this advantage will only increase with the upcoming models. Looking at the Cybercab versus the sixth-generation Waymo vehicle, we anticipate its cost advantage will reach 50%. Similarly, this is extremely important when scaling early platforms, but it is also crucial for competitively pricing these platforms to consumers. When it comes to pricing, we believe that the minimum price that a robotaxi platform might charge after scaling could be around 25 cents per mile. In contrast, this is less than one-tenth of the cost of human-driven ride-hailing in Western markets, less than half the cost of driving a private car, and cheaper than ride-hailing in China, which is already very inexpensive, at about 50 cents per mile. Therefore, we believe that the reduction in costs will significantly expand the current ride-hailing market, making low-cost point-to-point travel accessible to more people and ultimately making our roads safer. But we believe there is also tremendous market potential here. We estimate that by the end of this decade, robotaxis could create a $34 trillion enterprise value opportunity, and we believe this portion of value will accrue to what we call "autonomous driving technology providers" or platform operators. These companies are developing autonomous driving technology in-house to drive vehicles. They can provide this technology through ride-hailing services and robotaxi services. Casey mentioned that we believe this is also one of the largest revenue opportunities. The total potential market revenue for robotaxis could reach $10 trillion or more. By the end of this decade, we believe revenues and profits here could reach around $2 trillion. I won't go through each participant on this slide, but this is just to give you an idea of the companies globally that are researching this field. As I mentioned, you can see at the bottom that we believe platform operators, i.e., autonomous driving technology providers, are truly positioned to capture the largest economic share here. This is because these companies are the ones capable of lowering the per-mile price, thereby truly expanding the market. We also see electric vehicle manufacturers starting to get involved, collaborating with autonomous driving technology providers and ride-hailing companies as potential customer generators. At the same time, we see these companies transforming their business models; they are becoming maintenance service providers for autonomous driving technology companies and automakers. I just want to say that we expect this to fundamentally change the entire automotive industry. We believe that many participants currently operating on traditional fuel-powered platforms will undergo significant consolidation. We believe the future of robotaxis is electric. Electric vehicles are crucial for regaining attractive per-mile cost economics. In the U.S., we see ride-hailing price caps at around $2.80 per mile, which is quite good. In China, ride-hailing is much more competitive, which is driving many participants to turn to markets like the Middle East that may have greater profit potential Robotaxi has already been in operation today. I'm leaving this information for everyone. Dan is tracking every incremental mile we see on these platforms. Today, we have seen nearly one million miles driven on the robotaxi platform. So, the question is just when it will scale, and achieving scale is about the proliferation of fleets and, again, the lower cost per mile mentioned. Cathie Wood: I wasn't planning to speak after everyone, but I just wanted to add an exclamation point. I've been to Europe many times, and I realize that when I talk to people about robotaxis and the incredible research our team is doing, they can't resonate because the regulators in Europe are not yet in place. But we believe they will eventually be in place because the safety statistics are so astonishing. In the long run, it may be unethical for regulators to block this trend. OID: Or more directly, I think you should interpret certain regulatory actions as actively killing people. Cathie Wood: (Laughter) Very diplomatic. Just to return to the "Big Acceleration" part, in the U.S., the hidden labor cost of unpaid human driving exceeds $4 trillion annually. Think about it, converting $4 trillion of non-monetized activity into something you can pay others to do for you at a price lower than your time cost will ultimately bring about an economic transformation. You are transforming an activity that was previously unmeasured into economic activity, which is GDP. OID: That is precisely the answer to our first question. Tasha, which regions are most likely to achieve large-scale autonomous driving deployment first or last? How decisive is regulatory coordination for commercial success? Tasha Keeney: Absolutely. Interestingly, the U.S., because regulation has always been state-based, is one of the earliest markets to allow large-scale testing of robotaxis, which also makes it one of the earliest markets to commercialize robotaxis. Now, China is also placing great emphasis on the opportunities in autonomous driving, so we see significant scale from Chinese participants. So, I think these two markets will be the first to reach scale. But as I mentioned, the Middle East is also an attractive opportunity, especially for those Chinese participants who may see overseas profits exceeding domestic profits. Regarding regulation, as Cathie mentioned, we have already seen these platforms are much safer than human drivers. Years ago, we estimated that robotaxis could be over 80% safer than human drivers based on accident rates. Today, we have actual data from Waymo and other platforms to prove this. Tesla also regularly releases safety statistics for its full self-driving software. So, we know it is already safer. Therefore, the technology is mature. We expect regulation to be very important for allowing its proliferation. But today we have already seen it on the road, and we expect this to happen in the next 5 to 10 years. OID: We have talked about how regulation may become a bottleneck in this field. So perhaps you could briefly emphasize what you think the key bottlenecks for widespread deployment are? Is it regulation, safety validation, computing power, mapping, or telecommunications infrastructure? Tasha Keeney: Yes. Again, the most important takeaway is that the technology is already there. So, it's no longer a technological barrier. However, to truly scale robotaxi services—because they are currently only available in certain cities with relatively small fleets, compared to the scale we expect to reach in about five years—it requires companies like Tesla that can bring cars to market. Of course, Waymo needs to collaborate with other automakers, and Chinese participants need to both manufacture their own cars and work with car manufacturers. So, again, this emphasizes the importance of low-cost automotive platforms to expand robotaxi fleets and provide attractive products to consumers. Regulation is certainly important, but we already have safety proof points showing that autonomous driving technology is far superior to the current state of human driving. So, it's really just a matter of companies executing on scaling, and as we scale, we expect the cost per mile to continue to be lower than today's ride-hailing prices, which is what will truly expand the market. OID: Thank you, Tasha. Next, let's hear from Dan. Last but not least, shifting from the autonomous driving we see more and more in our daily lives to something we might not see as often—reusable rockets. Dan Muir: Of course. (Laughter) Yes, you will see more of them in the future. So, the reusability of rockets is really opening up the space economy. 2025 is a significant year, with annual orbital mass reaching an all-time high. This is largely thanks to SpaceX. SpaceX currently has over 9,000 active Starlink satellites operating in Earth orbit, which accounts for more than two-thirds of all satellites in orbit. They have this dominance because they are ten years ahead of the industry. What do I mean? In 2015, SpaceX successfully landed an orbital-class booster for the first time. Since then, they have executed near-perfectly in partial reusability, while their closest competitors only landed boosters for the first time at the end of last year. So, while other companies are still struggling to master partial reusability, SpaceX is moving full speed ahead with full reusability, which is really translating into lower launch costs. A key concept from Ark's research is the Wright Law. In the context of launch costs, the Wright Law states that for every doubling of cumulative orbital mass, the cost per kilogram of launch decreases by 17%. SpaceX's Falcon 9 rocket has proven this to date. According to our research, we estimate that since 2008, they have reduced launch costs by about 95%. The result is that it has opened up many opportunities in the space age, including the orbital data centers Brett mentioned earlier, and zero-gravity testing in the fusion field for advancing medical progress in the multi-omics area we discussed. Today, the cost is about $1,000 per kilogram. According to our research, when SpaceX achieves full reusability of its rockets, namely its Starship rocket, we believe costs could drop below $100 per kilogram. At that point, discussions about orbital data centers will truly become exciting. I know Brett has done a lot of research in this area At that scale, we believe that orbital data centers could be about 25% cheaper than ground computing. But there is still a long way to go to get there. We are excited to focus on the progress of Starship this year. I think this is a quick overview of reusable rockets. I know we could talk about it all day, but we have covered a lot of topics today, and I believe you have some questions that I would be happy to answer. OID: Great, Dan. Thank you. The first question: Has reusable launches permanently reset the cost curve for entering space? Or does it need to be reduced by another order of magnitude to unlock the next wave of orbital infrastructure? Dan McQuaid: Absolutely. We have a bad habit of answering questions before they are asked, but (laughter) yes, perhaps just to emphasize again, the key is complete reusability. Today, the Falcon 9 has a two-stage rocket, and the upper stage is lost after launch, while the lower stage is recovered. The goal of Starship is to have a two-stage rocket where both stages are recovered after launch. The key is that this will reduce costs by an order of magnitude. We believe it can achieve below $100 per kilogram. Reiterating the orbital data centers, this indeed opens up many opportunities that were not economically viable before. So, launch costs are indeed critical for all orbital infrastructure in space. Brett Winton: Yes, just to reiterate, from about $700 for Falcon 9 to around $100 for Starship, it brings about space-based AI computing. The number of satellites required for space-based computing is at least ten times the number needed for today's communication constellations like Starlink, and possibly even more. So, just this alone expands the market by about an order of magnitude. Then, if we colonize the Moon or establish a lunar base and can use mass-based launchers, we could potentially service Earth satellite constellations at about $10 per kilogram. This could further reduce the cost of sending things into Earth orbit by another order of magnitude. But this requires building a lot of infrastructure on the Moon. So, that would be another completely different infrastructure investment opportunity. Cathie Wood: I like to say that many people are worried that AI and automation will destroy job opportunities. But we have a whole new world opening up. Space is one, and another is actually related to blockchain technology and immutable digital property rights. I think we will see an explosion there. So, we are very excited about the AI era, and we believe it will net create job opportunities. OID: Speaking of net job creation and long-term changes, where does long-term value accumulate in the reusable rocket ecosystem? Is it in launch providers, satellites, satellite networks, or downstream data and services? Dan McQuaid: Certainly. I mentioned launch providers and launch costs. We believe that the near-term cash flow opportunities are in satellite connectivity. Today, I believe many people have heard of Starlink, which has just surpassed 10 million active users. This explosive growth goes back to Wright's Law, which is key to our research. We believe that for every doubling of on-orbit terabits per second, satellite costs will decrease by about 44% This is a very steep cost decline curve, leading to this explosive growth. When scaled, we believe this could be a $160 billion revenue opportunity annually. That's why you see many space companies entering the public market, trying to capture this huge share. OID: Great. Thank you, Dan. We have completed all parts. I have some general questions here that might belong to the rapid Q&A session. I'll throw them to the group and see who wants to take them; multiple people can participate. I'll start with the first one: How does Ark's fusion stack thesis unfold between AI, robotics, energy systems, and public blockchains from 2026 to 2030? What are the most significant bottlenecks? We've already asked several questions about bottlenecks today: power, computing power, talent, regulation, capital efficiency. Brett Winton: Well, I think we've answered how it unfolds across all technology domains. What I want to say is that from a diversification perspective, Ark's exposure to multiple technologies is crucial. You might be fully invested in AI enterprise software, but you could encounter bumps along the way. And those bumps should not be related to the successful pricing of therapies in multiple omics fields in the market. So, while AI is the accelerator for all technologies, they have different commercialization friction points and market opportunities. We are increasing the momentum across all technology domains and then trying to predict how it translates into cash flow momentum, which then gets reinvested. As for bottlenecks, I do believe the world needs more computing power. So, whether it's space-based data centers (which is an orthogonal direction that can incentivize computing power) or the continued expansion of U.S. foundries outside of Taiwan, we have chips to continue building computing power, and many companies are making a lot of money from it. For example, there is a supersonic civilian aircraft company that will actually benefit from this because its engines are very suitable for powering AI data centers. So, Boom has suddenly emerged with a large business providing power for computing. The capital markets are flooding in to fuel this opportunity. We believe this is the most important catalyst among all that is happening. Unit growth is crucial, as Daniel has mentioned several times regarding Wright's Law, unit growth. OID: So, if the question is about bottlenecks or what might hinder development, of course, disasters like global wars would hinder. But interestingly, even in those tough times, businesses and consumers are willing to change the way they do things. They are looking for better, cheaper, faster, more productive, more efficient, and more creative solutions. Ironically, COVID-19 is actually an example. The supply chain was once blocked, but now we have moved past that and are moving forward. Cathie Wood: Yes, what I want to say is that the biggest competitor to disruptive technology is actually inertia and the status quo. That is the biggest competition. So, I think the world is a bit tumultuous, and everyone is aware of AI and concerned about what it means, which is prompting people to take action and embrace this technology. This provides us with evidence that we need to invest hundreds of billions of dollars in more computing power here because these companies lack the computing power to serve their customers OID： Brett, you mentioned concerns. We've certainly heard some over the past few weeks. So, what is our view on the future of enterprise software and SaaS in an agent-based AI world? Which business models will be disrupted, and which will be strengthened? Brett Winton: Interestingly, you mentioned enterprise software, as if everyone is avoiding it now. But I think our view is that AI is transformative for software, but it won't necessarily destroy the entire existing landscape. This is because AI makes it easier to create new software than ever before. Some enterprises will choose to create their own software or enhance their software using internal capabilities. But I think more likely, we will see a wave of new competitors emerging from many of the existing companies. Instead of everyone building their own CRM, there will be many new competitors that are potentially more AI-native, more agile, and more targeted to specific industries, which will change the future expectations for revenue growth and pricing power of existing companies, and that's why the market is avoiding them. We haven't seen some AI-native companies enter the public market yet, but we've seen many seedlings in the private market, involving many different industries or functions, whether it's Sierra in customer service, Harvey in legal, or Cursor in software development. For example, Cursor just broke a $2 billion annualized revenue run rate. This company has reached a $2 billion annualized revenue run rate in just three years. In the cloud computing era, reaching a $100 million annualized revenue run rate is a significant milestone. Companies like Twilio, I remember it took six years and 500 people. Cursor achieved 20 times the revenue in a shorter time with fewer people. This is possible. Some of these new software companies will be outstanding, just not necessarily the ones currently in the public market. Cathie Wood: I think a very important point is that despite what Frank just said, we believe an entrepreneurial explosion is about to happen because we can all program with natural language now. So, go start a business. OID： In the spirit of the entrepreneurial explosion, let's wrap up here today. Thank you all for participating today. We hope you enjoyed the deep dive into "The Grand Vision for 2026." If you haven't downloaded it yet, please do so and read it, and feel free to reach out to any of us through the website or social media. We hope you enjoyed this time, and have a great rest of your day. Let's embrace innovation and move forward