Cathie Wood Shares Latest Views: In-Depth Discussion on AI Healthcare, Robotaxi, and the Open-Source vs Closed-Source Model Debate

Deep News10:52

Renowned American investor and founder of ARK Invest, Cathie Wood—widely known as "Wood Sister" and a benchmark figure in global active ETFs—has long been famous for her forward-looking bets on disruptive technologies and bold predictions. Her assessments of cutting-edge tech industries and capital markets have remained distinctive and continue to draw widespread investor attention.

Recently, at the "Dialogue with ARK (Season Two): Wood Sister and Huabao Fund Overseas Investment and Active ETF Institutional Private Sharing Session," hosted by leading domestic ETF institution Huabao Fund and co-organized by ARK Invest and Oming Asset Management, Cathie Wood made a surprise appearance via a trans-Pacific video link and held an in-depth conversation with Zhou Jing, Chief Investment Officer of Huabao Fund. Cathie Wood answered a wide range of hot topics in capital markets and the tech industry in one sitting, such as: "Beyond chatbots and coding, where is the biggest money-making opportunity in AI applications?" "What competitive landscape has Robotaxi already shown or is likely to show?" "In the closely watched AI healthcare field, who will capture the value?" "How do you view the debate between open-source and closed-source paths for large AI models? How will these two paths shape the AI industry landscape?" In response to these questions, Cathie Wood held nothing back, offering clear and distinctive views backed by research data, making the entire conversation highly informative and compelling.

Reshaping Healthcare: Companies Embracing New AI Technologies May See Significantly Higher Returns on Investment

Zhou Jing: You have mentioned that AI applications are not just about chatbots or coding, and that you believe the biggest money-making opportunity is in the healthcare sector. We have noticed that you have an ARKG ETF, and especially this year, this fund has delivered considerable returns for investors, with some outstanding holdings in its portfolio. So, could you tell us how AI actually saves money for the pharmaceutical industry? Does it help scientists find promising new drug targets faster, or shorten the years required for clinical trials? In terms of saving time and cost, which companies will benefit the most?

Cathie Wood: We believe the most profound application of AI is not chatting or writing code, but multi-omics, carried by biopharmaceuticals. The market cannot understand it, and the reason is not technology but two analyst cultures: healthcare analysts are used to regulation, reimbursement, and a slow pace; tech analysts are used to "move fast and break things." The two systems are merging, but this also creates confusion—which is precisely the source of pricing dislocation. However, there is a series of opportunities here that will completely transform healthcare.

The first is the research, discovery, and development of new drug targets. Today, a large pharmaceutical or biotech company needs to spend $2.4 billion and 13 years to bring a new drug to market. We believe this will change dramatically: over the next five years, we expect the cost of a new drug from research, discovery to development to drop to $600-700 million, and the time to shorten from 13 years to 8 years or even less. Considering that patent protection lasts 21, 22, or even 24 years, an extra 5 years of revenue generation is clearly a huge opportunity.

The second is extremely early disease diagnosis. This refers to diagnosing diseases through a single blood test at a much earlier stage than currently possible. Take colorectal cancer as an example—it can be diagnosed at stage one or even before stage one, because tumors release substances into the blood, and so do polyps. Thanks to the integration of sequencing technology and artificial intelligence, we can identify these cancers and other diseases very early. And if diseases can be detected at an earlier stage, survival rates will increase significantly.

The third and perhaps most important application—curing diseases. Thanks to sequencing technology, AI, and CRISPR gene editing technology, we already have the ability to cure diseases. Currently, CRISPR Therapeutics has been able to cure two rare diseases—sickle cell disease and beta-thalassemia. These are rare diseases, but they are also targeting some common conditions, such as "bad cholesterol." We see better targets, better molecules, better trials, and better decision-making throughout the development process: we can identify failed hypotheses earlier, avoiding years of time and large amounts of money spent on wrong biological assumptions; we can design drugs more efficiently, exploring more candidate molecules in a digital environment and only sending a few into physical testing; we can also improve clinical development—with better biomarkers, better patient screening, and better trial design. Of course, safety and efficacy still require clinical evidence. What AI can do is make the path to evidence smoother. We believe development cycles will be significantly compressed, and the ultimate winners will be those companies that combine proprietary data, powerful models, and clinical execution capabilities. Therefore, we believe that companies actively using AI tools will see their return on investment rise from the current approximately 5% (or lower) to 30% to 40% or even higher. We saw a similar situation in the 1990s, at the dawn of the biotech era, when monoclonal antibodies rose and Genentech stood at the forefront of that wave, with its return on investment entering the 30% to 40% range. We believe that companies willing to actively embrace these technologies will see significantly higher returns. On this point, we salute China. China's regulatory liberalization has been very substantial, and it has become a major source of licensed targets for many of our biotech and pharmaceutical companies. As far as I know, the United States is also beginning to follow China's example in easing regulation, so that we can enter this new world faster.

Robotaxi: China and the US Will Each Lead in Their Respective Markets

Zhou Jing: Let's talk about an area you are very interested in—Robotaxi. We know you invested in Tesla very early, and to this day, Tesla remains one of the largest holdings in your portfolio. We also know that Musk launched the Robotaxi business in Austin, Texas last year; in addition, Google's Waymo has been operating in San Francisco for years. Could you walk us through the Robotaxi market landscape, especially in the United States? Also, we know you have invested in some Chinese Robotaxi companies, including WeRide and Pony.ai. Please also share your views on how Chinese Robotaxi companies compare with global peers in terms of cost and international expansion.

Cathie Wood: In the United States, Waymo first achieved commercialization around 2018, but has been advancing very slowly and very cautiously, having to go through a lot of regulatory red tape. I think Tesla is happy to let Waymo go first through all the regulations, essentially paving the way for autonomous mobility. In the United States today, Waymo completes more than 500,000 fully driverless trips per week, and is growing very fast—in fact, their goal is to reach 1 million fully driverless trips per week by the end of the year, and we think they can do it. They currently operate in 15 cities. However, the number of autonomous vehicles they have on the road is still less than the number of vehicles Tesla produces in a single day. As you mentioned, Tesla launched its Robotaxi business in Austin in June 2025, and has now entered 8 cities, accumulating approximately 2 million miles of driving data. Although Tesla's Robotaxi business started seven years later than Waymo's, it reached 1 million miles of driverless driving six times faster than Waymo did for its first 1 million miles. As you know, Tesla recently unveiled the Cybercab, and we believe that by 2030 or 2031, when Waymo's next-generation vehicles scale up in five years, the Cybercab's operating cost per mile could be 40% lower than Waymo's. So in AI, the "winner-takes-all" logic exists: the company that can get passengers from point A to point B most safely is extremely important, and I know this is also very important in China. At that point, the safest, fastest, and lowest-cost company will win. In the United States, we believe that company is Tesla, and we believe the market evolution will be very similar to Uber and Lyft: in the ride-hailing market, Uber took the largest share, and Lyft came in second but with a much smaller share; in autonomous driving, we believe Tesla will take the largest share, and Waymo will take most of the rest. As for the rest of the world, it is indeed dominated by China. Our portfolio includes Baidu, WeRide, and Pony.ai. In Europe, we hold Wave, which partners with OEMs and is effectively replacing a company called Mobileye. We believe this partnership could potentially land with Volkswagen. Wave has been serving OEMs and building partnerships, starting from lower levels of autonomous driving, but still has L4 or even L5 ambitions. China is very focused on its domestic market, but interestingly, in China, the ride-hailing market with drivers already has extremely low travel costs. I know in Wuhan it is 35 cents per mile, and the China average is about 50 cents per mile. So China's autonomous driving situation is: since they cannot provide much higher economic value compared to ride-hailing, they are looking to other parts of the world, especially Europe and the Middle East, to develop their autonomous driving networks. As I said, ride-hailing with drivers averages 50 cents per mile in China; in the United States, the cost of ride-hailing operations is $2.80 per mile, five to six times higher. So if autonomous driving can be achieved at 50 cents per mile, or even $1 per mile, it would be incredibly cost-effective. That is why I think autonomous taxi networks will take off faster in the US, Middle East, and Europe than in China. As for how Baidu, WeRide, and Pony.ai perform in Europe and the Middle East, we will wait and see, and we are very excited to see them enter this space, because we believe it will accelerate the transition from "human ride-hailing" to "autonomous ride-hailing." I think China can become a leader—in the rest of the world outside the US; and in the US, given the competitive landscape between Tesla and Waymo, we believe they will be the leaders here.

Who Will Capture the Value of AI Healthcare? Track Who Anthropic and OpenAI Are Partnering With

Zhou Jing: Among all the areas you follow, such as Robotaxi, AI pharma, and autonomous driving, which one do you think is currently the most undervalued?

Cathie Wood: This year is the first year that "multi-omics" has gained recognition. As I said at the beginning, we believe the "multi-omics" field—biotech, pharmaceuticals, healthcare—will provide the most profound application of AI. We believe better data will lead to better predictions and better treatments. We also believe that weaving together these layers of biology will improve disease detection and target identification. This was impossible in the past, and it was impossible because we did not have the corresponding knowledge. On top of that, there is a collapse in costs. To give you a sense: every time the cumulative number of human genomes sequenced doubles (1 to 2, 2 to 4, 4 to 8), the cost drops by 40%. On the AI side, I mentioned that AI training costs are falling 75% per year, and inference costs are falling 99% per year, thanks also to DeepSeek, which showed us how to bring costs down—we originally estimated an 85% annual decline, but the emergence of DeepSeek brought it to a 99% annual decline. So, we believe we will see: earlier disease detection, higher drug development success rates, and possibly "cures," creating value far beyond what traditional forecasts can imagine.

So who will capture this value? We focus on and track companies like Anthropic and OpenAI—who are they partnering with? They are partnering with Twist Bioscience, 10x Genomics, and Tempus AI. These three are not only three of the largest holdings in our ARKG (Genomic Revolution ETF), but also important holdings in our flagship strategy ARKK. Our flagship strategy allocates approximately 28% of its portfolio to this AI revolution in healthcare, which is very different from most tech and enabling innovation funds, and we are very proud of that.

The Debate Between Open-Source and Closed-Source: Each Has Its Strengths; Focus on Proprietary Data, Customer Relationships, Workflows, and Pricing Power

Question: As is well known, the US model ecosystem is basically closed-source, while China has shifted to open-source. How do you view the impact of these two paths on the AI industry as a whole? What are their respective comparative advantages?

Cathie Wood: The "DeepSeek moment" helped us recognize how mature China's open-source ecosystem is. I have always liked open-source. I watched Linux grow up and dominate data centers, so I have great respect for open-source. However, most of the US is closed-source. We can see that both paths have their own roles. From the perspective of closed-source models, these companies can provide integrated products, management, services, and monetize directly. From the perspective of open-source models, we see customizability, deployment control, and broader availability. In addition, lower costs expand adoption, and open-source clearly helps in this regard. They can also put pressure on the pricing of closed-source models and improve the economics of downstream applications, which is a good thing. The investment test will be: who has differentiated data, that is, proprietary data; who controls customer relationships, who controls workflows, and where pricing power lies.

At the end of the conversation, Cathie Wood also posed two questions to Zhou Jing—

Cathie Wood: When building AI exposure through ETFs, how do you choose between broad tech funds and more focused themes (such as semiconductors, software, or robotics)?

Zhou Jing: We usually prefer more targeted thematic ETFs, especially when the goal is clearly to build AI investment exposure. Our approach is top-down. Our ETF portfolio fund is a fund focused on the tech sector. Therefore, we first need to determine the proportion we want to allocate to AI. Then, we form a macro judgment on the AI investment cycle and identify which segments of the value chain—such as computing power, large language models (LLM), or applications—have the most attractive risk-reward characteristics. Then, we determine the allocation ratio for each segment. After that, we study the composition of each ETF in our database and select the most suitable ETF for each sub-segment. This approach allows us to combine our top-down sector allocation advantages with the stock-picking advantages of overseas portfolio managers.

Cathie Wood: From a global investor's perspective, the most striking thing about China's AI is that open-source models like DeepSeek quickly changed the global cost curve. I would love to hear your views: in China, which areas are AI truly landing, making money, and scaling? Is it enterprise software, autonomous driving, robotics, or something else? Which AI applications can win in the Chinese market, and how is the market pricing these winners?

Zhou Jing: China has an extremely large consumer base, and we have always believed that consumer-facing AI applications will have a bright future in China. Unlike the US development in AI healthcare and autonomous driving, China's AI short dramas are remarkable and have already achieved a commercial closed loop. Represented by ByteDance, on the one hand, it provides the Seedance multimodal model to help developers create AI short dramas, and on the other hand, it sells advertising slots within AI short dramas.

Risk Disclosure: The content and opinions in this article are only customer service information and are not intended to provide investors with reference for investment decisions regarding market trends, individual stocks, or funds. Our company makes no guarantee regarding the completeness and accuracy of this information, nor does it guarantee that relevant views or analytical judgments will not change or be updated, and it does not represent the official views of our company or other affiliated institutions. The views, analyses, and forecasts in this article do not constitute investment advice for readers. Individual stocks mentioned are for display only, and descriptions of individual stocks do not constitute investment advice in any form, nor do they represent the holdings or trading activities of any fund under the manager. Our company and employees make no form of risk commitment or return guarantee regarding any investment based on this content, and assume no responsibility for direct or indirect losses arising from the use of the content. Before investing in a fund, please carefully read legal documents such as the fund contract and prospectus, understand the fund product situation and investment scope, and choose a fund product suitable for your own risk preference. Past performance of a fund does not predict its future performance, and fund investment requires caution. MACD golden cross signals have formed, and these stocks are performing well!

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

We need your insight to fill this gap
Leave a comment