저자: CathieWood · 외국인
요약 보기
결론: Cathie Wood는 AI·로보틱스·헬스케어·비트코인에 대해 여전히 장기적으로 낙관적이며, 특히 헬스케어 AI 영역을 가장 저평가된 기회로 꼽았다. [AI 발전 속도] - AI 발전 속도가 예상보다 빠르며 ARK도 놀랐다고 밝혔다. 기업들이 경쟁에서 뒤처지지 않으려 AI를 도입하고, 개인도 생산성을 높이려 AI를 쓰는 상향식·하향식 확산이 동시에 일어난다고 설명했다. [자율주행·임바디드 AI] - 로보택시는 2018년 Waymo가 처음 상용화했으나, Tesla가 예상보다 빨리 Waymo를 앞서고 있다고 평가했다. 테슬라 차량이 실주행 데이터를 수집해 학습한다는 점을 강조했다. - 전기차 충전 생태계가 로보택시 확산의 전제 조건이라고 언급했다. [휴머노이드 로봇] - 가정용 휴머노이드 로봇이 약 5년 내 가정에 등장할 것으로 전망했다(일부 구간 3~5년). 생산량 증가로 원가가 낮아져야 가능하다고 조건을 달았다. - 장애물은 데이터 부족이며, 벤처펀드 투자처 Figure AI 등이 가사 노동 데이터를 수집하고 있다고 소개했다. [헬스케어 AI] - 암을 1기 이전에 진단하는 것이 가능해지고 있다고 했다. Freenome의 혈액검사로 대장암을 초기에 진단하는 사례를 제시했다. - 단일세포 시퀀싱, AI, CRISPR 유전자 편집을 결합하면 질병 치료가 가능해진다고 주장했다. 베타 지중해빈혈과 镰형 적혈구 빈혈은 이미 치료 가능하다고 했다. - CRISPR Therapeutics는 나쁜 콜레스테롤을 치료하는 요법을 개발 중이며, 임상에서 혈중 지질 약 50% 감소를 보였다고 언급했다. - 신약 개발 비용은 현재 약 24억 달러, 기간 13년이며, AI와 시퀀싱으로 5년 내 약 6~7억 달러, 8년 이하로 줄어들 것으로 전망했다. 이는 특허 만료 전 수익 증가로 이어진다고 봤다. - 헬스케어 분석가와 기술 분석가의 관점 차이로 헬스케어 AI가 가장 저평가되었다고 평가했다. CRISPR Therapeutics와 Intellia 등을 예로 들었으며, Intellia는 2020년 시총 약 2,000억 달러에서 현재 50억 달러 미만으로 하락했다고 언급했다. [엔비디아] - 2014년 GPU가 자율주행 두뇌가 될 것이라고 보고 엔비디아에 투자했던 사례를 소개했다. 당시 시총은 100억 달러 미만이었다고 회고했다. - 차세대 엔비디아 후보로 헬스케어 분야를 꼽았다. [비트코인] - 최근 약 1년간 하락장이었으며, 비트코인은 약 50% 하락했다고 언급했다. 하락 원인으로 2025년 10월 10일 바이낸스 플래시 크래시, 양자컴퓨팅 위협, AI로 인한 채굴 기업의 데이터센터 전환을 들었다. - 기본 시나리오 약 70만 달러, 강세 시나리오 약 150만 달러를 제시했으며, 2030년경 도달 가능성을 언급했다. 디지털 화폐·21백만 개 한정 발행·새로운 자산군이라는 세 가지 가치를 강조했다. - 스테이블코인은 신흥국 사용자의 달러 수요를 충족하는 중간 단계로 보며, 장기적으로 비트코인으로 이동할 것이라고 봤다. [기타 암호화폐] - 주요 4개 체인으로 비트코인, 이더리움, 솔라나, Hyperliquid를 꼽았다. [AI 인프라·중국] - 한 기업이 데이터센터를 기가와트당 250~300억 달러 비용으로 건설한 뒤 Anthropic에 기가와트당 340억 달러에 매각한 사례를 언급하며 수요가 크다고 설명했다. 이후 Google이 기가와트당 500억 달러에 계약했다고 들었다고 말했다. - 중국 AI 모델이 가격이 1/30 수준이지만 실질적 수혜는 클라우드 제공자에게 돌아간다고 봤다. 안드로이드·iOS 구조에 비유했다. [테슬라·스페이스X·뉴럴링크] - 머스크의 기업들이 융합되고 있으며, 테슬라·스페이스X 등이 수십조 달러 가치를 가질 수 있다고 언급했다. 뉴럴링크는 1조 달러 시총에 도달할 수 있다고 봤다. 다만 명시적 시점은 언급하지 않았다. [리스크·조건] - 로봇 및 AI 확산은 생산 규모와 데이터 확보에 달려 있다. 비트코인은 양자컴퓨팅과 규제 불확실성이 리스크다. - 출연자 Katie Wood는 본인 펀드 투자 관점에서 말하며 투자 조언이 아니라는 면책이 방송 말미에 있다.
다듬은 글
We are capable of curing diseases, and this is already happening. I think in about 5 years we will have humanoid robots at home. We believe we are in the early stages of a revolution in artificial intelligence. Anything that is useful to humanity will be highly valued. Think about it. What I just described is 25 years of stretching a rubber band.
Welcome to another episode of Equity Mates, a podcast where we explore opportunities in the world of investing. If you're joining us for the first time, we have a very exciting episode for you today. My name is Ren, and today we're talking to Katie Wood. She is the co-founder and CEO of ARK Invest. She's an investor known for buying cutting-edge innovative companies, and we'll be talking all things technology, the pace of change, and where the world is headed.
I'm really looking forward to this episode . We're going to get her opinion not only on AI, which is where we'll start the conversation, but she's also very bullish on Bitcoin, so we'll get an update on that . She is also optimistic about Elon Musk. Elon Musk, yes. Robot taxis, humanoid robots, Tesla.
She's certainly not one to shy away from making confident predictions about where the future of technology will take us. Or about the future price of Bitcoin, we asked . Yes. And Cathie Wood rose to prominence in 2020 and 2021 during the COVID period, when the market boomed and so did her funds . And then the funds collapsed in 2022 and 2023 .
We asked her what she learned from the experience, how she manages risk in her portfolios, and what has changed as we enter the next tech boom in 2026. Fascinating interview. ARK has funds that invest in both the listed market and venture funds that invest in the unlisted market. So we get a good cross-section of companies at different stages of their life cycles, but overall, Katie is optimistic.
Yes, Katie is optimistic. Okay, nothing more to say. Let's move on to our interview with Katie Wood. Well, Katie, welcome to Equity Mates. Thank you, Bryce. Glad to be here. We want to start with a theme from ARK's Big Ideas 2026 report: accelerating the pace of development. Yes. People apparently have an intuitive sense that the pace of change is accelerating.
Just look at what has happened with AI in the last couple of years. But can you tell us how fast technology is actually advancing? We were even surprised ourselves, and it's very difficult to surprise us because we conduct deep fundamental research, and seeing how quickly AI is developing is incredible.
We have some great leaders who, you know, are very aggressive. Then we have China, which is really focused on this area, and I think there's nothing that's really pushing us both forward and accelerating the pace. I think the reason it's happening so quickly is because it impacts every industry. So there are a lot of opportunities, but we believe that every company can use AI.
And this leads to a huge increase in productivity. And what happens is that CEOs realize, " Okay, we have to do this because we might get left behind in the competition ." So it's top-down , but it's also bottom-up, because people, especially in entry-level positions or near them, are saying, "Wait, AI might take my job.
So I'd rather use it to become more productive, and I always say, " Make your boss look brilliant." “And I guarantee myself a place in this revolution. So, from top to bottom, from bottom to top , by company. This is the first thing that happens. But then something else happens —this is what's called embodied AI.
And this means that the physical and digital worlds are truly merging. This is also surprising because there is a lot of inertia in certain areas. In transport, for example in autonomous mobility, the pace of change has accelerated significantly. It seems like everything is happening very quickly, but the first commercialization of robotaxis took place in 2018.
It was a company called Waymo in the US. And so, when we did our research, we thought Tesla would get there earlier than last June. Now we see that it moves faster than Waymo. And this helps us to explain or illustrate that everything happened gradually and then suddenly. It didn't happen overnight.
In terms of data collection, Tesla has about 6-7 million robots collecting data. I have two of these robots ." Model Y and Model 3 collect data. This is my first car like this since 2018. It collects data and sends it to Tesla daily so that Tesla can learn our roads based on real driving, not just simulations.
So it was a long time. You know, for me it was in 2018. Actually it was in 2015, so 11 years ago. So, little by little, little by little, people said, "Yeah, yeah, yeah, that's never going to happen." “And then, suddenly, everything at once. And I think this will happen with robotaxis. Perhaps in Australia this will not be felt as much because the electric vehicle charging system needs to be developed.
And it took us a long time. We also had a lot of concerns about the range of electric vehicles. Now robotaxis will be electric . So we needed this charging system, this charging ecosystem, so that they would be ubiquitous and wouldn't break down. So this is one of the manifestations of embodied AI. It seems like it's happening quickly now, but it took many years.
Yeah. Another one is humanoid robots. Again, Tesla, SpaceX will be, and probably will merge, will be at the forefront of this. Would n't we all love to have a humanoid robot at home that would do all our household chores, right? This is a dream. And many people, including Elon, dream about this." They say it will happen in a couple of years.
Uh, it's Elon's time. This will probably happen in 3-5 years. Uh, and when I say that in order for us all to be able to afford a humanoid robot, we need the growth in production to accelerate to bring the cost down. Ah -ah. Uh, I think it will be closer to 5 years before humanoid robots appear in our homes.
This will be exciting. Again , slowly, slowly. What will become an obstacle? Data. Uh, we have a company in our venture fund called Figure AI. And she just gave people cameras and other sensors that they wear while they do housework. And why does she do this? Because there is no data. Yes. Yes. Yes.
Yes. And for AI to really take off, it needs a lot of data. Uh, and so I went to Physical Intelligence in San Francisco. Uh, the CEO doesn't allow any The fund won't participate in his funding rounds unless someone from the investment committee visits the CEO in person, which is what I did. Fine. And what was so interesting was that when I walked in, there were people with physical intelligence, humanoid robots, and so on.
The same . All these people physically performed some actions so that the robot could learn using computer vision . But the data is at a very early stage. And in healthcare, this also happens slowly, gradually and suddenly. I think healthcare is the most profound application of AI. We are now diagnosing cancer at the first stage.
And soon, thanks to the convergence of sequencing and AI technologies, we will be able to diagnose cancer and other diseases before their first stage. And to give you an example, Freenome in our venture fund was founded by one of the fund's leaders, a surgical oncologist, who said, "This is ridiculous.
We need to get more data about cancer so we can diagnose it much earlier." " So he's become an AI expert and his company, Freenome, now diagnoses colorectal cancer in its early stages, but he believes it can only be done through blood tests. Colonoscopy is not required. Too young for colonoscopy. I would be very happy to do without a colonoscopy.
But with blood tests, we can now , and increasingly do so in trials, diagnose even the earliest stages of cancer or precancerous conditions such as polyps, which are also released into the blood if they begin to develop into cancer cells. Think about it. Without AI this would not be possible. Without sequencing technologies it is impossible.
These two factors, when combined to diagnose and treat diseases, allow us to add another factor. Sequencing technologies... do you know how many cells are in each of your bodies? No, unfortunately not. Oh, just guess. We usually ask the question: " Trillions." Yes, trillions. Certainly. This is very good.
Very well informed. The answer is: You are very close. From 35 to 40 trillion dollars. I mean, the cells in each of our bodies. And what do we represent? We are walking data factories because there is now a technology called single-cell sequencing. Think about it. We will be able to scientifically determine what is wrong with our body .
We couldn't do this before. But now, thanks to sequencing, AI, and CRISPR gene editing, we can cure diseases, and it's already happening. Beta thalassemia and sickle cell anemia, two rare blood disorders, are curable. CRISPR Therapeutics is working on this, and we will be able to treat not only rare diseases.
I say “we will be able to treat” in a figurative sense. But we will also be able to treat common diseases. CRISPR Therapeutics has a therapy in development that essentially treats bad cholesterol. This will truly be a killer in its category. And by the way, nobody believes it. Yeah . These stocks are priced as if they are not going anywhere.
This is what is happening in the field of innovation. Many people doubt at first, doubt, doubt, doubt, and then - bam! Yes, we wanted to ask you this question, and this is perhaps a good example to illustrate. When it comes to innovation, a lot of companies claim to be innovative, a lot of technologies show potential, but obviously not every technology makes a breakthrough and, you know, reaches mass adoption.
Your job is to invest in innovation, and hopefully, in real innovation, not in fads. Yes. How do you distinguish between these two concepts? What are the hallmarks of something that will make a breakthrough and change the world? Well, first of all, we do research. So when we see a new technology, we go and study it.
Let me give you an example from the very early days of Arc in 2014. We told Tasha Keeney: "Elon Musk talks about autonomous vehicles in his master plan ." What is an autonomous car? What does this mean? “So Tasha hit the road, and since we do a lot of work with our audience on social media, we share the results of our research.
Not after they're finished, they never end, but as they evolve on social media, because we want to meet those who are innovating in this area, and we do that. When we started the company, I couldn't have imagined that Twitter, which was then aimed at teenagers and celebrities, would become our most important social network for research.
And so, through research, social media, academia , technology analysts, she went everywhere to try to figure it out. And one day she came back and said, "It looks like the brains of the autonomous car are going to be graphics processing units (GPUs)." “It was in 2014. You might find it hard to believe, but Nvidia's stock was plummeting back then.
Its market capitalization was below $10 billion . The fall happened because it was just a company that made chips for PC games. And PCs were being replaced by laptops, so their shares fell at a double-digit rate. And when she came and said this, I said, “Oh my God, nobody knows this. Let's continue our research.
We will conduct research from the top down. We will conduct research from the bottom up. We are working hard on this. And if you're on the right track, this should become a leading position in our portfolios." And in many of our portfolios. And we did it, and the rest is history. We were there . Nobody paid any attention to us.
We said, " GPUs, autonomous driving, AI, robotics." " For many years it was in 2014. The moment when the GPT platform began to be actively discussed in the chat did not come until late 2022. Nobody listened to us. We were there, and you know, we made a lot of money. Some would say it's premature. It was premature, but you always have to look at the other side.
What are we going to do with this profit? What about Palantir when it was selling for $10 or $9? It was the same topic. We just approached it from a different angle. Well, in the spirit of sharing your research, Katie, I'd be remiss if I didn't ask: What's the next Nvidia? Oh, the next Nvidia is ...
Or at least what do you think is happening lately that's less reported and what will happen in 5 years? I think it's the healthcare sector. And the reason for this is that the DNA of a healthcare financial analyst and a technology analyst are like night and day. Technology analysts don't like working with healthcare because it's..." A very slow, regulated, bureaucratic, politicized process.
It's held hostage by insurance companies and governments, and tech analysts want nothing to do with it . Healthcare analysts, on the other hand, strive to act quickly and break stereotypes. They don't want this. No, no, they are regulated. They are politicized, they are in a political environment. So I think it's the most undervalued part of the market when it comes to AI.
And so, I gave you one name—CRISPR Therapeutics. Another name in this field that has the most robust patents is Intellia. Intellia is at an earlier stage of development than CRISPR and has already been destroyed, to give you an example. When we first got our footing, in the late 19s, I don't know if it was 18 or 19 when we made our first purchase.
By 2020, Intellia and CRISPR both had market capitalizations of $200 billion. Today, Intellia is worth less than $5 billion . billion. Wow. And the reason is this gap between health and technology analysts. I've heard some health care analysts say, "Ah, disease management is a terrible model. Why? Because we won't be getting this annual payment, this pill every day, or this injection every month." "That's a ridiculous answer.
Firstly, everything that is useful for humanity will be highly valued. And in this case, the cure for the disease, insurance companies today have no problem paying $2.2 million for its treatment. Of course, yes. Why? Why? Because it significantly saves money on treating diseases. Sickle cell anemia begins at an early age, and for 50 to 60 years there was no cure for these poor people .
They had to go to the hospital 10-20 times a year on an emergency basis for blood transfusions. Of course they want to get cured. And of course, insurance companies want them to get cured because it is very expensive to treat diseases. Transforming disease treatment into healthcare is, I think, one of the greatest opportunities we have.
I would say that 75-80% of all funds allocated to healthcare go towards treating diseases. With the advent of drugs, this proportion will decrease, and we will have more means for prevention, early diagnosis and actual treatment ." So we're very excited about this area. Yes, it is really exciting. I mean , going back to what you were talking about earlier, about cholesterol therapy with CRISPR.
Yes, for example. I just googled it while you were talking. A single gene edit , and in trials it reduced blood lipid levels by about 50%. Yes. This is simply incredible. Simply incredible. And how can we get some of this? What other innovations in healthcare are brewing in one way or another that you're really excited about, but that few people talk about?
Right. We talked about early diagnosis of diseases. We talked about treatment. What about the discovery and development of new drugs? People are very surprised to learn that it costs a large pharmaceutical or biotech company $2.4 billion and takes 13 years to discover and develop one new drug, including trial failures along the way.
So, with the help of AI and sequencing technology, we believe that next year...In 5 years, the cost will go down from 2.4 billion to 600-700 million, and the time will go down from 13 years to 8 years or less. The patents for these drugs are valid for 21-22 years. If you add 5 years of income, that's a huge amount.
A huge sum. So we believe, and this is where active stock management comes into play , that some companies will be successful at this and some will not. They won't act fast enough, and by the way, they'll lose $300 billion in revenue because the patents on their drugs will be cut short and they'll switch to generic prices.
So they need to do this, and we are encouraged by what is happening in healthcare. Thank you for giving me the opportunity to talk about this because, you know, it's a terrible area to invest in in most cases, and because of this gap that I told you about, it's probably the most undervalued and the most undervalued part.
The market when it comes to how much AI will transform it. Wow. Yeah. Yes. So, Katie, there are a number of, I would say, very strong beliefs that you've talked about a lot , looking at your career. We've touched on robotaxis and humanoid robots, but we just wanted to mention a couple more ideas to get your general opinion on how optimistic you still are about these ideas, or if you've perhaps moved on to other ones.
So, let's start with Bitcoin. Yes. Still optimistic? Still optimistic. Over the past almost a year, we have experienced yet another crypto winter. And we're down, at least as far as Bitcoin is concerned, by 50%. In previous crypto winters, Bitcoin fell by 75-95%. Some progress. But the reason for this winter, firstly, is the so- called “ flash crash” in October, October 10th last year, when Binance had a software crash and something else.
Automatic leverage reduction occurred when a hedge fund or DeFi company operated on two exchanges: long on one and short on the other, to perfectly hedge their positions. A software glitch on Binance caused an automatic leverage reduction . As a result, they found themselves at a complete disadvantage .
And I think the damage was between 15 and 30 billion dollars, according to various estimates. Many simply went bankrupt. This was the first. The second is quantum computing . Many believe that Bitcoin is the most attractive target for quantum computing. And the reason is that it is a cryptographic technology and Satoshi's coins are there.
We don't know whether Satoshi is a group or an individual. But this group or person has not moved these coins since their creation. So they are essentially easy prey. They didn't move their wallets or move on to more complex quantum computing. resistance. And of course, there are a lot of lost bitcoins on the network .
So it's a big and attractive target. However, the more we work on quantum technologies, the more years and years it takes. AI research is currently accelerating progress, but not that much. To give you an idea, they are currently working with 100 qubits at a time. This is the maximum. We need hundreds of thousands of qubits to, so to speak, crack the code .
This won't happen for a long time. And the third event is AI. AI eclipsed everything else, and Bitcoin miners took the lead . They were moving from mining Bitcoin to powering data centers , right? So, these three events happened in the last year. I think we've hit rock bottom . We went from 60,000 , just below, to around 80,000.
We still believe that Bitcoin is three revolutions in one. First, he introduced technologies that were native to the Internet and a currency that was native to the Internet, which we did n't have before. Why? Because in the early 90s, when internet developers were programming again, no one knew that financial services and trading would be carried out on the internet.
Therefore, there was no built-in level that would provide this. Now it is. This technological revolution is a global monetary system. First, a digital, private, non- government, rules-based, mission- critical, mathematically measured, 21 million unit limited currency that guarantees long-term purchasing power and wealth creation in a way that fiat currencies cannot.
And finally, the third is a new asset class. This is really the first in a new asset class, and we think it will be the biggest because these are three very important ideas. We write in the press: the base case scenario is about 700,000, the bullish scenario is 1.5. But when to buy? We talked about 2030.
This could very well happen, especially if institutional investors begin to actively implement this technology. And the concept of tokenization is introducing more and more institutional investors to decentralized finance. The only thing that has slightly overshadowed Bitcoin in terms of fundamentals is stablecoins.
And yet, if you're an OG of Bitcoin, you've got it. OG. Yes, yes, yes. Fine. Yes, yes. Fine. By the way, I first heard this when Tasha and I were recording a podcast with Elon Musk. He used the word OG and I looked at Tasha: “ What is that?” So now I'm an expert. So, even Bitcoin veterans are saying, " Okay, we need stablecoins because in emerging markets, people are living paycheck to paycheck.
They are very concerned that their currencies will destroy their purchasing power, destroy their wealth. They can't afford Bitcoin's volatility. Therefore, this is an intermediate step. But , they say, over time these people will realize that Bitcoin is better than the dollar, which is the main basis of stablecoins.
So they will grow their wealth with Bitcoin." So it's interesting . We talked a little about Bitcoin and stablecoins. What about the rest of the crypto ecosystem? Do you have an opinion about Ethereum and other coins? Yes, I have. We believe the big four are Bitcoin, Ethereum, Solana, and Hyperliquid.
And during a bear market, builders actually build , and we watch what they build and how ecosystems adapt to each other and become more specialized. So, Hyperliquid is all about perpetual futures. Solana also offers high-performance financial transactions. Ethereum is smart contracts of all kinds. I think this will be very important for agency commerce, but as we're hearing from companies like Cloudflare, Cloudflare says, "To power agency commerce and agency payments, we're going to need an ecosystem, a blockchain, that can handle 20 to 100 million transactions per second ." So these are microtransactions, where agents communicate with each other and make deals in micro- denominations, you know?
For example, 1/1000, 1/1,000,000 of a dollar, and so on . So Cloudflare CEO Matthew Prince says, "We need 20 to 100 million." Well, we see that Visa , the traditional payments ecosystem , processes 20,000 payments per second. Think about it. He's talking about 20-100 million. Ethereum and Solana have very low throughput .
I am an advisor to the Layer Zero blockchain and they are aiming for 2 million and will increase that number as they grow. That's why we're monitoring the payments space, agency payments , agency commerce, to understand how quickly it will develop and scale, and so on. And I think that's going to be very important, especially for Ethereum, to see if it can handle that kind of volume.
Now he can't . This is when founders get really creative. Well, Katie, we'll take a short break. We'll be back soon. Katie, we wanted to talk about risk management because, you know, we're in a period right now where, depending on which side of this debate you're on, we're either about to see the bubble burst or we're going to see it continue to grow.
And, you know, Arc rose to fame in 2021 with an incredible bull market, and then fell again in 2022 and 2023 during the inflationary sell-off. So, using this experience, what did it teach you about risk management? Yes. This is a very unusual period. I have never seen anything like it. Worse than the tech and telecom crash , by the way, which didn't make any sense because the tech and telecom crash that happened 20-26 years ago was caused by the technology not being ready.
The seeds were sown for everything that is happening now, but the technology was not ready and was too expensive. Fast forward to today , all of this has changed and we are now in a technological revolution. So it was very surprising to see such a big drop, but here's what I learned. Typically, when we enter a bear market, we start focusing on the safest assets.
We have a rating system, the most reliable assets. What do we do during a bull market? As in 2020, we are diversifying our portfolio because we know a bear market may be coming. Correction can occur at any time. We are in a volatile market segment, and volatility is not a measure of risk, which is very important for your audience to understand .
It's a function of uncertainty, and our research allows us to have a higher level of confidence than I would say most investors. However, during a bear market, this strategy usually works very well. During a bear market, we should continue to invest in Mag 6 and other cash-rich companies. They have less money now because they are building , but these are real fortresses of money.
But we had no idea that the supply chain problems would last 3 years and that we would be ravaged by a slowdown in production growth . For example, electric cars have been hit hard, right? There are supply chain issues and this has slowed down their growth significantly. Raising the stakes. Rate increases of 22 or 24 times have never happened before, and this has exacerbated supply chain problems.
We can usually withstand rate increases. But the rapid rise in rates is much more difficult. Although MAG 6 performed well. It was much more difficult to survive this . I don't think we'll see that again. First of all, I hope COVID doesn't happen again. But even if that happens, the reason rates have risen so much is because governments around the world have been doing everything they can at the same time, both fiscally and monetary- wise, and that has caused a lot of problems.
They probably won't do that again. That was a lesson of sorts from the last major bull market. Fast forward to today, and Bryce raised the question of whether the AI data center trade is a bubble or just the beginning. What is your view on this issue, and in particular, how do you position your portfolio?
Do you diversify your portfolio more due to risk or focus more on one thing ? We believe we are in the early stages of an AI revolution. I'm really glad that a lot of people are talking about the hype around AI. Doesn't this sound like a bubble in the tech and telecom sectors? Wo n't this end badly?
Typically, these things end badly because too much capital is chasing an opportunity, which reduces the return on invested capital. We are far from that now. If you look at Anthropic's revenue growth rate (we own shares in the company in our venture fund), you'll see that its revenue has grown from $9 billion in December to $65 billion in six months.
So, Anthropic has grown its revenue in 7 months by the same amount as Salesforce, and I understand that we are at the same level that Salesforce has been in its entire 25-year history. And what we're learning, and this is just a rumor, I think the Financial Times wrote about it , or I can't remember , but we think the rumors are well founded: Anthropic's gross margin is 80%, and the company has been operationally profitable for two quarters in a row.
Of course, she's getting ready to go public, so I'm sure she wants to show off, and that partly explains the situation. But, you know, beyond Anthropic, look what happened with SpaceX's xAI. The company built data centers, and then decided... Well, first of all, SpaceX bought it and decided to reorganize because it was lagging far behind the big companies that were operating under the Frontier model.
After the reorganization, skeptics said, " Yeah, he just bought it because it was losing a lot of money and he wanted to disguise it as a much bigger company." And now they laugh at them. What happened was he built a data center that cost, by various estimates, between $25 billion and $30 billion per gigawatt.
Since it was no longer in use, he sold it, I think to Anthropic. Yes, exactly Anthropic. They sold it for $34 billion a year per gigawatt, given its cost. So you get over 100% return on investment in the first year, and then Google comes along and pays $50 billion per gigawatt. We are at a very early stage.
The demand is huge. The profit is enormous. This is quite a provocative technology, and I've given you a lot of examples in terms of embodied AI that show that this is not just about productivity in companies and organizations. This in itself is a huge opportunity. This is autonomous mobility of all kinds, not just robot taxis, but also drones, trucks, air taxis, and so on .
So we live in a world that is going through a real transformation, and I think many of them are the elders of the industry today . They lived through the tech and telecom bubble and bust in their youth, and are essentially saying, "We've been through this before. We wo n't do this anymore. I'm the head of this company, or something like that .
I've been through this . I understand what they are saying. It was a crazy period." And it feels just as crazy now, or it felt just as crazy now, but it's true. The seeds were sown then. They were not ready to sprout and blossom. They took 25 years to grow. Cloud technologies only appeared in 2006. The first major breakthrough in AI didn't happen until 2012—deep learning.
The second major breakthrough—the Transformers architecture—came in 2017. Large language models. It took almost 20 years for all of this to happen. And now we are ready to begin. Yeah. Think about it. What I just described is 25 years of stretching a rubber band. Yeah. If we think about the biggest risk to this story, like if AI does become a reality and it does happen, and I think then the question becomes who makes money and where, like who wins on cost—that's a big part, and that kind of raises the question of China.
We've all obviously seen those Chinese models that cost 30 or 1/30 times less than some American models. Price, 1/30 of the price, not the cost. Sorry, sorry. Yes, this is a symbolic cost. So how do you assess the risk of becoming a commodity and the risk of cheaper competition from China? Do you know what's interesting ?
If you listen to Alex Karp from Palantir, he has been saying for quite some time : “These are just consumer goods. These large language models are mass- market products. There will be a pond of them. They don't have data, blah-blah-blah." Now he even says: “Okay, maybe not consumer goods. They create junk for AI." You know, even he changes his mind.
We actually think that if you look at what's happening in China compared to the US, you'll find it's a bit similar to iOS and Android. Android is open source . iOS is a closed model, right? Android occupies 80% of the market. Apple gets 80% of the profits, right? We think it's a bit similar because the cost structure for these large language models is pretty much the same no matter where you are.
Is the Chinese government subsidizing them? I don't know, maybe, maybe . But the real winners in this whole situation are the cloud providers, because even in the US we use Chinese models. Not us at Arc, we work in the financial services sector, so we are regulated and can't use them, but many researchers do.
And who receives this income? Not to Chinese providers, but to cloud providers . That's where all the activity happens. So we think we're starting to understand, you know, if we use the analogy with Android and iOS, it's becoming a little bit clearer what's going on. They don't have the most advanced chips.
So they have to compensate for this by simply increasing the power, which they have more of than we do. So both sides have their challenges in terms of cost structure, but what we're seeing in terms of this revenue growth is the market is saying, "This is so valuable. This is so valuable." I'll tell you, our chief futurist, Brett Winton, I think he spends $4,000 a month, not $20 a month, but $4,000 a month.
And you know what? He doesn't want to train an intern or a junior analyst, he can do it himself. He likes to do it himself. He is an engineer by training from MIT, and he enjoys it immensely. He can work on 10 projects at the same time instead of one. Well, Katie, in closing, I will say that this was an incredibly pleasant conversation.
If you look at today's stock markets, and if you had to pick the next company to reach a trillion-dollar valuation, what would you pick? In the stock markets? Well, maybe first stock and then non-stock. Oh my god, a trillion dollar market. I mean, we have a lot of trillion-dollar candidates in our portfolios, both public and private, because the world moves so fast.
Which companies will reach this level first? In the private markets, you know, I will never doubt Elon Musk, and we have Neuralink. Oh yeah. And that's relative to a trillion, that's a very low market cap. It won't be the next company, but if you ask me, will it reach a trillion? I think it could reach a trillion dollars.
So, this will be his fourth company, no, third company, that will reach a trillion-dollar market cap. What's so interesting about it? He realized that very early on, and that's why we get along so well , not that I wanted to be too careful. No major non-public information or anything like that, but from a conceptual standpoint, from the very first days when we recorded the podcast with him in 2019, we were already talking about the convergence of technologies.
He understood that all these companies he was bringing together would become sources of differentiated data, and what's the point of the AI game? Who will win ? This is the company with the deepest industry expertise. He has it and he has very high standards. And the company with the largest volume of high-quality data.
He has this high-quality, differentiated data. Think about it. He has X, he has SpaceX, he has Tesla, he has Boring. Yes, which will also take him to the Moon and Mars. He has some robotics data that he is starting to collect. He has Neuralink, which means he has multi-omics data. It has brain data and AI or neural networks are modeled after the brain.
So he has all this data from all these companies that will take him to places we can only dream of. But this idea of convergence...about nine months ago he said, "Hmm, you know, I think my companies are converging more than I even thought, and faster." And then, of course, he absorbed XAI. And, uh, we hear about Tesla.
Just keep an eye on SpaceX. I think it will be one of the most valuable companies. Again, we are talking about tens of trillions of dollars. Is it true? Yeah. Wow. Wow. Well, it's an exciting world we're entering, we're already in it, but the future is even more exciting. Katie, thank you for joining us and sharing your thoughts.
We value your time. Oh, thank you, Alex and Bryce. It was nice. Thank you, Katie. Thank you. You've been listening to the Equity Mates Media podcast. This podcast is intended for educational and entertainment purposes. Any advice provided is general and does not take into account your personal financial circumstances, needs or objectives.
Before acting on general advice, you should evaluate how it suits your needs and read the relevant product disclosure statement. If you are unsure, please consult a financial professional. The host of this podcast and his guests may have positions at the companies mentioned . Equity Mates Media operates under Australian Financial Services License No. 540697.
Comments 0