저자: CathieWood · 외국인
요약 보기
AI·로보틱스·에너지저장·블록체인·멀티오믹스 등 15개 기술이 수렴하며 5대 혁신 플랫폼이 성장을 가속한다는 주장. 로보택시(테슬라 최대 보유, 마진 개선), CRISPR 유전자 치료, AI 학습·추론 비용 급락에 따른 디플레이션 가능성 제시. 미래 GDP 성장률을 7~8%로 전망하며, 궁극적으로 우주 데이터센터(2030~31년)와 원전 확대가 필요하다고 봄. 비트코인은 디지털 금으로서 장기 가치저장 수단으로 긍정 평가. 단기 밸류에이션 압축 리스크는 인정하나 5년간 연 16% 수익률 목표 유지.
다듬은 글
Welcome to Merryn Talks Money, the podcast in which people who know the markets explain the markets. I am Merryn Somerset Webb. And with me today is Cathie Wood, CEO and CIO of investment management firm Ark Invest. Now Ark Invest, I think many of you will know in disruptive and innovative technologies of all kinds.
So we're going to talk about quite a few of those today. AI, robotics, cryptocurrencies, all this kind of thing. Cathie, absolute pleasure to have you with us today in the London studio. Welcome. Thank you, Merryn. Very, very happy to be here. Right. We have got a lot to cover, you know, big ideas. There are a lot of big ideas out there.
That's why I want to start with the big ideas. So you put out at the beginning of every year a big idea report. But you talk about what what's what's going on, which is a lot. And this year, one of the the main ideas in it was that there are there's this amazing convergence of different types of technology coming together to create a wave of tech innovation, the likes of which we really haven't seen before.
And the consequences are going to be enormous. So can we just talk through to get us started those technologies and the convergence that you're expecting? Sure. So there are 14, I mean, 15 different technologies and they're included in the five major innovation or five innovation platforms that are evolving today.
So AI, the ChatGPT moment really did awaken investors to, you know, something's going on here. So AI technology now and you're right the 15 technologies are converging. Two of the biggest trends coming out of this convergence have to do with AI applications. So embodied AI, the physical and the digital world coming together.
So if we look at autonomous mobility, so robotaxis that that involves three of the major innovation platforms converging. So autonomous vehicles are robots, they will be electric and they will be powered by a AI. So robotics, energy storage. Yes. And so if you think about those, those are three different S curves in the world of innovation.
Slowly, slowly, then all at once, and they're beginning to feed each other. Tesla launched its Robotaxi service in Austin in June, and we're seeing now a quickening of the pace. Waymo is now moving more quickly into different cities, so there's a competitive dynamic that is is going to accelerate the trend towards robot taxis in the United States.
And because robotaxis are safer, we've just crossed the human threshold of safety and will continue to be. Yeah, that's true. That's true. So that's one. And we think that's the biggest application of AI from a revenue generation point of view other than the productivity tools that we're using right now, which is.
Yes, absolutely. And and we're paying for them and we're going to pay more and more because they are that useful. The second major trend is in health care. It's the most profound application of AI, and it includes the convergence of sequencing technologies, AI and CRISPR gene editing. So sequencing, we have 35 to 40 trillion cells in our body generating data, and AI is a data project.
Proprietary data is key to success. So sequencing technologies, A.I. now that we can find mutations which are programming errors in our body, we have 6 billion bits of code. If you think about 3 billion base pairs of DNA pairs, so 6 billion bits of code, we can now see when one of them has gone awry, when there's been a mutation.
And now and we can see it because of A.I., we couldn't have done any of this five, ten, 15 years ago. Now that we can see it, we can use CRISPR gene editing to reprogram the genome. And we're already seeing cures for beta thalassemia and sickle cell disease. Those are and we're seeing revenue generation CRISPR therapeutics Is the company and Vertex, its partner, generating revenues and curing disease.
Those are rare diseases. We think this tech, these technologies will impact common diseases in the not too distant future. And CRISPR is one of your big holdings in the innovation ETF. It is. It's I think it's our number two holding. It's earned, right? Yes, yes, yes, yes. So and and one of the reasons it's up there is in its pipeline is a cure.
Now it's in trials for bad cholesterol. Think about that. Cardiovascular disease is the biggest killer in the world. And what we're going to do here is shift sick care dollars. So just treating chronic disease. Shifting sick care dollars or pounds or euros into really therapies that cure or prevent or into diagnostic tools that diagnose cancer in stage one or before stage one with blood tests.
So we think health care is going to transform completely here. The idea there being that people will end up with a much not necessarily a much longer life, but a longer health span, a longer period of healthy living. Yes. And we are we are writing a white paper about longevity. But you're absolutely right, Merryn.
The first order here is to extend healthy life, you know, increase the quality of life while we are alive. And if that happens, that is transformative for public finances apart from anything else. So at the end of end of life, that last ten years is the most expensive bit for any health system and hence for any government.
You can take that out. Suddenly, possibly the fiscal situations of the Western economies don't look quite so awful. Absolutely. And the U.S. is the worst that way because we spend nearly 20% of GDP on health care. I think in the U.K., it's it's less than 10%. I think around 11 oh oh, you know, it's huge.
Okay. So it's it's huge. And in the U.S., there's a lot because of our litigious ness there. There's a lot of unnecessary spending. And so we will get those answers, you know? Yes. Spend and no Don't spend. Okay. So we have this convergence of these extraordinary technologies, which gives us this acceleration in everything acceleration and productivity acceleration and GDP growth.
I know you're very optimistic about growth across the board as a result of all these innovations. Yes. If you look in big ideas, you'll see there's a chart, a timeline, and it's in the section called the Great Acceleration. So our chief futurist, Brett Winton, named his section the great acceleration thinking.
People would understand that, no, we're not going into the Great Depression, we're going into the great acceleration. I have to explain that. I've told him because I didn't get it. So anyway, so the great acceleration, if you look, you'll find a GDP chart and it's broke that the timeline is broken down by tech breakthroughs, you know, tech revolutions.
So if you look in the years from 1500 to 1900, nothing much changed. And as best as as Brett could ascertain with, you know, consulting, academia and so forth, global real GDP growth in that time period was about 0.6%. So very slow. And then we have really the industrial revolution, first the railroads and then building on the railroads, we had telephone electricity and the internal combustion engine.
And we had then a step function increase in growth to 3% real GDP growth globally for 125 years on average, with China joining in at the end. And now we believe with these five innovation platforms. So there were three major ones on top of railroads. There are five major ones on top of the Internet.
The Internet was sort of the prelude, much like the railroads were. So five innovation platforms which involve Ford, I mean, 15 different technologies, which we think are going to take growth into the 7 to 8% range. Now, many people think that sounds ridiculous, but that's not the last technology revolution, which we we don't think.
Was as big as this one's going to be. We had a five fold increase. We're saying maybe a two and a half fold increase. So we think we're being conservative. But the other thing to note is these technologies are deflationary. So if we were to enter this 7 to 8% growth range, it probably would be accompanied by deflation.
And many people just don't believe that, but that the deflation being caused by AI alone. AI training costs are dropping 75% per year. AI inference cost so what it costs to deliver an answer to your query. Those are dropping 85 to if you believe DeepSeek 98% per year. This is this is technology is called learning curves.
So the more units produced in this case tokens in in the world of AI, the lower costs go. It's the learning curves associated with technologies. So when a technology is mature, like the internal combustion engine autos, there isn't much unit growth, so the costs don't come down any more. Right now, the unit growth in AI is exploding and therefore the costs and sort of the the undertow, the deflationary undertones we think are picking up here.
Now, of course, we have oil on the other side of this. We think this is temporary. Better be temporary, right? Yes. And so so it's this deflation is being somewhat obfuscated, but there is a measure. It's it's in the blockchain world. It's called truflation. It measures. 10,000 items, 24/7. I keep an eye on that, and it's often very, very different to the official inflation numbers.
Yes, it was higher coming out of COVID. It got to 11 12%, whereas the CPI in the U.S. got to 9%. Now it's one and a half percent. Whereas the inflation measures, especially with oil and tariffs, they're kind of sticky in this 2 to 3% range. Truflation, we think, is getting it right. It did drop below 1% and now we've had a bit of a bounce because of energy prices.
But even even if the oil price rise is temporary, even if if the Middle East is sorted out and everything's fine or price falls back down again, still, one of the main limitations to a lot of this is energy right now. And we know how energy intensive anything to do with AI is. So that still remains one of the limitations.
Absolutely. And one of the limitations, of course, in the United States and elsewhere in the world is getting connected to the grid. It takes it takes five years. So Elon has gotten very creative and brought in generators and batteries. And so he was able to get his Memphis data centers xAI's up in I think it was it was fewer than eight months and most people thought it would take him years.
And now he's thinking about data centers in space. And we're we're doing that work. In fact, we have published our Space X model, and anyone can go to our site or to GitHub and and find it. And if they disagree with our with our assumptions, they can change them and see a different number which Space X is going to go public.
So this is going to become more important. We are adding in data centers that will be additive to the model and we do think it's possible. And again, if you have your data centers in space, then they simply use solar energy, Solar and everything is very straightforward, very straightforward. And this brings into play reusable rockets that just go up and down and hugely.
And that's been the real breakthrough here. Without reusable rockets, it would be prohibitively expensive to put satellites into space. Space X has about two thirds of the satellites in space right now, and other rocket providers are attempting the reusable rocket space. But space X is ten years ahead.
Mm hmm. Okay. So if if we get to the point where most of the data centers are in space, then that takes away the energy limitation. It takes away the the land limitation, the grid limitation. All of this, the bureaucracy, there's no bureaucracy takes away the regulatory. I mean, right now there is no bureaucracy or regulatory environment in space, but the more full space becomes, eventually there will have to be a regulatory regime around it, right?
Yes. Place your own space. We don't know, do we? Exactly. That's that's going to be a comfort. I mean, it almost seems that this is a technology race, and that is why China is now racing to the moon and and experimenting with reusable rockets as well. Okay. So how far away do you think that is? By the way, if we move on.
Yes. Orbital data centers. Yes. By our calculations of 2030, 2031. So we still will be dealing with the bureaucracy and the regulation here. So, yes, it's that that time period is is where we consider the energy drain. The shift comes so fast, doesn't it? Last time we talked, Cathie, when we talked about the energy limitations on on air, we talked about nuclear energy and about how that was the solution.
But we've already pushed that to one side because it doesn't matter anymore because we're going to space. Well, what's interesting is the Trump administration. Recognizes that economic activity is energy transformed. Now, what what has happened in the last 15, 20 years was a movement towards more efficient energy use.
So that was good and that was very good. And we've become much more efficient. But now we're we're. Because we believe GDP growth is going to accelerate. We need more, more energy. And so nuclear absolutely is an answer. And I am shocked at the regulatory the speed of regulatory change in the United States.
Now, what what's interesting, this is another race. China has about 30 nuclear plants, big ones being built today. The U.S. has zero. And so this is shocking to us who those of us who believe that real GDP growth is going to accelerate and we need the energy to do so. So we're now seeing some approvals, but we're also seeing breakthroughs in small modular reactors.
And of course, OpenAI with Arklow had really started that movement in the public equity markets. There is a lot more going on in the private equity markets. It's interesting, isn't it, how we've as you said, we've been through this phase where the the driver behind everything has been to try and use less energy.
And that is not the way that wealth lies, is it? That's right. I do think going through that period was important because there was a lot of waste and we have gotten a lot better. But now we're going to have to step up energy production of all kinds in order to accommodate this new world. Let's talk about the winners and the losers here, because it sounds like everything is becoming a little bit more clear.
We have more of an idea of what the future is going to look like than maybe we did even a year or two ago as these things converge. But when you look at stock markets and listed companies, then you can now begin to see who might win and who might lose. Yes. Well, I have to start with Tesla. That's our largest position, at least in our flagship fund and our funds here.
I mean, Europe is really, I think, approaching our and maybe it's because of your podcast, but approaching our strategy in the correct way as a satellite strategy. So our largest position in the flagship strategy and in our air and robotics strategy, I mean, there's there's a bit of a competition there, but it's it's it's up there is Tesla because that is the largest AI project on Earth.
And it embodies this idea of the convergence among technologies. And in fact, Elon very recently kind of giving us a clue that he was going to combine Space X and xAI. He said, you know, they're more of my companies are going to converge more than I ever thought. And we thought that was a victory for us because we've been using this word convergence for a while and, and, and we've been describing Tesla in this way.
And ultimately it could be the Tesla X, A.I. and Space X do get together. But we don't think yet, because I do think Elon wants to reward those of us who have been patient shareholders with the Robotaxi take off. And and now that analysts are being forced to look at this new world, they are seeing Tesla's model shift from 16% gross margins in electric vehicles to ultimately with robotaxis just that opportunity segmented, that's 80 to 90% gross margins.
It's still a very competitive market, isn't it? It well, it seems so. There are a lot of companies trying to enter. We do think this is a winner take most market because the company to get passengers from point A to point B, the safest, the fastest and the least extensively. And that last one is where Tesla will ultimately win will win the lion's share.
It's a little bit like Uber and Lyft. There's a network effect and and and so we think Tesla will win the lion's share. So if we're right, then globally the revenue generation from autonomous taxis is going to get globally including China is going to scale to north of ten. Trillion within the next 5 to 10 years.
And we're talking about only a billion dollars right now. And again, the lion's share will go to a few companies. The platform providers will get most of the profits. Tesla's a platform provider. Waymo is as well. But Tesla, because it's vertically integrated from a manufacturing point of view. We believe that by 2030 its costs will be 50% lower than waymo's costs.
Given what we know now and so it will be able to pass those lower costs on and provide a less expensive ride. And then, of course, China's where a lot of other companies are evolving. So Baidu's Apollo pony AI we ride. And then of course, Uber is is partnering with a lot of auto manufacturers here in the U.K.
You have wave. We had the pleasure yesterday of taking a ride in in wave. And I have to tell you, I was truly impressed, very impressed. And, you know, we've been we cut our teeth on Cyber Cab and Waymo and so forth. So here in London, the streets being as complicated as they are and pedestrians all over the place, I was truly impressed.
And it is partnering with all the OEMs because there's a sense of urgency now. They all pulled away during COVID from these investments in the future and now they realize they should not have done that. So now they're scrambling and wave seems to be in the catbird seat. Okay. Interesting. And winners across the OpenAI space?
You know, we've moved we've moved slightly on from the days of the Mag7, haven't we? And we saw all the disruption in the software space, etc.. It's becoming more obvious that not everyone's going to win it. And this has been a wake up call. So we noted from our big ideas that if you if you split the tech stack into its three major components, so infrastructure that data centers chips and so forth, platform as a service which is really customized software often generated internally.
But now talent here is turbocharging the customisation of company data and company ontology. You know, every company has its own language so that's in the middle. And then the third was applications. So infrastructure platform as a service applications. And we saw all of them growing very nicely over the past five years and that.
But when you do the second derivative, we learned that wait a minute, platform as a service is capturing much more of the incremental growth than applications. What's going on here? It was that it was the SAS co-pilots, it is Palantir and all of these customization efforts that are possible with AI now is taking share from SAS, which is one size fits all.
And so that was an important clue to us. So we did pull away in recognition, not understanding that there would be SAS copyleft, but just that there would be more opportunity in the platform and the infrastructure part of the tech stack. Okay. Some of that was a surprise to other people. And when you look ahead now, what is the thing that you're most uncertain about that you look ahead because we've talked a lot about what you can see and what you can expect.
What is it that you look at? And then I just don't know how that's going to unfold. What are the big unknowns in these sectors? Well, I think the most important question for us thank you is unit growth. What what would what would impede the unit growth of these technologies? And we learned an important lesson in COVID.
We had no idea that there would be that this supply shock would cause supply chain issues for three years. And that did slow down a lot of our technologies. So we learned in an important lesson there, anything whether it's geopolitical risk. And sure, we have we have this Iran war. And the question coming out of it is, oh, my gosh, is this another supply shock that's going to cause another few years?
We don't think so, but it is the right question. One of the reasons we don't think so is what did we learn from COVID? We learned that meant just in time so jit manufacturing had gone too far. We. Need buffers. So we need just in case. And I think that's what we have done during the last few years, is we've started building in those buffers.
But that's the most important unknown to us. And I suppose the other thing is the talking about supply crunches is the possibility of supply problems with rare earth metals and various commodities, not just oil. Right? Yes. About fossil fuels. Yes. And I do think that President Trump, in his negotiations with President Xi Jinping, is focused on this quid pro quo.
We have the oil that they need, especially if if Iran and others energy is shut down in some way. I think China gets 80% of whatever flows through the Straits of Hormuz. Right. So we can be that fallback. They need to give their rare earths and we in the United States need to deregulate. Why don't we have rare earths?
Because in 1980, our regulators decided that any waste from rare earths should be treated like nuclear waste. So everyone stopped. That is changing. Yes. Changing. Yeah. Yes. Well, and that brings me to the other limitation that you might find around pretty much everything, which, of course, is regulation.
Yes. And, you know, increasingly, the whole conversation around, I guess, quite emotional, doesn't it? Yes. And if you look on social media, you'll see these stories playing out with these are terrible and games at one of them, of course, being mass unemployment, etc.. So does it. Is there a concern that the regulatory build up around the use of AI might turn into a headwind?
Well, in the United States, there was a regulatory buildup. There was an executive order which President Trump rescinded first thing. And now we're seeing in New York ridiculous legislation being proposed saying no corporation can use large language models. It's like, that's crazy. If New York wants to continue the exodus from New York to my state of Florida or Texas or other, more business friendly, capital friendly states, that's the way to do that.
That's the way to do it. And I think, you know, this mass unemployment that is it is the unions who are behind this legislation. Mass unemployment, I think, is exaggerated. There will be displacement. And in fact, already we're seeing in the cohort 16 to 24 years old that unemployment rate is going up.
It's at nine and a half percent in the U.S. versus 4.4 for the overall unemployment rate. So those are those entry level jobs. Those are disappearing and giving way to what's happening in the U.S., though, interestingly, is despite what we just heard from metro platforms laying off 20% of its workforce to afford the capital spending.
We are seeing more and more people holding their jobs. They're not they're not layoffs. And so we're already starting to see entry level jobs disappearing. But those who are at work are taking up that challenge and saying, I'll run with this. I know in our firm everyone's really excited about what we can do.
And in fact, everyone's asking for bigger budgets so they can do more. So I think the saying goes that, you know, those who lose their jobs will it won't be because of AI. It will because be because of someone else using AI and becoming much more productive. So that's a clarion call for everyone to just, you know, really delve into this opportunity.
It is a huge opportunity for everyone to to increase productivity. And we you know, the history of technology is net job creation long term. Sure. Short term, some displaced with a lost generation along the way. Yes. Well, but they have an opportunity. You know, what's a wonderful about this opportunity or a technology opportunity?
This isn't going to coders and developers. Sure. The developers the most the most productive ones are benefiting. It's going to almost anyone. If you speak a natural language, you can vibe what's called Vibe code. Vibe code with Chatty Beauty or Claude and start your own business. Solve a problem that you see out there.
And I'm counseling young people especially to do that because. Maybe it takes six months on average to find a job if you're unemployed. In the U.S., the median is three months. But if you look for the job, but in the meantime, try and start your own business. Using a AI, you're going to become much more attractive to employers out there, anyone with AI initiative.
And again, this is all natural language. So it's about initiative, not skill. Exactly. And creativity and imagination. Yeah. Speaking of which, the big ideas that everyone's talking about. Space X Anthropic, Open AI, are you still expecting those this year or do you think maybe in the Middle East, etc.
will have pushed it out a bit? I think the reason we'll see more this year than otherwise might have been the case is because of the midterm elections. If the Republicans lose the House, it could be that we'll be in a more stringent regulatory environment. Yes, they might come forward. Right, right, right.
Interesting. Let me then connect your question. Valuations. Now, one of the things that is often talked about, of course, is the high valuations. And of course, lots of companies that you invest in. And I suspect that your view, obviously, is that all companies will grow into their valuations. The other view, of course, is that they will collapse into their valuation, into the correct valuation.
Yes. It's a very good contrast to help me explain what we're doing. Yes, our companies do sell at premium multiples. Why? Because we want them to sacrifice short term profitability and invest aggressively to capitalize on these really massive opportunities. So while they do have premium multiples right now, our our assumption is that that valuation will compress to a market market multiple during the next five years.
And our analysts and portfolio managers must believe that the revenue growth and margin expansion is going to overwhelm that headwind. And you're right, grow into the valuations is another way of saying it. But to visualize it mathematically, we're saying no, that is a big headwind for the next five years.
And our analysts and portfolio managers must believe that revenue growth margin expansion will overwhelm the valuation compression and deliver a minimum hurdle rate of return of 16% at a compound annual rate during the next five years. And if that happens, everything really is fine. Yes. Yeah. Yes. Okay.
Let me ask you one, one last question, because our listeners will not forgive me if I do not have Bitcoin. The one thing we haven't talked about is the monetary system and possible changes within. Are you still fully confident in your view? Yeah. Yes. On bitcoin, absolutely. Bitcoin we believe is going to be the most important crypto asset for three reasons.
It's it's really three big ideas in one. The first is it's a new technology. We have now currency native to the internet, which is going to take out a lot of friction from the buying process, especially as we move to a gigantic AI where, you know, machines will be will be buying from other machines.
Right. So that's the first thing. The second is it is a global monetary system, and we're beginning to see it react to the events in the Middle East. And it has stabilized after something called a flash crash from last October, which caused auto deleveraging, which caused a lot of people offsides. We think that's all washed through.
We think Iran now is taking center stage and that Bitcoin will serve its role as a store of value. We just hit 20 million Bitcoin minted ever and we're going only to 21 million gold with the increase in price. I guarantee you miners have accelerated their search for more gold and will be producing more.
Bitcoin is mathematically metered and its growth rate has already slipped below that of gold. And we also think with the huge intergenerational wealth transfer taking place, that Bitcoin will be the store of value of choice. And then the third big idea is Bitcoin, and we wrote this paper in 2016. I'm very proud of our team Bitcoin, the first of its kind in a new asset class.
If you look at bitcoin, its correlation relative to other assets is very low, even to gold. Many, we call it digital gold. But if you look at the correlation since 2019, really since institutions started thinking about this, the correlation between gold and Bitcoin is 0.14, hardly any correlation. And if you look at a matrix, I put it in my New Year's letter in January, we have a matrix.
You'll see the correlation with other assets is very low. So what does this mean? Asset allocators are looking for low correlation assets because if they in the satellite strategy that we offer, if they put a lower correlation asset in their mix, their risk adjusted returns go up over time. So maybe everyone should just hold Bitcoin in gold and be done with it.
It's a store of value. Yes. Kathy, thank you very much for joining us today. Really appreciate it. Thank you very much.
Comments 0