저자: CathieWood · 외국인
요약 보기
ARK은 AI·로보틱스·에너지저장·멀티오믹스·블록체인 5대 플랫폼에 투자 중. Anthropic 매출이 예상을 크게 웃돌며 흑자 전환, AI 학습·추론 비용 급락으로 AI 수요 폭발 주장. 중국 오픈웨이트 모델보다 미국 클라우드 기업이 수익을 가져간다고 진단. 비트코인은 고점 대비 약 50% 하락했으나 저점 형성, 장기 130만~150만 달러 가능성 제시. 기술 혁명은 디플레이션 압력이며 장기 금리는 명목 GDP 성장에 수렴할 것으로 봄. 과거 5년 성과 부진은 금리 상승 탓이 아니라 공급망 문제와 연준 정책 탓이라 주장.
다듬은 글
The revenue generation started immediately and it is so much as I described with Anthropic so much uh higher in terms of revenues generated than we expected and you know this is all we do so this is a big surprise to us if it's a big surprise to us it's a shock to everybody else. Kathy Wood, thank you for joining the business.
Well, thank you for inviting me, Michael. Um, now you're in Australia to Spruak Arcs Venture Fund. Um, what are you investing in right now? Uh, in the venture fund, we are investing actually in venture and all of our funds in five major innovation platforms. Uh, and actually that involve 15 different technologies that we think are going to transform the world.
So, uh, AI, that's the obvious one and it's the biggest catalyst to everything innovation today. And what are some of the names in AI that you're investing in? The names in AI, almost everything that we're investing in. So, we own uh anthropic and open AI and we're direct to cap table, which means we do own the underlying equity, which you will not find in a lot of other venture funds.
It's very hard to access. uh um but because of the kind of research we do and the focused and dedicated effort um uh in terms of researching these new technologies uh I think founders um respect our work and um are happy to give our investors access uh broader access to uh to these new technologies.
Uh so of course in AI proper open AI and anthropic but you know a name like Laya which no one's heard of Laya uh it is aiming to become basically a super intelligent scientist a scientific platform uh for not only research and discovery of uh new therapies uh but also focused on material science. ience as well.
So a general scientific uh platform could not have been done without AI. Almost all names in all of our portfolios, we put them through an AI lens first because AI is going to be ubiquitous and the companies that harness these tools the most aggressively and creatively are going to win. And you mentioned four other platforms related to I guess technology and disruption.
Yes. So robotics and in the venture fund uh one of the premier robotics companies is figure AI uh for humanoid robots. Uh in fact it uh did something very interesting recently. it uh is has basically hired people to clean their own houses uh but wear cameras and other sensors so robots can learn from them.
Uh so uh that's just telling us that the big need in terms of uh the robotics revolution is data uh and uh data is lacking um but there are lots of creative uh attempts to change that. uh then energy storage. Energy storage uh is really effectively battery technology to enable solar and wind to become a part of the ecosystem supporting power the power needs associated with AI most importantly there that's uh one of the biggest shortages um and also in that realm power uh we have focused a lot on nuclear power uh and especially small modular reactors and micro reactors.
Radiant would be a name focused on that. Uh then we have multiomics in the health care space. Um we believe that health care is the most profound application of AI. So curing disease uh and developing new drugs as I mentioned uh with LILA um freome diagnosing disease very early stage one uh is uh and in this case colurectal cancer with a blood test um soon they're well right now they're triing the same for pancreatic cancer which is really
it's a very nasty disease nasty and not diagnosed typically till stage three, mostly stage four. Uh so it definitely um it is a huge unmet need. Uh so think about that with a blood test being able to diagnose cancer because tumors shed into the blood. So huge breakthroughs there. Curing disease is uh something else we're focused on.
uh the crisper gene editing names uh that we don't have one in the private fund. It's in we have two and more in our public funds uh but patents are very dear and the two in our public funds so crisper therapeutics and uh and Intellia uh are uh in very good positions to cure already are curing cickle cell disease and betathalmia.
So those are two rare blood disorders. Uh but the killer app is going to be able to cure bad cholesterol. Uh and that is hopefully in the pipeline. And then the fifth platform is blockchain technology. Uh and there's a huge amount of infrastructure that needs to be built out as the decentralized financial world and the traditional financial world uh converge.
So block damon is a name in in our portfolios on the infrastructure side. Architect is a name uh focused on one of the fastest growing spaces within blockchain technology is called per uh perpetual futures. Um it was invented onchain. We have futures that expire in the public markets. These do not expire.
Uh so lots of different companies surfacing to capitalize on what we believe are going to be massive opportunities. Now there are massive investments going into AI and related infrastructure. Now they're going to require trillions of dollars in investment. Yes. Why are you so confident that the sector will produce the revenue required to make that payoff?
Well, I think we were all looking for proof points and we are getting some already and they're astounding. So, in the case of Anthropic, Anthropic's annualized revenue run rate in December of last year was $9 billion. In July, it was 65 billion. Um, you may know of uh Claude Code and Claude Co-work.
uh these caused a a hockey stick in terms of Anthropic's revenue. uh and not only that there are rumors and we think they are well founded uh that Anthropics gross margins are 80%. Now is that sustainable though an 80% gross margin? That's a softwarelike gross margin. uh but maybe even more important at this early stage given how much in R&D it's spending it is has been profitable two consecutive quarters on an operating basis uh so I mean nobody expected that nobody expected that so I think what we're learning is AI is in making the world a much more efficient place and creating new products and services is incredibly valuable.
I mean, at a media briefing that I was at earlier today, you said that techn technology is inherently deflationary. So, how do AI and other tech companies grow their profits in a deflationary environment? Yes, it is. So, AI training costs are dropping 85% per year. They were starting from a high level, but they're dropping 85% per year.
Think about that. AI inference cost, how much does it cost uh Open AI to answer one of your chat GBT uh questions? Those costs are lower and they're dropping 99.99% per year. So, these are input costs to the frontier model providers that are collapsing. So that's a big big benefit. Uh if you look at the top line, what's happening to do you want to say pricing there started out at $20 a month, right, for enterprises and the businesses were finding so much value.
Uh I know that our chief futurist Brett Winton almost immediately went to $200 a month because he was able to do a number of research projects at the same time. And now with a Gentic AI, he has agents like I think he might have 10 projects going on and I think he might be paying uh we might be paying ARC [clears throat] $4,000 a month.
That's um uh that's $48,000 a year. That's less than half of a research associate. uh it is the the the returns in terms of productivity are enormous. We couldn't imagine this in the old days. But the AI companies are going to have to seriously up prices from there, aren't they, to recapture the the trillions of dollars in investment in physical infrastructure and the depreciation of the chips which just keep getting better that they have to keep updating.
Yes. Uh but what what I think the surprise has been that those it's not that they're raising prices. It is that in the case of Brett, he's just using more of the service and willing to pay for more tokens um because his productivity is exploding. Uh so they won't have to increase prices. He's just going to use more.
That in itself has been a bit of a change though rather than blanket pricing moving to a pay for what you use. Yes. Usage based pricing. Now historically when we've had previous technology revolutions whether it's railroads, electricity, telefan, the internet. Um history has shown that it's not always the developers of the technology or the people who build the infrastructure that end up profiting from it.
and profits often end up going down once we get to the maturity of the technology. Are you not worried that the same might be the case for some of the AI leaders? The the AI revolution is very different. Let's take the um railroad revolution. Um between 1873 and 1893 in the United States, 200 railroads went out of business.
And as you know today there are just a few left. Um so the industry restructured, consolidated and so forth. The railroad industry was built out with a hope and a prayer. Hope they come because they were not generating revenue right away. It took years. It took years for that platform to become really useful.
Uh so there was a lot of hope and a lot of capital deployed and investors just didn't know if it would pay off. Ultimately it did but too late for most of those railroad companies. Um in the case of AI we've already narrowed the market down so much to four big players. You've got anthropic in the US
in the US. in the US and there are from a sovereign AI point of view other AI um companies out there. Uh Mistral in France is a big one in Europe. Uh and so um and of course China threatening to in some senses eat the lunch of the Yes. Well, first I'll answer the um this is not on a wing and a prayer like railroads were.
The revenue generation started immediately and it is so much as I described with anthropic so much uh higher in terms of revenues generated than we expected and you know this is all we do so this is a big surprise to us. If it's a big surprise to us it's a shock to everybody else and you're you're an optimist.
Yes. consider the source. But uh and that's right in terms of the openw weight models. Um what's fascinating about China is they don't have the most sophisticated chips. They do have the power. So they're having they're having to make up with power sources, especially nuclear. They're they're uh in overdrive when it comes to nuclear.
um make up for the lack of sophistication in chips uh with brute force power. Um so what we're finding is from Deep Seek, Moonshot and everything. You know, many companies in the United States are using those models. Who's getting the revenue? It's not Deep Seek or Moonshot. It's the cloud companies who are running those models.
uh they have to be paid, right? Otherwise, they're not going to run the models. Uh so uh and and you know the open weight nature of them sounds free. It sounds free, but it's not free. They are now charging customers. uh and and yet deepseek's revenues pale in comparison to open ai and um and anthropic.
So we're happy the competition is here. There's nothing like competition to um to move a new space along even more quickly than it might otherwise. Um but we we if you look at the Chinese models across the price uh performance uh paro frontier uh open AI uh from a price performance point of view is uh now out competing a lot of the Chinese models.
Moving on to the blockchain and you're known for your bold calls. You recently said that Bitcoin could be worth as much as $1.5 million US by the end of the decade. It peaked last year and it's been on the heading in the other direction since. Why are you so confident that it's going to rally? Yes. So, uh, Bitcoin in particular had it goes through crypto winters they're called.
And, uh, so we're down 50 we were down 50% on Bitcoin. Um, which I I I know people will find this hard to believe. This the volatility is actually diminishing in Bitcoin because in past crypto winters, we were down 75, 85, 95%. Uh so institutions are moving in we do think the price has bottomed uh and if we get the clarity act and by the time this is aired something might be happening in Washington I don't know there are rumblings maybe maybe not uh that should give this new life institutions uh want to get involved with this new technology because it helps lower their costs makes them more efficient And you've got Larry Frink, CEO of Black Rockck, say saying everything's going to be tokenized, so it can trade 24/7 and come become collateral inside the decentralized uh financial world for other activity.
Um, in the case of Bitcoin, uh, it's three revolutions in one. That has not changed. It introduced a new technology for the first time, a currency native to the internet. In the early days of the internet, developers had no idea that financial services and commerce would take place uh on the internet.
Uh and so we finally have that um which is exciting uh because of agentic AI, agentic payments, agentic commerce. If we're going to have machines to machines uh machine-to-achine uh dialogue and uh and commerce and payments, uh we're probably going to need stable coins uh which are, you know, divisible many many many more times uh in order to make that new world happen.
Microtransactions is what there will be. Um the stable coins are part of the crypto ecosystem. They're the on-ramp into decentralized financial services. So very important for Bitcoin, Ether, Salana, Hyperl liquid and so on. Uh so first revolution was technology. Uh it's going to transform financial services.
Second is it's a global monetary system. Uh it's the first private digital that's important. Private also is important. No government oversight. rules-based, probably the most important, mathematically metered to top out at 21 million units. Its inflation rate is about 0.9% in about two years. That will go down to in terms of the supply, it'll go down to 0.4 or 0.45%.
In US dollar terms, the inflation rate's pretty high. Yes. Yes. And compared to That's right. That's right. Uh no, US inflation at 3.7% which is what uh Fed Chairman Wars is watching the PCE deflator is much higher than the inflation rate in the supply of uh bitcoin. The inflation rate in if you look at money growth in the US it's up about 5.4%.
Uh so uh but the inflation rate in the price of Bitcoin is has been in the past quite considerable. Obviously deflated from the record. Yes. Well that means that Bitcoin's purchasing power has gone up. You know as the price has gone up that means its purchasing power has gone up. That's a good thing uh from a wealth generation point of view.
So we do think it's a store of value asset much like gold. It is called digital gold but its correlation to digital gold is over the over the last seven years is 0.1. There's hardly any correlation. So they're both good sources uh from a store of value point of view if you believe that this diminishing uh supply growth in Bitcoin will acrue to its benefit as an asset over time and we do um many people in the developed world don't understand how important something like Bitcoin will be to people in emerging markets whose purchasing power and wealth are destroyed.
destroyed regularly uh by errant fiscal and monetary policies, not just in the developing world at the moment. Oh, well that's true. That's true. You know, I mean, the dollar has lost 99% of its value since the Fed was created in 1913. So, um you know, it's uh endemic to a fiat currency system uh which is controlled by human beings.
uh bitcoin is not controlled by human beings and uh we think long-term will be a hugely important store of value and then the third revolution so technology uh monetary system is that it's the first of its kind in a new asset class I mentioned that even relative to gold the correlation is very low uh I think every asset allocator has to look at this new asset class uh if they want to increase risk risk adjusted returns.
The way to do that is to add assets that are not highly correlated to the other assets in a portfolio and Bitcoin fills that role beautifully. So, is it going to be 1.5 million in uh in 2030? It could be. One thing that has uh usurped some of its role in the short term uh is stable coins. uh stable coins are now being used in the emerging markets uh you know to avoid the the the travailes their own currencies go through and the reason it's working is you know a lot of people live hand-to-mouth in the emerging markets uh and so they cannot afford the volatility associated with bitcoin uh stable coins are much less volatile uh so right now they are usurpering one of the roles we expected Bitcoin to play.
Uh but we think longer term as all emerging markets if this technology revolution is as powerful as we think it's going to be it will increase the wealth in emerging markets as well and they will look for other places to put their money so that it can acrue value and bitcoin uh we think will be an important one of those
and just finally it is fair to say that long-term investors in arcs funds have been on something of a roller coaster ride of returns from 150% gains in the great years to losing twothirds of the value in the the the at the trough. Why should people place faith in your vision of the future? So volatility many people think that's a measure of risk.
It is not. It is a measure of uncertainty and our research is designed to give us a certain level of certainty, a higher level of certainty than the average investor in these new technologies. Now, the period you're pointing to um was very unusual. I'll get into that in a moment. Uh but if you even Morning Star who's looked at our results over the years on a one-year, a three-year, and a 10-year basis, and since inception, I believe as well.
Um we've delivered uh since inception about 14% compound annual rate of return. Uh, Morning Star puts that in our category in the top five to six percent over those various time periods. The one you're mentioning is the five-year time period. Uh, definitely a lot of endpoint sensitivity there. Definitely um anomalous.
I think um many people after that period think thought think that ARC is ARC's performance is inversely related to interest rates. It is not. In 2017, interest rates went up. We had one of our best years. In 2018, interest rates went up even more. It was a down year for the market, we had an up year.
What happened in that COVID and postcoid period. And if there were lessons that I learned from that period was I never would have dreamed that uh the supply chain problems that came out of COVID would would last three years. Um because as a consumate believer in technology, I thought tech would take care of this problem and it certainly did take care of some problems.
It sequenced the corona virus. It developed a test for the corona virus. it uh developed a vaccine for the corona virus. So uh innovation solves problems. It couldn't solve the supply chain problem. Why? Because the Fed was exacerbating it. It took interest rates up 22 23 fold uh in a very short period of time.
It exacerbated the supply chain problems. I would not have have expected that either. So what we ended up in terms of technology back then, we ended up in a just in time world perfected by the Japanese uh just in time manufacturing system. Uh just in time is not what we need in this uncertain world.
We need more just in case. And that I think we have now. So I'd be very surprised to see that mistake uh repeated. If we had known with perfect foresight that that that supply chain issue would last for three years, we would have continued what we started during 2020 when we saw this bull market go crazy to the upside to the point where I said hold your powder dry.
This is not the real world. we would have continued diversifying into uh more cashrich uh cash fortress type companies involved in innovation the mag six and they did very well during that period relative to others we had started doing that also big biotech um uh typically during a bare market we concentrate towards our highest conviction names if we get into another bare market where policy is exacerbating the problem
which we may be heading for the Federal Reserve's meeting this week. um you know if the Fed raises rates um I don't think it will be a mistake and one of the reasons for that is we believe that underlying real GDP growth is accelerating here and if I'm going to compare what could happen during the next few years because of this technology revolution which is accelerating real GDP growth but we believe will lower inflation, maybe even push us into deflation because technology is deflationary.
Good deflation. When you cut prices, you get more units. So, not saying deflation like 0809 at all. Um, if we end up in that environment and real growth and inflation do what we believe, we think short rates will um will approximate nominal GDP growth, which was true during the industrial revolution.
the last time we had a technology revolution of this size. So back then 4.8% for short rates and 3.8% for long rates. So we had an inverted yield curve more than 60% of the time during the industrial revolution. We believe that nominal GDP growth could approach this is nominal real growth could move into the 7 to 8% range.
nominal may be the 6 to 7 1/2% range. Short rates would over time, I'm not saying tomorrow, move towards that uh 6 and 1/2 7 and 12% rate. Long rates, they'd still go up as well, but not as much. um the equity markets hitting new highs even with the recent rise in long-term interest rates and the anticipation of the Fed potentially increasing short rates.
Uh so what that is telling us and we can see it in bond yields is that the market is responding more to real GDP growth. real yields are going up more than inflation as um as it's being discounted by the bond market uh is going up. So um we think that the market is not going to react like it has certainly did in 22.
We're not in a supply chain crisis anymore. Sure, energy is causing issues. Um, but the US is a net energy producer and exports 6 million barrels a day. So, um, it's not going to impact us the way other oil shocks have historically. Kathy Wood, thank you so much for joining the business. Thank you, Michael. Very happy to join you.
Comments 0