저자: CathieWood · 외국인
요약 보기
AI 추론 수요가 '수요 쇼크'를 만들고 있으며 인프라 공급은 수년간 부족할 전망. 기업들의 AI 비용 최적화(효율화)가 수요 충격을 일부 완화할 수 있음. 팔란티어가 효율화 수혜 사례로 언급. 제약 특허절벽으로 M&A가 늘고 있고, 유전자 편집·멀티오믹스 기반 신약(예: 콜레스테롤 치료)이 큰 시장을 열 것으로 봄. 벤처캐피털 붐과 M&A 증가를 생태계 확산 신호로 제시. 인텔은 주가 반등 신호가 있으나 과거 경험상 품귀는 공급 증가 신호로 보며 이익 실현을 고려할 시점이라는 입장. 한국 기업들은 특정 분야에 집중 투자 시 공급을 빠르게 늘린다고 평가.
다듬은 글
So, next session we want to talk about the market. In your recent media interviews, you mentioned that that the inference demand should reshape the industry structure in the future, right? Yes. Yeah, but so far this year, the stock market the market is still um obsessed with chips. Yes. Not potato chips, but GPU, CPU, memory chips, etc.
Yeah. How do you explain this phenomenon like this? Um well, I think one thing we're learning is unlike when Chat GPT first came out. 2022 yes, in late 2022 and then started scaling in 23 and 24. Yes. Um we're not talking about subscriptions at $30 per month [snorts] anymore. In fact, even at Arc uh we're paying uh something like for for our for our heavy users, power users uh we're paying $2,000 per month.
Yes, our chief futurist Brett Winton I think is going through more than that because he's got so many different projects going on at the same time that these uh frontier models are just able to take and then run. Uh so he's in heaven. He doesn't have to train a lot of interns. Uh and so it makes a lot of sense.
$2,000 a month uh that would be $24,000 a year. And that wouldn't even pay for 1/3 of an of an assistant. So uh he could go even higher. That's what people are beginning to understand. The major productivity unlock that is going to make knowledge workers um uh more valuable. They're they're going to be creating uh more products and services uh faster uh and and are leveraging uh the productivity of AI to become even more creative.
Human beings becoming even more creative. Imagining uh that we can do infinitely more than we ever dreamed before. That's what's happening to us. How much is that worth? Well, uh um as more people understand it's worth a lot to a certain kind of company. Probably every company, but certainly our kind of company.
Uh that um that it that the cloud providers uh understand they can't keep up with the capacity, so they have to keep investing. So we're in that wow, this This a demand shock. It's a demand shock. Uh and it's going to take a few years to uh for for supply, meaning hardware, uh to catch up with demand.
Now, as I'm saying that to you, we're already seeing another phenomenon. And that is some companies, especially in the startup world, are beginning uh to run up bills in with the the frontier models that are are much higher than their venture capital partners uh believe they should be spending. And so, what they're saying is, you know, there are more efficient ways for you to do this.
And I believe Anthropic doesn't lead you to the more efficient ways, at least not yet. But OpenAI has found out uh that it can um it can garner the trust of enterprises by saying, "Hey, you could become more efficient if you uh if you do things this way instead of that way." Um or not using the the leading edge of the frontier model, but just step back cuz for that particular task, you don't need the leading edge.
So, I think there's going to be some efficiency moves by enterprises um to solve some of the demand shock, maybe faster than people expect. Oh, it seems that the uh if you want to do the use AI to do the more efficient and more productive way, it seems that the more integration of platforms and techni- tech- uh technologies is needed, right?
Absolutely. Absolutely. Palantir's a great one. Yeah. Palantir uh puts guardrails on in terms of governance and compliance, but also understands the language, the distinct language of each enterprise, and is able to evolve more efficient ways of doing things. Again, this is all part of the productivity move, but that that will help with the conservation and efficiency when it comes to hardware.
Now, you put the AI, robotics, energy storage, region, blockchain, and the multi-omics sequencing into a such a broader disruptive innovation framework, right? You I'll show we today we show we all know we all know that it is not just a tech ready I mean stock market. But technology revolution now. But for the investors, this is can give some ideas of what kind of a signals and the investors should notice to know if the copy capital is starting to rotate from the infrastructure level to the to application level.
Yes. Well, I think um in the case of multi-omics, for example, um pharmaceutical companies have a big problem they need to solve. Uh they they I think the patent cliff uh meaning their their drugs Yeah, they no longer no longer can get so much revenues from the market. Right. They go generic and they lose revenue generating power.
So, that's a a $300 billion hole during the next 5 years in their revenues. So, we're beginning to see M&A, and what are they doing? They are buying into companies who are using AI in drug discovery uh and and development in companion diagnostics uh and and they're aiming to actually cure disease now that we have the knowledge associating associated with sequencing technologies to know what is wrong in each cell of our body.
Now, I say that each cell of our body, how many cells do we have in our bodies? 35 to 40 trillion. So, this is going to be the biggest AI project and the longest lasting one because what do we all want? We all want to live healthy lives, not everybody, but we definitely want to live longer and be healthy.
And I think curing disease the more people understand that this is happening now and that it's going to become even more frequent as AI and sequencing technologies converge with CRISPR gene editing. Uh I think that uh the market will begin because these stocks have been hurt so badly over the last 5 years.
Um but we're seeing pharma now starting to buy some of them. Lilly bought a company called Verve. Verve Verve is a a gene editing company that is going after bad cholesterol and eliminating it. Think about that. Uh and we believe that this could be worth 2.8 trillion dollars in revenues over time. I know that sounds crazy.
Crazy. Uh yes. Yes. Uh it is uh more than 10 times what Lipitor, Pfizer's Lipitor, The drug for high cholesterol for Yes, before it went generic. And so now they're seeing a new way that is you don't have to take a a drug every day. You just cure it. And you know what uh we think that will be worth the one and done?
Uh $160,000. And there are something like 17 million uh patients in the US that uh have a form of ath- atherosclerosis. Uh just that one part of this uh uh population with a bad cholesterol. That would be worth that $2.8 trillion. Exactly. As soon as some of these revenue opportunities present themselves, because companies like Lilly are buying companies making this happen, uh I think uh I think we're going to see a rush into the health care space.
Now, consider the source. I have been saying this for a while. Uh but as we've been saying this, what have we learned? It's safe. Uh the FDA is approving it and is actually expediting the trials and the review process. Uh and insurance companies are willing to pay because they avoid a lot of sick care costs, which are more than 80% of the health care budget here in the United States.
Okay. So, I believe in the M&A activities and not only just in the multi-omics sequencing, but in other Very much so. very good signals. If we see in the future someone in the future more and more M&A activities, maybe this is a signal. Yes, I think it is a signal and I think the other signal here is that the venture capital market is beginning to boom again.
And sure, it started with chips and these frontier models, but venture capital, once they see M&A activity, that's a signal to them to invest more in that area and that stimulates the whole ecosystem. Okay. So, we're very We've been very patient because we've seen all these milestones along the way play out and now we're seeing revenue generation.
It's early, but it's starting and we think this opportunity around cholesterol could be the killer app. As more firms begin to understand this is where it's going, not just rare diseases like sickle cell disease or beta thalassemia, but bad cholesterol, that's a monstrous market. Uh next for end of the this session about the market, maybe the short question maybe forward to you.
If we look back at this year the market so crazy so so far in the United States and here in Taiwan, but however, the disruptive innovation things seems to have some somewhat decouple with such a big variety. Um what do you think this reason behind it and this innovation complex remains under appreciated or under prices for a period of time.
Um well, I think one of the reasons is the the demand shock that I spoke about has been so powerful. Now, I do think we're going to be hearing more about efficiency, too. So, um and and we think this demand shock uh will it will um equilibrate. We'll We'll reach some equilibrium. Um whenever I hear the word shortage, whenever I hear the word shortage,
I begin to say, "All right, I'm going to start taking some profits because every shortage that I have seen in all of my years in the business uh has signaled to the market, "Do more, more, more, more." So, that today what's happening? Well, I it's it's like old is new. Intel Yeah. is getting a boost.
Yes, yes. Right? Uh now, that's the last company I would have expected to get a big boost because it has lost so much share Yes. in the CPU space to uh AMD. It It was uh felt Challenger. Pardon? The challenger. Yes, and and it never got the GPU space right. Uh but, it is starting to work with Tesla.
So, um I think there the the share price rise has signaled, "Okay, let's see if we can work with these other companies that people don't expect to provide a lot of supply." Uh and and you'll get the supply coming, uh I think, in in pretty good um in in and especially when I see uh in Korea the focus.
I've been through other cycles with the Korean companies, and when they decide to put their uh time and money into something very specific, we get so much more of it more quickly than we in the US believe just because of the efficiencies. We are family.
Comments 0