Thomas Laffont、Coatue：Anthropic、Citrini报告、AI波动与下一个Mag 7

In 3 or 4 years, if Claude Code can rewrite their entire business, that’s harder for companies to control. It’s one of those companies where, depending on which day you’re picking, you’re going to have a different metric. If you want to outperform the index over a long period of time, you’re going to need exposure to these companies.

Some of them will probably go public in the next 12–24 months. It is unbelievable, the amount of innovation that is now coming from this group of companies. Most companies are now reporting back to their boards the adoption of these tools inside their organizations. They all want to make sure that they’re using the best tools, that they’re being the most AI-forward. They don’t want to be outcompeted by someone who’s using those tools.

And so there was a slide in one of the board meetings that said, “Look, we’re spending X on this tool, and we think it’s way too low. We want it to be much bigger. We expect the spend to at least triple.”

I read the Century paper. I don’t think that screaming “Fire!” in a crowded room is obviously productive or safe, or frankly something you should do. Some people have made that analogy to the report, and I don’t share it.

You and your brother, Philippe, run the firm. I promise not to make a brother joke, but why does he have a French accent and you don’t?

First, before we start, a big thank-you to Mark and Carrie. That was a very lovely introduction. I’m quite flattered. I’m so excited to be back at the Upfront Summit this year. There’s clearly a lot of volatility in the market, and a lot of fun, exciting things happening in AI.

We have someone here who knows a lot about that and studies both the private and the public side of things. So today, we have Thomas Laffont, partner at Coatue. They manage around $70 billion on the private side. They manage around $30 billion. Most recently, I think this is your first interview since leading Anthropic’s $30 billion round. So let’s start there. Did you expect, when you invested in Anthropic, that every Claude release would break the market?

1. Claude Code Drives Adoption

It’s been amazing to watch the evolution of the company, even from when we first started discussing this fundraise that just got announced to when it eventually did get announced, which, in most of these processes, takes about 2–3 months before a company formally announces the fundraising.

What was interesting about this one is that the projections and the scale of the business grew materially between the fundraising being discussed and eventually being announced. I think that speaks to just the incredible adoption of Claude Code in particular, which we can dive into.

No, I don’t think we predicted that Claude Code would take off as quickly as it did. I think it’s indicative of a very powerful trend that’s underlying it, which we can discuss. I’m actually really proud that the inventor of Claude Code, Boris, who’s a friend, worked at Coatue for 2.5 years developing software for us. He’s been on an incredible trajectory.

It’s funny: I was in a board meeting yesterday, and I’m just off the cycle of having done 6 or 7 board meetings in the past few weeks. Most companies are now reporting back to their boards the adoption of these tools inside their organizations.

They all want to make sure that they’re using the best tools, that they’re being the most AI-forward. They don’t want to be outcompeted by someone who’s using those tools. And so there was a slide in one of the board meetings that said, “Look, we’re spending X on this tool, and we think it’s way too low. We want it to be much bigger. We expect the spend to at least triple next year on these tools.”

When you see board decks, they’re such a treasure trove of information and insights, right? When you see the same pattern repeating itself across companies, you know that you’re onto something big. All of them, by the way—whether it’s Cortex, Claude Code, or others—are benefiting from that.

There are some quite viral charts about their growth. So when you invested, what were those metrics like?

It’s one of those companies where you can’t even pin it down. Depending on which day you’re picking, you’re going to have a different metric, right? They publicly released, as part of this announcement, what the revenue was. I think they disclosed something in excess of $13 or $14 billion or something like that. It was definitely materially lower when we started.

The fact is, these companies do live in a bit of a quasi-public or private environment, especially these very late-stage companies. People do tend to know the revenue scales. Stripe publishes an annual letter, which they just did yesterday. They go on CNBC. They disclose a lot of their metrics—not all the metrics the way a public company does—but they did disclose, in Stripe’s case, accelerating year-over-year revenue. They disclosed total payment volume growth.

So even though these companies are private and not in a public market, you do have some disclosure and insights into them.

2. Private Markets Need Access

Since we’re in a room full of LPs and institutions, how do you think, as a pretty famous crossover fund investing on both sides, value and capital are shifting in the private markets over the next 5 years?

Look, my default has been as a public market investor. That’s where we started. We started the fund in December 1999. From when we started to about 2.5 years later, the market was down 80% over that timeframe. So I do have to remind some of my colleagues who weren’t there, maybe even in 2008, that markets can go down that much.

My default view has always been that the public market is the best valuation mechanism. It offers transparency, liquidity, and opportunity of access, which, in a world where we now have Trump Accounts, as was discussed in the State of the Union yesterday, are essentially accounts given to children when they’re born and hopefully can grow over a long period of time, is incredibly important.

I do think giving the broader public access to all of these companies is incredibly important. So I think that’s either going to happen one of 2 ways. It’s either going to happen where companies create incentives to go public, or we’re going to have to create more methods to democratize access to private companies. I think it’ll probably come from both ends.

Regardless, if you think about the innovation of these late-stage private companies, one thing we look at is the Magnificent Seven, which has been a significant driver of returns in the public market over the past few years. It has essentially been flat over the past year or so, and that’s because Microsoft, as an example, has lost almost $1 trillion of value over that timeframe as people are questioning its positioning in AI.

That leads you to think, well, what would the next Magnificent Seven look like, or who would be other candidates to fit into the index of the future? I think the names that all of us in this room would probably think of are SpaceX, OpenAI, Anthropic, Revolut, and Databricks.

I do think it’s a really important class of companies. I do think if you want to outperform the index over a long period of time, you’re going to need exposure to these companies. Some of them will probably go public in the next 12–24 months, so that’ll be one impact of it. But it is unbelievable, the amount of innovation that is now coming from this group of companies.

3. AI Volatility Reprices SaaS

We used to see startups disrupt other startups, and now we’re seeing startups and viral X posts and, I don’t know, the Centrine research paper apparently clobber markets. So with that kind of hot-flash-type volatility, how do you, as an investor, think about management?

Yeah, I read the Centrine paper, obviously. Look, I’m of multiple minds on this. I don’t think that screaming “Fire!” in a crowded room is obviously productive or safe, or frankly something you should do. However, I don’t view the report that way. Some people have made that analogy to the report, and I don’t share it.

I do think bringing up these conversations early is really important. By definition, if everyone thinks we’re in a bubble, then we’re not in a bubble. So I think these points being brought up—preparing investors, preparing companies—is really important.

I’m very happy that, in all of our companies, the sense of awareness about AI is incredibly high. That means that our companies aren’t keeping their heads in the sand. I think the fact that governments, regulators across the world, and big companies are already thinking about where this could be going is actually incredibly healthy.

I know that the volatility, trust me, is difficult on a daily basis, and I live through it every day. But I would much rather have daily volatility and daily questioning than no volatility or no questioning, followed by a massive crash 3 years later.

I think the fact that all these questions are being brought up forces governments, companies, executives, and founders to constantly be worried and aggressive about what AI could do to their business.

I think that's actually probably pretty healthy.

I made a joke when we started that each Claude release is clobbering the markets and erasing hundreds of billions of dollars. Each Claude release is going after different categories of SaaS, and so SaaS has been the pinpoint of the volatility. Do you think SaaS in the public markets is going to stabilize at a different premium? Do you think it'll always have a premium? Where do you think it lands?

Yeah. I think it's a question with a lot of different variables, so I'll try and unpack my view. One of the things I try and explain to companies is that you have to think about the opportunity cost and who's the buyer, right? The public market will continuously be comparing the value of your equity and the return on your equity versus others in the market.

If you looked at SaaS, part of why it was so popular among investors for a long period of time is that SaaS just grew faster than other sectors. You could compound. A lot of SaaS companies were compounding in the mid-20s to low 30s for a long period of time. There were no other companies in the market that could offer that kind of growth, and so obviously that was really attractive to investors.

I think what's happened now is, by and large, SaaS companies have significantly decelerated. I was on the Workday earnings call yesterday, which is an interesting example of a founder stepping back in to help lead this company through its next chapter. It's now growing organic revenues at about 13%.

I think now investors are saying, “Well, you're not growing 30% anymore; you're growing 13%.” If I look at your multiple of earnings and I look at GAAP earnings—which investors are increasingly turning to as the gold standard—you're still trading in the high 20s, maybe 28 to 30 times, in that range.

So investors are now saying, “Well, hold on. I can own a semiconductor company that's probably growing. Avago, as an example, is growing almost 40%, and it's trading at a cheaper multiple of GAAP earnings.” I think it's a combination of decelerating growth and expensive valuation.

One of 2 things is going to have to change. Either companies are going to have to benefit from AI and reaccelerate the top line, or the multiples are just going to start to re-rate to where other companies in the market trade, which is some version of 20-ish times GAAP earnings.

If you listen to the Workday earnings call yesterday, Aneel, the CEO, essentially said his job is to come in and reaccelerate the company through AI. He's a product guy, and I think he could do extremely well at that. We'll see what happens.

That's the dynamic that I see, and we haven't even talked about the threat of AI yet. That's a third bucket. But the threat of AI for these companies isn't actually related to their current valuation or, frankly, even their current business. It's more related to the terminal value: Maybe some of these companies are not benefiting from AI today. They haven't reaccelerated, so it's not impacting the business today. But in 3 or 4 years, if Claude Code can rewrite their entire business, what happens?

That's a much harder question. It's much more sentiment-driven, and that's harder for companies to control, especially in the near term. Markets will flip a little bit over whether a company is well positioned or not, and ultimately its product execution will determine that. But I think a lot of it is the combination of the first 2 factors, now combined with questioning of the terminal value, that's leading to the significant re-rating that we've seen in these companies.

4. AI Changes The Workforce

There are around 400,000 software engineers in the Bay Area. Do you think there are going to be more or fewer of them in the next 5 years?

Yeah. That's the 64-trillion-dollar question of today, right? What I can tell you is that not a single one of the companies we're involved with is saying, “Wow, we're seeing so much more efficiency. We want to cut our engineering staff in half.” What they are saying is, “We hope our engineers are significantly more productive so that we can do way more things, so that we can do features that have never been enabled before.”

You could think of companies like Cursor in R&D or Rippling in SG&A—payroll. What if they move from selling you software, which is what they do today, to selling you work, which is different? In one, you're an HR software company, and in the other, you're saying, “I'm actually selling you HR.”

What does that mean? Well, companies have to hire HR people who have to handle a lot of incoming requests from employees: “Why is my payroll different this month than last month?” “Why weren't my commuter benefits approved this month?” Or, “Why wasn't I reimbursed for this?” There are tons of daily actions that are generated.

What if I could have the system handle most of those for you? Now I'm not selling you software that an HR person will use. I'm selling you the work of an HR professional. That might mean that my current HR professional can be repurposed from something that was not strategic but important operationally into something that's much more strategic.

Maybe we need to redo our review process, or maybe we need to rethink how we recruit our engineers, or whatever. I ultimately believe—I'm not a doomer, right?—and I love the bank teller example. I think it was cited in the report, in case you haven't read it.

In the 1970s, when ATMs started being introduced, there was a famous New York Times article that said, “Look, branch tellers are dead. We're going to see a 70% reduction in branch teller jobs.” Actually, what ended up happening from the '70s pretty much through the early 2000s was an explosion in bank teller jobs.

What happened? The ATM brought down the cost of branches by a lot, which meant companies were able to introduce way more branches. Maybe you had fewer employees per branch, but you had so many more branches that the overall TAM increased, right?

So I think it still remains to be seen what the impact will be. If engineers in the US become so much more productive, maybe you'll have fewer outsourced engineers in India, as an example. There's just a lot of different dynamics at play.

I can tell you for us, we're not looking to cut our investment staff in half. We're hoping that they can do significantly more things, analyze more companies, and just be better at their job. If that's possible, we'll want to hire more of them, not less.

5. Coatue Embraces Autonomous Agents

I had Michael Barton, sector head at the hedge fund, on the podcast a couple of months back, and he was saying—and I'll clip this and it'll go viral—that 85% of his job could be automated and he could get AI agents to automate that work. How are you and Coatue thinking about experimenting with autonomous agents? If you are, and if you have a ton of Mac minis around, how are you thinking about that within your own organization?

We do. We brought in someone recently from Goldman Sachs who's Claude-native and is really pushing us everywhere in the organization to adopt coding-first approaches. So we're definitely spending a lot of time on that. I do think there's an element—especially to big-idea investing, which is something that I spend a lot of my time on and, quite frankly, enjoy the most—that is creative. Ultimately, how machines will do that, we'll see. Are they just assisting the creative process, or are they replacing the creative process?

To me, big-idea investing is both creative and actually reflective of someone's taste at the end of the day. I remember when the iPhone first came out. Some people liked it, and some people thought, “No, it needs a keyboard,” or, “It doesn't support Flash,” or, “It doesn't have 3G.” These were all the things that were pitched against the iPhone in 2007, and obviously, we know how that turned out.

I see it, definitely for myself, as enabling me to express myself in much more interesting, coherent ways. I use all of these tools every single day for different purposes, whether it's communicating an idea, replying to an email, or thinking through a difficult situation. If you don't use these tools for that, I really encourage you to. They're incredible at teasing your brain and evaluating different scenarios. For now, I'm investing a lot of my personal time in how to use these tools, and I've found that they make me better.

6. Big Ideas Expand The Market

I did hear from a couple of your employees that you are the big-idea guy. You were the one who brought in NVIDIA. You got conviction on it. Can you talk about NVIDIA for a second, and then also how big ideas permeate throughout the organization?

Yeah, I love big-idea investing. I think we have a moniker internally, BFI, which you might guess stands for “big fucking idea.” The reason that we keep the swear word in the middle is that when you hear BFI and you hear “big fucking idea,” it's jolting for a little bit, right? It's this, “What? Hold on.” That's what a big idea should do.

I personally have a view that a lot of entrepreneurs, when they pitch you an idea, will come and pitch you a TAM. It's usually big, and it's hundreds of billions or whatever. I have a personal view that I've developed, which is that the size of the TAM is irrelevant. Whenever an entrepreneur will pitch me a TAM, I really don't think about the TAM. I think about 2 things. Whatever number you want to give me for a TAM—100 billion, 7 trillion—it doesn't matter to me. What I do think is, is that TAM going to grow between now and, let's say, the next 5 or 10 years? Pick whatever baseline you want. Is the TAM going to be 2 or 3× larger over that timeframe? So that's number 1.

A canonical example is the taxi TAM. It didn't really matter what the taxi TAM was. What mattered was that it actually ended up growing 5 or 10× because Uber created less friction and grew the entire market. So I care a lot about TAM growth over time. That's point number 1. Then I care a lot about additional TAMs. You had 1 TAM initially; now you've added another TAM. Continuing the Uber example, you've added grocery and you've added food.

To finish that example, what that means to me is that the TAM that Uber had initially wasn't super compelling. What was compelling was the fact that, number 1, the TAM grew significantly because of its product, and they added additional TAMs over the course of their life. To me, the best companies—Apple and the iPhone are another phenomenal example—do that. I was very lucky to be the analyst on the iPhone and Apple, basically starting in 2003 and for almost the next 20 years.

It's hard to imagine, but one of the bear cases as the iPhone was getting started and building momentum was that there just wasn't enough TAM for the handset manufacturers. They already represented 150% of the gross profit of the handset industry, right? In other words, they were making money and all the others were losing. What ended up happening? That TAM grew massively.

In fact, the number 1 thing that we got wrong in our analysis of Apple in the early days is that we had the price of the phone declining 5% in 5 years, because that's kind of what you did as an analyst. You had to put declining ASPs, and in fact, the opposite happened. The price increased. So the TAM for phones increased massively, and then, guess what? They added additional TAMs through services and other things like that.

7. The Firm Needs Risk Discipline

So you and your brother Philippe run the firm. I promised not to make a brother joke, but why does he have a French accent and you don't?

Sometimes people really wonder if we are related, and it's purely a function of age and when we learned English. I was lucky to learn basic English when I was 10, and I think I was just old enough—or young enough—to be able to somewhat mimic a US accent. He's 9 years older. By the time he really started to become fluent in English and was learning English, the vocal cords were just more set. So you can see it both ways.

But I'm still able to do it once in a while if I'm in a bar, maybe with some friends. So there you go.

What is the biggest lesson that you've learned from Philippe?

I think that if you think about our firm, I always say there's 2 key components, and 1 is talked about a lot and the other isn't. The first one is innovation investing, big-idea investing, and trying to find trends early, like NVIDIA and others. But that's really half.

I think the other half is risk management. I mentioned to you that the market was down 80% when we started. I think your formative years as an investment manager will stick with you like a face tattoo over the next decades. We think our ability to manage risk and to constantly be thinking about risk is why we're still around almost 3 decades later.

We don't get everything right, and we've certainly made mistakes over time, but we think the ability to endure and compound is what really defines generational investing firms. We're continuously thinking about different risks. That could be in not making an investment. It could be in seeking liquidity in a secondary or in a public market when maybe it's not the most in-vogue thing to do.

But that focus on risk management—I think he's one of the best in the world at that—and I think it has kept us in business for that period of time.

Amazing. Well, we are out of time. So thank you so much, Thomas.

All right. Thank you.