And if you and I want to win championships and build a championship basketball team, we should not care where in the world the basketball player comes from. We just need to get the best players on our team to win the championship. And we ought to take the same approach to AI and technology.
Hey Bill, great to see you.
I'm doing great, I'm doing great. You know, I'm heading back to Boston this weekend for my 25th HBS reunion.
25? It's crazy. I mean, you're old, you're way older than me, but I can't believe it's 25 years. And I'm doing this talk on what's happening in AI these days. And there's a lot of questions about and comparisons because remember we were there during the 99-2000 kind of boom and bust. And so a lot of my classmates are wondering whether or not AI is kind of like that again. And so I was going back and pulling together these slides and I have to say a few things just shocked me frankly. I forgot how much has changed. One was I was looking and doing some analysis on Amazon. You know, I was day trading Amazon out of the back of the classroom back then and you were out here working on the IPO. So I know you were paying attention to what the share price was in 98, 99, and 2000. And a couple interesting points, right? Amazon, if you recall Bill, and I know you do, peaked at 243 bucks a share in 1998, but at the start of 2000 is at 150 bucks a share. Okay? And Henry Blodget famously makes the call. He calls $400 a share and he literally kind of top ticks it from the year 2000 and the thing plummets and goes down to about 26 bucks a share, right? He gets ridiculed and all this stuff. But I wonder, by the way, it went public at 17 and broke issue.
So it traded under issue for about two months, right? And I'm sure you didn't get any calls from the company about what the f was going on. But here's the crazy thing, right? Remember he gets ridiculed for making that call. So I wanted to know and I went and asked our ChatGPT friend to help me with some split-adjusted math on this. So split-adjusted from the high in 2000, which was this 150, that's equivalent to about 47 cents split-adjusted today. So it's up about 440x from where it was in 2000. And on a split-adjusted basis from the low in 2000, it's up about 1,800x, right? I mean, those are shocking numbers. And so that was one of the things that did it. So I said, well, what's happened to the NASDAQ since we graduated? Well, from the peak in 2000, the NASDAQ is still up 5x, and from the trough in 2000, it's up 10x. And so I think, you know, what's the punchline? The punchline is, you know, we were a bunch of dreamy-eyed big thinkers. We thought we knew everything about the internet. We knew it was going to change the world forever. And it turns out we overestimated what it was going to do in the short term, right? Over the next two to three years, adoption was slower. We had the terrorist attacks in 2001. We had an economic recession. So things definitely grew slower. But the biggest problem was there just weren't that many people connected to a high-speed internet. So all the things we dreamed of occurring were just inefficient to happen at that time. But what's probably even more surprising, Bill, is how dramatically we underestimated the long term. Over the next 20 to 25 years, it blew away all of our estimates and forecasts in terms of how big these companies would be. And so now we sit here at kind of the dawn of the age of AI. And I think people are asking a lot of these same questions. But that was probably the biggest punchline on my 25-year reflection.
All right. Well, now if they listen to the podcast, they're going to miss your talk. Maybe, maybe, you know, talking about this pace of AI, you and I were sending some things back and forth just about some strategic shifts going on in the world. Talk us through that.
Well, obviously, I mean, the pace continues like crazy. Every time, you know, between just the podcast episodes we do, which are about two to three weeks, there's new news always and constantly. And it's hard to keep up with it. But I do try and pay attention to the stuff that spikes in my brain as being strategic and something you should really pay attention to. And one thing that I've seen in the past week, so this is very recent, is what I might call data walls. And so everyone's recognizing the value of AI. Everybody wants to have AI work against their data, against, you know, and everybody wants to wow the consumer or the customer of their product. And so all of a sudden though, we're seeing things pop up where people are trying to wall off data. And one example would be Reddit just sued Anthropic this morning despite having a deal. So it'd be interesting to unpack what's in there. I think there's a tiff between Windsurf and Anthropic where Anthropic cut them off to all their models, you know, when Windsurf was bought by OpenAI. Not that surprising, but just seeing these walls pop up. I think a more interesting one in the enterprise case is Salesforce changed their terms of service in a way that includes not only the CRM data but Slack data, which is a company they bought. And they're putting an MCP connector on top of it which allows AI to query it, but they're saying that you can't train on the data that's in there.
And hold on a second. So Altimeter, Slack data, I can't train on my own data? That's what the change in the terms of service say? You can query the MCP?
Yeah, I think maybe we'll start to talk about enterprise applications as either being open data or closed data and people are going to need to declare that. But my guess is that if you're a competitor to Salesforce, you're immediately going to declare yourself open data and try and steal as many customers as possible. I can't fathom in my own brain how upset I would be if I were paying seven, eight-figure license to Salesforce and they told me I couldn't train on my own data, which is basically everything about my customers and everything that I would want to analyze. Now, I'm sure they're going to train on it and give you their AI agentic view of it, which is why they're doing this. But battle lines are being drawn, you know, going back to the old song. And one of the things you didn't mention, Bill, I just saw Kevin Weil, our friend, tweeted from OpenAI that deep research is now searching across GitHub, Google Docs, Gmail, Outlook, HubSpot, Dropbox. Now, I think a lot of that's using MCP, but I do agree with you that this all seems to be happening faster rather than slower. You know, it reminds me, you and I back in the day sitting in a Zillow board meeting and talking about people building dependencies on Google, and you were very much aligned with, you can't build a brand in the underbelly of Google because over time they will have to take that territory back, right? And I remember TripAdvisor peaked at $20 billion in value building a search engine, a recommendation engine in the underbelly of Google. And then when Google decided it was time to do recommendations on its own, TripAdvisor's value went from 20 billion to 1 billion. And that happened over a period of many, many years, Bill. And what I see happening here is kind of in part because AI is going so much faster than the internet did, right? I saw some analysis from my team this week, we'll post that, OpenAI reached 400 billion annual searches 8 years faster than Google, right? So they're doing over a billion searches a day now, right? So all of this is happening in hyperspeed. And so the strategic plays by all of these companies to vertically integrate, to shut down access to data, because they all know that they need to monetize by kind of offering that full stack. And so that to me is something for these companies. When you look at Windsurf, they're going to have to build their own models. When you look at Cursor, they're building their own models, right? The dependencies on all of these different open models, I think, is changing very rapidly. And I'm sure Anthropic woke up on the Windsurf announcement and said, 'Oh wait, OpenAI is going into verticals and we're good at coding, so how do we think about this?' And yeah, I mean, it's a reordering. I think people really need to pay attention to this data thing. If you look at where OpenAI wants to go on the consumer side, access to your contacts, your calendar, your mail, like all that's going to matter and you're going to want your personal assistant to be able to do that. And whoever owns those systems, whether or not they try to block access to it or not, will be interesting. You know, Google has an advantage in that they own their own phone platform and they own their own Gmail platform and they own their own alternative to the office stack. And so they should be able to make that a competitive advantage. But whether or not they'll be willing to put up a wall and say, 'OpenAI can't scrub this,' it'll all be very interesting. It's something that's super important to watch.
I mean, listen, I think it would be a bad development, right? You know, I think you and I both agree that MCP is a good development and allowing more open access. At the end of the day, it should be my data. It's my Gmail that I'm paying for. It's my Google Docs that I'm creating and I'm paying for. And so if you tell me that I'm not going to be able to use my AI of choice accessing this information to make my life better, to answer questions, that will lead to a lot of disruption. So, but that is the question, right? Because here we have Deep Research announcing today that they're going to access all those things using MCP on my behalf, which I'm excited about, right? Because I am using all of those Google services, but I do like the idea of using ChatGPT that has a lot of built-up memory about me to access those services. So I think you and I will certainly be loud and proud on the side of keeping this all open. But we definitely see some early warning signs here of people closing down the ecosystem in order to try to protect some of those advantages.
Another topic, Bill. You know, we've talked a lot about China and one of the interesting parts of our dialogue from my perspective over the course of the last few months is just how well China is doing in robots, in autos, in batteries, in precision manufacturing. DeepSeek coming out of nowhere, Huawei building chips that are catching up very quickly. But you sent me this piece of research this week and it's really enlightening because it talks about let a thousand flowers bloom, like how they've in fact seeded these industries and there's probably some stuff we could learn here. So unpack it for me, Bill. Why has this been so successful?
Yeah, and for the record, I had ChatGPT do the analysis. It was version 4.5, the $200 a month version. It's an amazing piece of research. I guess it could have some errors in it, but we're going to post it so everybody can see it. But yes, I think it's imperative that we understand exactly why China is so competitive in so many industries and what led them to be successful. And what I uncovered, and I had heard this from a few other people in the past, but what I really uncovered the detail on is that in these industries where they want to succeed, they actually make sure that there are 500 competitors or a thousand competitors and then they let the market whittle down to what the best one is. So rather than, I think we have a perception that communist or authoritarian governments have a single state-sponsored company that's not very competitive. That's not what they're doing. They're doing something very different. They're letting entrepreneurism and kind of a Darwinian competition and survival of the fittest create these five companies that survive all of that as winners. And we all know because we believe in capitalism that that kind of system will lead to the very best shining at the end of the day.
Well, it's interesting, Bill, because you and I grew up kind of with this model in Soviet Russia where they picked a state actor and part of the reason we won the Cold War is they bankrupted themselves. All those state actors were terrible. They were inefficient. They were full of grift. They couldn't compete. Their products weren't any good. But we see the exact opposite out of China. We see products that are globally competitive and many that are in these different industries better than the US. So there has to be something systematic and structurally different that they're doing even though we refer to them both as communist systems.
Correct. And I would, you know, I've studied these types of systems my whole life and it, you know, I've mentioned before that I'm on the board of the Sante Fe Institute and this kind of Darwinian competition is exactly the kind of thing that if you told me they were doing this before they did it, I'd say, well, that might actually work, right? Like this is sensible to me that this works. One thing that came out of it, in addition to just helping to identify winners, there's a couple of other benefits you get from this. One, you're exposed to way more optionality because 500 startups will try a bunch of different approaches. And one thing I've uncovered in the past few weeks is Chinese LiDAR is solid state and designed very differently than the LiDAR that Waymo is using. And it's actually in China right now being priced at like $130 a car, this solid state LiDAR, whereas the Waymo LiDAR is like $5,000 a car. So you end up in completely different places because of this type of competition and you have this many people trying different things. And then the other big thing that happens is your supply chain develops in a lot more robust way because if I'm a supplier to a part that's important to a solar panel or to an EV, I now have 50 or 100 competitors. So you end up, and that spurs more and more competitors for that part as well. And so you end up with a much more robust supply chain, more players at each step along the way. And I think the data suggests and this piece of research would suggest that's why they've been so successful in EVs where there were over 500 EV startups. 500. In the US, what has there been? Four legitimate ones? A handful.
So here's my question. Right, in Silicon Valley we have this dynamic ecosystem of risk capital. We're probably, I don't know, fourth, fifth generation of risk capital and risk takers. They find each other. There's not a lot of government interaction or coordination with regard to that. It just kind of happens. What are we saying? Like, how has this ecosystem developed? There's definitely venture capital in China, but it seems to me as a fraction of the venture capital that exists in the US or certainly the tightness of the ecosystem over the course of the last four or five years has been reduced dramatically because a lot of the US players backed off from China. So is it the different states and provinces that are seeding or subsidizing to get these 500 startups rolling? Like, how are they getting going?
Yeah. So the government involvement is at a provincial level and that's part of why there's such a high number, like a high number of startups in the area. I mean, it's interesting if you take this as a conclusion, oh, this worked, this was very successful for them, and then you turn and say what should the West do? I think it's hard because when our government's tried to get involved in startups, it's never been about helping to ensure there were a thousand of them. It's usually you create some program. The one that's stuck in my brain was solar because I actually had a company in the space where there were probably 10 solar startups that had raised over $500 million and then three of them were able to court the players in DC long enough to get money. Solyndra being the one that people remember the most because they eventually went bankrupt. But that doesn't feel like this thing, right? That feels more like regulatory capture and who can win at the highest level. This is at the beginning of the game. So I don't know if there's a US equivalent of this. I'll need to think about that. I think my first step was just to take a fresh look at why they've been successful. And when I saw that the approach was so novel from what I had imagined, it was very eye-opening.
Well, one of the things you pointed out to me was that they deprioritize market caps or I might even say over the last five years that they've kind of attacked the largest market cap companies in favor perhaps of diversifying competition. And so it seems like the one thing I think we can definitively say, there's a hell of a lot more national coordination about what industries are important, like their industrial policy identifies an industry. Then what it seems like they do from the central governing authority is they encourage all of the provinces to seed these companies within their different areas, allow that competition to occur. So I guess the question is, when you look at the US where it's much more just unfettered competition, is the point here just be aware of this or are you suggesting that there are things that the US needs to do? I mean, it seems in some ways like the coordinated industrial policy that's now coming out of Washington, the stuff that the president is talking about where we have to reshore critical national industries, precision manufacturing, some medical supplies, some chips, some aluminum and steel, that would seem to me to be fairly aligned with the industrial policy that you're discussing here.
Well, there's so much, like you just asked a question that might take four hours to answer, but let me just try and be as terse as possible in responding to it. So I had ChatGPT do two more pieces of research. The first one, or this would be part two. Part two is worth mentioning. So I think most people believe and talk about China subsidization. So once winners are identified, there are situations where the government has helped subsidize. BYD was in a situation where I think they were given $2 billion. So that second part walks through cases of that. I don't want to shy away from that part because that is the criticism that a lot of people bring to the table and so this will give a dump of that. But the third thing, what you hinted at, which I think is very important to think about, and before I get to any type of response, I want to make sure I fully understand what's happening. And there does seem to be in China, and part three of this research walks through this, a deprioritization of market cap of successful companies. And this is something that I think should be important for policy makers to understand but also investors. If you're buying stocks of Chinese companies hoping that they will, like the Mag 7, turn into these $3 trillion entities, the Chinese government may not consider that part of the objective function of what a win is. And in the past three weeks we saw BYD take prices down 30%. That may have had encouragement from the government. I don't have proof of that, but it looks synonymous with the kinds of things that matter to them. And so if your government cared mostly about high employment and cared mostly about the durability of the competitiveness of your companies globally, you might take what I would call the Amazon approach and say, your margin is my opportunity. I'm going to be the low-cost producer and that's going to make my competitive position relative to other countries around the world the best it can be. And if I don't have a government that's dependent upon whether or not those market caps are high or not, then I might encourage price competition in an industry that we're already winning at.
Well, and that's consistent. Like if you look at, as winners begin to emerge, China definitely plays a heavier hand, right? You have golden shares and veto rights by the government. You have preferential procurement by the government. You have regulatory approvals that are required. And in this post-DiDi era where they went public, I mean, we've seen the government literally, I mean, they disappeared Jack Ma with respect to Alibaba. They've stepped in, right? We have ByteDance, it still is not public. So clearly they exert way more control once the winners emerge. We know that they're doing this at some level with Huawei as well. And all those topics that you just mentioned, they're all covered in these three pieces. So if people have more interest, I would encourage them to read it. But I think in the US we have a mindset that having $3 trillion winners is a positive sign. And I think it's important to understand that that may not be the attitude over there and that can lead to different decision-making.
You're making me think about some of the relative valuation comparisons between Chinese internet companies and US companies if you really think that there's going to be an obstacle to allowing them to grow bigger. That's something US investors have to take into account.
Well, I think the number one thing for me, Bill, the so what on all of this is that we're in this competition with China. And I think we would be very naive to think that they're going to do anything but be extraordinarily competitive. You and I have argued they're on the frontier of AI already. They're gaining ground quickly on chips and right on our heels. And so I appreciated unpacking a little bit the why, like why have they been so successful there. So that was my takeaway. Bill, are there any other takeaways from the research that you have?
