Marc Andreessen 揭示其在 a16z 最大的成功与失误

It’s like a process of falling up the stairs. Just when you think you’ve got everything figured out, some weird issue pops up and you’ve got some new thing to prove. Every one of these companies that has ended up being a global world-beater always has one of these stories of the path not taken.

So, the firm was started—we raised the first fund in 2009, which was in the depths of the stock-market crash after the height of the financial crisis of 2008. We had the first LP meeting in 2010.

So this is, what, number 15?

Wow. Yeah. The first one was folding chairs. We had about 20 LPs to start with.

Wow.

Folding chairs in our office, in our little conference area, with our flat-panel TV on the wall. Actually, about the same people are still with us. A bunch of our original LPs are still here. This has grown to be a somewhat larger production.

Take us to some of the first LPs. They must have been true believers at the time, not just because of the strategy, but because of the market timing. Take us back a little bit to what that was like at first.

The thing to remember is that the dot-com boom and bubble was sort of 1995 to 2000, although it had its own ups and downs even during that period that got forgotten later. But in 2000, of course, there was a catastrophic crash, and then it took 3 or 4 years to dig out of the dot-com crash, so it was 2003 or 2004 before anything really started to happen.

Immediately, any sign of anything working in tech was met by screams in the press: “Bubble 2.0. Bubble 2.0.” It was just, “These crazy idiots are back at it again. Everything’s going to come crashing down.”

There were 2 big M&A events in 2005, I think. Yahoo bought Flickr and del.icio.us. They were $25 million tickets each, and the press went bananas. Facebook was getting started, and the reaction was, “Social media is a joke. What did your cat have for breakfast? Nobody cares, right? How are they going to make money? They’ll never make money.”

It was this wall of negativity heading into 2008. Then the financial crisis happened, and everything was completely devastated. Everybody in the financial investment world was flat on their back.

Ben and I decided in March 2009 that it was the perfect time to start a venture capital firm, which was an absolutely unique view at the time. There were only 2 venture capital funds raised in all of 2009. It was us and a new Khosla Ventures fund, and Vinod, of course, was one of the legends of venture capital. He was able to do that, but we were literally the only other venture fund raised that year.

At one point, were you considering starting another company together, or was it like, “Hey, we really want to support entrepreneurs. This is the company”?

We were done. We had started multiple companies, and we were done. We had completed that part of our life journey.

We decided to start a firm. Some people were really nice to us and really helped us get underway. Then we started meeting the LPs.

We’ll get to the evolution of the firm, but just because you mentioned it, the Zuckerberg acquisition story: You, of course, have been on the board of Facebook for a long time. Take us back to that memory, your conversations with Mark, and what your perspective was at the time.

Basically, this has been chronicled over the years, but what happened was that the negativity at that time wasn’t, “This technology is evil and it’s going to destroy the world.” The negativity was, “This technology is absolutely useless. It has no point and no purpose. It’s a joke. It’s a farce. It’ll never make any money.” That was the uniform public conversation around it at the time.

Facebook took off, and they had ad revenue early on, but it was remnant ad revenue. Facebook was literally selling remnant banner ads from the Bing ad network, so they were getting these super-low CPMs. Mark always had this theory of how he could turn it into a big thing, but he was facing this wall of negativity.

At that time, he had a lot of people around him. It had been a fast rush from 2004 to 2008 to get to the point where, all of a sudden, they were starting to get offers in the hundreds of millions of dollars.

Was this after a down round or before the down round?

I believe there was only one. The Microsoft down round was after that, because that was after the $15 billion round. It was a down round.

Yahoo at the time was doing a turnaround under Terry Semel. Terry correctly figured out that Facebook was a property they should buy. They actually struck a deal. I don’t know if it was literally signed or verbally agreed to, but they structured a deal for a $1 billion takeout.

Then the financial crisis hit. Advertising immediately collapsed, because advertising is the first thing that gets cut in a recession. Yahoo came back and essentially renegotiated the deal, trying to lower the price. That gave Mark the cover—the wherewithal—to walk away from the deal.

He was seriously considering it, or at least strongly considering it, at the $1 billion level. At the time, it was a spectacular home run for a company that was growing that fast, had started in a dorm 4 years earlier, and had done so during a time when this stuff was viewed as a complete joke.

The ultimate thing in the story is that, a few years later, somehow the internal Yahoo deck on the acquisition got leaked and published. It was Yahoo’s internal analysis of Facebook at that time. The Yahoo people were bullish on the company, which is why they offered to buy it, but they and everybody else radically underestimated the future growth of the company. It was an incredible lowball estimate.

Not only did Facebook grow really fast, but they figured out targeted advertising. By the way, they went through the exact same crisis again after they went public in 2012. They went public right as the transition from desktop to mobile was happening, really when the iPhone was hitting critical mass.

There was another wall of negativity. You read the articles at the time: “This is going to destroy Facebook,” because everybody knew that Internet ads were based on the amount of screen real estate you could sell. The ad rates literally were the number of pixels on the screen.

Therefore, because mobile screens were much smaller than desktop screens, logically the ad rates were going to shrink, and that was going to be it. The stock came out and got hammered right out of the gate. The argument was that the mobile ad opportunity was much less than the desktop ad opportunity.

What we know today is that mobile meant people were going to use this stuff far more.

Yeah.

Throughout the day, all-day usage went way up. Then we learned that the targeting works. Everybody knows today that a company like Facebook is a trove of personal data against which you can target ads, but at the time, people didn’t believe that to be the case.

I often describe these companies as—even the ones that, from the outside, look like they’re up and to the right the whole time—going through a process of falling up the stairs. It’s just that when you think you’ve got everything figured out, some weird issue pops up and you’ve got some new thing to prove. You’ve got some other thing to deal with.

Mark is one of the best who’s ever lived at working his way through every single thing that’s come along.

It’s funny that we went from “mobile ads won’t work” to organizations—Tristan Harris’s, for example—dedicated to stopping ads from working too well. Basically, ads have worked so well that we have an epidemic of people using them so much.

This is so fun.

I would even heighten that. It went, within 4 years, from “the ads don’t work at all” to “they are literally mind control.” They are literally mind control, right?

And, you know, this is political ads specifically.

They can literally put the whammy on people to vote for candidates nobody would ever vote for had they not been mind-controlled.

Exactly. Right—by the Facebook algorithm.

And the whole thing was so weird because it was like, wow, if the mind control works so well, why don't we have triple the ad rates on toothpaste? Why does it only work for Trump?

Yeah, exactly. $100,000 can swing the entire election.

Right. Exactly. That's the thing. The Russians can spend. By the way, I think the result of it was a result of $80,000 of ad spend by the Russians. In the relevant period of the election, it was a total of about $140,000, but during the actual run of the election, it was $80,000.

So the theory was that $80,000 of ads channeled into this magic mind-control device—

Bang for the buck.

Yeah. That literally swung an election that Hillary Clinton spent $3 billion on.

And the whole time we're just sitting there—I'm just sitting there—like, I can't even believe it. It doesn't make any sense. I went to Hillary's big speech, her first big speech after she lost, at Stanford, and she literally sat onstage and said, "Donald Trump is only president today because Vladimir Putin had Facebook." I'm like, "Vladimir Putin didn't have Facebook. What the fuck are you talking about?"

And so, literally, it went from 2008 being useless, to 2012 being still useless—although it did get some credit for the Arab Spring and for getting Obama elected. There were lots of headlines at that time saying, "Social media saves democracy," because of the Arab Spring. Then, by 2016, it was the evil mind-controlling death machine.

It turned out that the method supposedly used by the Russians or the Trump campaign to do the mind control—basically, that method—we now know never worked. The whole psychometric thing that Cambridge Analytica was supposedly doing isn't something people do today.

Yeah.

This is the other part of the narrative you're forced to believe: you're told it was magic mind control, but only for that election, and then nobody ever tried it again.

Exactly. It's fascinating to imagine alternative tech history. What if Yahoo had actually been able to buy Facebook? Obviously, there would have been a massive blow for Facebook, but sometimes it works the other way. Maybe Snap should have sold to Facebook. Maybe it would have been better for everyone else.

I have endless stories like that. Netscape—we almost bought Yahoo. There was a bid-ask spread between $3 million and $5 million. Yahoo almost bought Google, and there was a very low-priced deal at the time. Netflix almost sold to Blockbuster early on.

Every one of these companies that has ended up being a global world-beater always has one of these stories of the path not taken.

Uber and Lyft almost merged at one point.

Yeah, exactly. There's all that. So, I mean, look, the forces mean something like this: the smartphone was going to happen. There were going to be killer apps on the phone. There was going to be something like this. Somebody was going to do it.

Had Facebook sold to Yahoo and Yahoo let it languish, somebody else probably would have figured it out. In a sense, some of these things were bound to happen. But the specific companies involved—there are so many twists and turns along the way.

It's so contingent on very specific people doing very specific things at each point.

Totally. I think it's extremely sensitive to the microlevel decisions that are made along the way.

Speaking of twists and turns, let's go back to the firm, because the first fund was $300 million, correct? We didn't have a sense for how big it would become. You wrote "Why Software Is Eating the World," so you had a hypothesis that these companies were going to get bigger and there were going to be more of them. Take us through when you realized that scale was going to be a strategic asset for venture, and that the future of venture was going to look like this barbell.

I remember, actually, having a conversation with John Doerr in the mid-1990s, because Kleiner Perkins was an investor in Netscape at the time. In those days, when I was young, the model was very straightforward: Series A, Series B. Venture was Series A and Series B, and there wasn't even really seed investing that much.

You would raise a venture A round, which would be a $3 million, $4 million, or $5 million round—maybe $6 million or $7 million if you were super ambitious. Then there would be a B round of $20 million or $30 million, followed by what was called the mezzanine round, which was the C round. That was the pre-IPO round.

You would raise a total of 3 rounds, maybe $30 million or $40 million, and then you would go public. There were basically no follow-on rounds after the C round. You could go public in those days at around $50 million in revenue, and a 10x revenue multiple meant a $500 million market cap was kind of the bar to go public.

By the way, Amazon went public in 1997, and I think it was at a $400 million valuation. So it was right in line with that model. Netscape was similar in 1995. Venture had this very specific role.

What was becoming clear even in the 1990s—I think it was a conversation John and I had about Cisco originally—was that some of these companies were really going to rip. They were really going to run. At the time, it was like, if they were going to run to $100 billion, which was the big ceiling, all of a sudden the venture B round looked great. The mezzanine round looked great. Maybe the VCs should just keep re-upping.

If you could make as much absolute money investing $50 million in Cisco at its Series C as you could investing $5 million in its Series A, was it a mistake for the venture firms not to do that? In those days, it was a big statement to say maybe the venture firms should do that.

Of course, what's become clear since then is that the ceiling isn't $100 billion. It's much larger than that, much, much higher than that. That's just a consequence of tech proliferating and becoming more important.

Facebook was raising larger rounds by that point. DST kind of redefined growth investing when it did the Facebook round. This idea of venture-growth investing was starting to really materialize.

We defined the firm right up front. We said we wanted to be stage-agnostic. We wanted to do seed, venture, and growth. That was in the original pitch deck. It wasn't enough money to do a lot of growth investing, but we could start to get underway.

Our argument to the LPs was: look, it's not about being at a certain stage; it's about the total aggregate opportunity. If you're going to get into one of these world-beating companies, you can still do it with a B, C, or D, and you'll still have venture-scale returns.

When did that model originate? Our approach of specialized funds is different from some of these other partnerships that say, "Hey, we're just all generalists. We're a small team. We all do everything." What was that journey like?

We started that way. We started as generalists, and Ben and I had both worked in both consumer and enterprise, which were kind of the 2 big categories at the time. It was basically consumer software VC, enterprise software VC, and then bio-health tech.

The older venture firms actually did both what they called digital, or IT investing—software investing—and biotech. A lot of those firms had internal divorces and spun their groups off because those industries at the time were going in different directions.

But Ben and I had done consumer and enterprise, so we just said, "Look, we'll be a generalist firm," modeled after the Benchmark model. A big part of this was industry evolution. Around the time we started the firm, the tech industry fundamentally changed from primarily building tools to actually going directly into industry. We called them full-stack companies.

I think the reason the generalist model worked well for as long as it did is that, fundamentally, a database company, a router company, a word processor company, and an operating system company are all fundamentally tools.

They’re different if they sell to the consumer or enterprise, but a router is a piece of software in a box. Chips are a little different, but a database and an operating system are kind of similar. The technical challenges, the form of the company, and the organizational model for the company are kind of similar.

Word processors and video games at the time both cost $50 and were sold in boxes on retail store shelves. They were kind of similar, and so the generalist model worked really well when the industry was that earlier model. Then 2009–2010 was the pivot we now know: the rise of Uber and Lyft, the rise of Airbnb, the rise of Tesla, and the rise of SpaceX.

The world-beating companies of the 2010s, in many cases, were companies that were full-stack, direct insertion into end markets. We therefore leaned hard into that because we saw it happening. We started investing in everything from e-commerce marketplaces to new kinds of defense companies, and then we decided to take on biotech because we thought then, and believe now, that the convergence is happening. Biotech is becoming much more based on software, IT, data, and AI.

What we realized was, okay, the generalist model doesn’t work anymore because each of those things is now a very deep domain in and of itself, with very specific domain knowledge. The specific problem that a generalist has is that a generalist can sense heat, but a generalist has a very hard time getting into the specifics.

The reason the specifics really matter is that so much of venture works this way: if you invest in 1 company in a space, you can’t invest in the others. If you pick the space correctly and invest in the wrong company, you’re screwed because you can’t fix your mistake. You can’t then invest in the successful company because of the conflict issue.

What we realized was that the thing becoming very important in this new world was understanding, deep in the vertical, what was going on—specifically for the purpose of being able to tell which company was actually the one most likely to win. That’s just really hard to do if you don’t have domain knowledge.

That catalyzed us to verticalize, and we did that in 2 steps. We did a partial verticalization early, probably around 2013 or so, and then by 2017, I think we did the full verticalization—the one we have now.

It’s fascinating because 15 or 20 years ago, as things like Y Combinator were getting off the ground—and you were saying this yourself—people said there was an arbitrage around young technical founders. The idea was that we could teach them the business elements and they could be successful. You don’t need to go to an MBA program to be a successful founder. That’s accurate, and yet today maybe there’s a similar arbitrage around domain experts, where the technology has gotten easier and it’s easier to get things off the ground, and domain expertise and distribution are valued more than they were 15 years ago. Maybe that’s true on the founder side and on the investor side.

Although, at least until now, I think the question is whether AI is going to change this. At least until now, it was still the case that founders also needed to be very deep in the technology.

Even the founders who would go out—and I used to get these; there’s this wave of design founders, right?—it’s like, okay, that’s great. Good design founders are great. It turns out the successful design founders are also very strong technologists. They maybe pretended sometimes that they weren’t because they wanted the design cred, but it turned out they were actually very deep and substantive technologists.

Pinterest—these were Ben Silbermann and Brian Chesky. These guys are top-end technical founders, in addition to being good designers—great designers.

I would argue there’s really no escape from deep domain knowledge. To your point, I think the question is, okay, if AI makes deep domain knowledge instantly accessible to anybody—if everybody’s got o3 Deep Research at their fingertips—and if you’re a generalist or a person wearing many hats, can AI give you the depth when you need it? I don’t know. I’m excited to see people try that.

It’s a little bit like the debate happening around coding right now: Is vibe coding a substitute for actual writing? Sitting here today, vibe coding is super exciting for casual development, but at least right now, all of the really sharp technologists I know don’t think that you could vibe-code a top-end software company today.

The coding capabilities of these models are getting really good, and the agents are starting to work.

There’s a famous Dilbert in which Dilbert’s boss pitches Dilbert on starting a company together. The boss says, “I’m going to bring the idea and be the manager, and you’re going to do the coding.” Dilbert says, “And so what you’re saying is, you’re going to contribute nothing?”

Then you get these very entertaining thought experiments, like: Are agents going to get really good at writing code, and are you going to be able to have a nontechnical person supervising 1,000 AI coders and outranking a top-end technical person who’s supervising 100 great coders? It’s possible. We’ll see if it’s the era of the ideas guy.

I want to talk about areas where we’ve chosen to play and areas where we haven’t chosen to play. Let’s look internationally. We talked about how you presently didn’t enter China in ways that other firms did. As a U.S. firm, with American dynamism, and Europe—when it’s not shooting itself in the foot from a regulatory perspective—is increasing its defense spending. Could you imagine European dynamism? How have we thought about internationally, historically, and how do you think about it going forward?

The good news is that the world is globalizing. There is incredible activity and interest all over the world. Everywhere we go, people want to talk about tech; they want to learn about tech.

Ben saw this more than me, but when Ben wrote his book and went on tour, it didn’t matter what country we went to—thousands of kids showed up, and they wanted to learn how to do this. The internet had a big impact there: every kid anywhere in the world can now watch Peter Thiel talks and our podcast and everything else. Global knowledge is way up, global enthusiasm is way up, and relevance to societies is way up.

As you mentioned, Europe needs to rearm. Presumably, it would be better off rearming with the new systems as opposed to the old systems—drone swarms instead of aircraft carriers, that kind of thing. There are smart people all over the place, so there’s human capital everywhere.

I think historians will look back and say, “Wow, the 21st century was a really primitive society. They had all these smart people all over the world, and they never figured out how to actually utilize them. They never figured out how to identify them and use them.” The startup process is a way to get smart young people to do ambitious things.

That’s all fantastic. Against that, there’s this incredible drag from bad governments and bad policies. As you mentioned, Europe persists in shooting itself not just in the foot, but in the other foot, the ankle, the knee, and the gut. It’s on this absolute frenzy to regulate and kill tech in Europe and the U.K.

And they’re proud of it.

They’re proud of it. They’re very proud of it. The actual European line now is—we quote this from the Financial Times; this is an actual quote from a senior European politician—“We know we cannot be the global leader in tech innovation, so therefore we will be the global leader in tech regulation.”

You just imagine being a German or French tech founder and reading that: “Oh, God. Get to the U.S. embassy and apply for a visa as fast as possible.”

In fact, that’s what’s happened. As a consequence, a lot of these really curious, bright people basically end up moving to the U.S. We’ve been an enormous beneficiary, as a country and as a firm, of that. That remains a really big challenge.

It's interesting watching the U.K. right now because I thought that under the previous government, they were going to become enlightened, and they did the opposite. They tried to ban AI, and they said they were going to liberalize crypto, but they never did. The new government says they're going to open some of this stuff up. They still want to keep all the regulations, but they say they want economic growth. So we'll see the defense stuff.

Yeah. I mean, there's already a dispute—this is just in the headlines—over whether, if Europe cranks its defense spending up to 5% of GDP, as it's supposed to, it will buy any of that from the U.S., or whether it will only be purchased from indigenous European defense vendors. It's already becoming part of the whole trade war. Look, I think most of the state of the world is, at least for what we do, high-caliber people being held back by being in bad systems.

Yeah. What are our criteria for opening offices in other places more generally?

Basically, there are 2 kinds of activities that make sense for a firm like us to do internationally—maybe 3. One is just sales. The crude term is business development, dressed up with fancy words. Go-to-market matters because a lot of the market is global, and so it's about helping all of our companies succeed and build up businesses globally. We've been doing that from the beginning, and that's always an important thing.

The second is investing. We've always been—and I would say, other than China, we generally have been—open to investing in many places in the world. We have investments in a lot of places, very recently in Mistral in France.

Exactly.

Yeah, exactly. But we've made investments in the past in Vietnam and lots of other interesting places. Having said that, again, just because of this political dynamic, it has been striking over the last 30 years how many great founders who could have stayed in Country X and started a company there have just moved to the U.S. So, from a practical standpoint, we've tended to emphasize U.S. investing because we figure we'll get most of the best global founders anyway.

Yeah.

Maybe the 3rd thing—we're not really active on this right now—is policy. We're very active now in U.S. policy, but there are policy issues all over the world. Do we need to insert ourselves more into—God help us—European politics, or whatever? And then the question is: As Americans, are we allowed to?

Yeah. Well, I know we've only done this sort of kicking and screaming. This is not our intent, even on the U.S. side, but talk a little bit about how the little tech concept sort of emerged in our thinking as distinct from big tech.

Basically, this goes right back to the tech industry changing around 2010. In the era between, say, 1950 and 2010, when the industry was in what is called the tool phase—chips, word processors, routers, spreadsheets, and things like that—tech was never really a salient political topic. Startup tech, particularly, was never really a salient political topic. Really, the only time it would ever enter politics was with antitrust.

That was a really big deal in the 1990s when the government dug into Microsoft, because it was a big deal—the case was a big deal—but also because it was a very rare case of the government actually caring about tech and wanting to become involved. There was a state of benign neglect, or something, up until 2010. And, by the way, in both directions: The Valley companies never thought it was important to go to D.C., and the D.C. companies didn't really think the Valley companies mattered.

That all started changing around, I don't know, probably 2012 or 2013. There was increasing political energizing happening, in general. The anti-tech narrative started getting more political. Tech got wrapped up in the Occupy Wall Street 1% thing: Are these tech people just as bad as the finance people? The FT just ran a big story saying the tech people are now definitively worse than the finance people. I read the story twice, and I still don't understand it, but the gist of it was something like, at least the finance people you could have dinner with and it's fun, right? Whereas the tech people are just—

Sometimes they say, at least the finance people know that they're money-motivated or something, whereas the tech people have the delusion that—

Yeah. They're saying sanctimonious and insufferable, which sometimes maybe there's a little bit of truth to that.

But, yeah, it started to energize then, and then really the takeoff was 2015–2016. It was really Trump's nomination and election that radicalized a pretty big block of political actors in the country against tech. And, by the way, the way I would describe the last decade is: The left got super angry at tech because of Trump, inequality, and all these other things, and then the right got super mad at tech because the right thought the tech people were all on the left.

Yeah.

Right. And so we'd go talk to Republicans, and they would basically say, "Yeah, we agree with you, but you're all Democrats, so, you know, f-off." They thought it was very entertaining.

Social media was kind of the tip of the spear of that for about 5 years. Then, in the last 5 years, crypto became incredibly politicized, and the U.S. government basically launched a full-on war against it and tried to kill it. Then AI got incredibly politicized, starting basically right around 2021–2022. The government was moving in very hard on AI, and for us, it was the combination of crypto and AI that got us extremely alarmed. So we decided we had to become involved.

But, like I said, there were very specific policy issues in both of those domains, but there was also this general kind of thing: We always wanted to build things that matter. It turns out these things matter, right? As a consequence, we're either going to show up and explain ourselves, or other people who hate us are going to describe us. So we've generalized that idea into what we call the Little Tech Agenda.

And the first part of that is the Little Tech, and we say "Little Tech" specifically to differentiate it from Big Tech.

I like that it creates a cleavage.

Yes, that's right. That was another thing we discovered as a result of our early interactions 4 or 5 years ago in D.C. They were just like, "Well, tech is tech." We'd meet with them on behalf of startups, and they'd start yelling at us about Google or something. We'd say, "We're not here representing Google," and they'd say, "Well, that's tech." And, by the way, it was a startup 20 years ago, right?

We'd say, "Well, look, Google has its own people. They have 10,000 lawyers, policy people, and compliance people. They've got their own people; they can carry their own water. That's not what we're here for." In fact, a lot of what the venture ecosystem does is fund competitors to Big Tech—to disrupt Google.

Yeah. We're disrupting Big Tech.

That's the story I always tell in D.C., and it's 100% true: The caricature of reckless Silicon Valley is that we're starting companies that go disrupt whatever—health care or something. I say most of what our founders are trying to disrupt is Big Tech.

Frankly, Big Tech companies have a love-hate relationship with us because we're the escape hatch. If there's somebody at a Big Tech company who gets frustrated, they come to us to raise money to start a company to compete with their former employer.

Yeah. Right.

And so Big Tech is not always in favor of us. Big Tech companies have become very big, and they have their own agendas. A lot of big companies, over time, try to do things to prevent—try to form cartel-like structures to prevent new competitors, startups, from emerging. We noticed in more and more cases there was a divergence of interests. That actually helped a lot the minute we figured that out, and then we coined this Little Tech thing and ran with it.

It was basically immediate—that was the icebreaker in D.C., because everybody in D.C. hates Big Tech. They only agree on 2 things: Big Tech and China. These are the only 2 things they agree on, but they really agree on those things, and they all hate Big Tech.

The minute we say, “We’re not here on behalf of Big Tech; we’re on behalf of Little Tech. In fact, our companies attack Big Tech,” we get these huge smiles because they’re like, “Oh, wow. Tech startups are actually good, right? This is actually exciting.” That’s the Little Tech side.

The agenda side is essentially freedom to innovate, right? The ability for companies in new fields that aren’t yet completely well understood, that lawmakers of the past did not anticipate, to actually be able to operate.

The other maybe important thing is that we’re not lobbying for no regulation, right? Sometimes we get tagged as these crazy libertarians who want these companies to just run wild, commit fraud, and do all this crazy stuff. That’s not what we’re pushing for. We want clarity. We want clear guidelines and sensible rules. I always tell these people: consumer protection regulation, go crazy. 100% no problem. That's all good. In fact, we want that because we want our companies to be compliant. We certainly don't want our companies committing fraud. By the way, the other thing we don't want is we don't want our compliant companies competing with non-compliant companies that aren't getting regulated or prosecuted.

Yeah. We’re advocating on behalf of regulation as often as we’re doing anything else. So, it’s funny, because there was the Paul Graham–Palantir dust-up on Twitter the other week, which I thought was symbolic, or emblematic, of—not to pick on Paul—older Silicon Valley versus newer Silicon Valley, or just how the space has changed.

Palantir has to work with the government. They’re building hard tech. It’s not just putting something up and talking to customers right away. Paul Graham has been in England for quite a bit. YC is still innovating and doing fantastic things, but I thought that was such a fascinating example of how Silicon Valley is evolving: new kinds of companies, new kinds of go-to-market strategies, and new kinds of ethos.

That’s right. The specific dust-up, I think, was over Palantir working with ICE.

Yes, ICE, the U.S. government immigration agency.

So, yeah, there’s actually a deep—there’s kind of a deep generational change. I think it’s a very deep and profound change, and there’s a generational aspect to it.

Basically, the U.S. tech industry and startup industry in Silicon Valley were deeply intertwined with defense and intelligence—basically, the deep state—and the federal government, really from the 1920s onward. Steve Blank has these incredible videos and essays where he goes through all this. Silicon Valley actually started in the 1920s, and it was defense tech: radar, early missile-guidance systems, airplane avionics systems, and things like that in the 1920s and ’30s.

From that time through basically the 1960s, it was just assumed that there was this very deep relationship. With the Cold War, industry and government had to work together to build all these systems to defeat the Soviets. One of the first computers was literally called SAGE, which was a U.S. government project—a giant mainframe computer for missile early-warning systems. It was just assumed that this was all kind of hand in hand.

Then really what happened was Vietnam. Vietnam was the first modern war where you had this massive protest movement. College campuses went wild with this anti-war movement and all that stuff. If you talk to people who were involved back then, what happened was that the leading-edge American universities that did science and technology were doing military science and technology, and many of them just stopped.

Stanford split up. There’s this thing called Stanford Research Institute, SRI, that’s actually no longer associated with Stanford because it’s the remnants of what used to be Stanford’s military R&D. MIT had something similar with Lincoln Lab.

Yeah.

That created a culture. The Valley was always a mix of right- and left-wing, but the hippie anti-war movement was largely in California, especially at Stanford and Berkeley. There was this mood shift. I guess we have the vibe shift now, so in those days it was probably the groove shift. Call it the groove shift.

The groove shift 50 years ago was: no more work with the military, no more work with the U.S. government. The U.S. government was an evil colonial, oppressive power wreaking havoc overseas, and it would be better to have disarmament and pacifism. The U.S. should give up all these foreign engagements. That was a very hot political topic because the Soviet Union was still up and running.

As a result, a lot of these universities, and then a lot of the companies, basically disengaged. They just weren’t willing to do that. Another topic that became very hot at the time was so-called ROTC recruiting. The military used to actively recruit new officers on campuses, and a lot of these colleges outlawed that 50 years ago.

Anyway, Paul is a very bright guy. He’s kind of from that era—not the Vietnam era, but the arc that followed, in the ’80s and ’90s, when it was universally assumed that, of course, you don’t try to enable the military-industrial complex or defense agencies.

I never really felt that way. Maybe it’s just—I don’t know where I grew up or whatever—but I never really felt that way. But I knew I was swimming in those waters for a long time. My company, Netscape, was always very active with defense and intelligence. We always leaned hard into it.

Alex Karp, I think, is a national hero. The one thing he does is say that Palantir is the first company that’s ever done this, and I’m like, “No, no, we were doing it 10 years earlier,” but that’s okay. He gets tons of credit.

I would say Paul kind of represents that neo-hippie ’80s and ’90s phase where it was just clearly assumed that you don’t do this. Now, more recently, the vibe shift is happening. With Palantir and Alex Karp—and again, give Alex Karp huge credit, give Brian Schimpf and Palmer Luckey huge credit, and Peter Thiel—these guys really, really led the way on this.

The new model is: of course you work with them, because, number 1, you’re a business and they’re a customer, and you shouldn’t be sitting there as a moral judge for every single customer. How are you going to do business if you’re just sitting there having these moral conversations all day long?

Number 2, these national missions actually matter, right? The border issue is very inflammatory, but most people in the history of civilization have thought that borders are an important and valid thing. Is the world a safer place because the U.S. is in it versus not, and because the U.S. military is strong versus weak? Do we want to protect ourselves against terror attacks?

I think there are a lot of us who are like, “Clearly, we want to do those things.” I’m encouraged by it because I think things are headed in the right direction according to what I want. But I also recognize that these are serious underlying moral, philosophical, and ethical questions.

I don’t question Paul’s intent. I think when he says those things, he’s operating out of a very strong sense of ethics and morals.

Totally. I disagree with him, yes. Software is eating the world, and it didn’t stop at social networks or marketplaces, right? It got into Internet marketplaces. It got into everything.

That’s right. In fact, what happened—if you go to D.C. and talk to people who work there, if you talk to the people who run the defense and intelligence agencies in D.C.—is that almost every topic they talk about or think about has a major tech component to it.

Like almost every single one. Yeah.

Obviously, the Ukrainian battlefield is a completely new kind of battlefield because of all this new drone technology and all these things. And that's basically true for every area of national geopolitical policy: It has a big tech component to it, right?

Of course, my approach is obviously that we should be in the middle of that, and we should be investing against that and trying to help those things happen. But I understand why people get uncomfortable with it.

That's a perfect place to wrap our conversation at LP Summit. Marc, thanks so much.

Good. Thank you, Erik.