Bill Maris: How Google Could Crush AI Competitors, Why Small Funds Win, and AI's Atari Stage

After saying he was out, Bill Maris is returning to the investing world. The founding CEO of Google Ventures has raised $150 million for his new fund, called Section 32.

With a smaller fund, I have the advantage of being very selective in the companies that I invest in and the people that I hire. We're going to invest for a financial return. Any other metric is impossible to measure and, therefore, won't succeed.

Think of the change that has happened just in the last 100 years and what's about to happen in the next 100 years with the advent of AI. The world's going to change by orders of magnitude.

Thank you very much for that warm welcome. I am Bill Maris. I'm the founder of Section 32. Prior to that, I was the founder and CEO of Google Ventures. I was also Google's vice president of special projects, where I incubated Waymo, Google X, Calico, and many other projects as well. Before that, I founded a web hosting and data center company, which we're going to talk a little bit about.

Today, I think I'm going to talk to you about a few of the lessons I've learned from these interesting experiences I've had in life. We're going to have 4 lessons, and we're going to go back to 1997 to start, when I was a fresh college graduate.

I had a degree in neuroscience, and I found myself on Wall Street. I somehow managed to land a job there, but I was miserable having to wear a suit and trudge to work in the heat. One good thing came of that, which was that I looked in the closet of the office one day and saw a server. I asked, "What is this thing beneath our jackets?" They said, "That's where our email and websites live."

As can happen to many of us, I had a moment where I felt like I was bathed in the light of inspiration. I thought, "I think I've glimpsed the future. I think I can maybe make a business out of this, because if you can have our website and email in your closet, how many websites and emails could I put in my closet?"

I immediately quit my job because I had glimpsed through a keyhole, and through that keyhole I thought I saw the internet. I saw a data center, and it looked something like this. Or maybe when I say "data center," you think of something like this or something like this. But in 1997, the state-of-the-art data center looked almost exactly like this.

We had 3 servers: a small, medium, and large. Business grew, and we eventually had 5 servers. This isn't a data center at all. This was my apartment, where I founded the company with credit cards, and the servers lived in 1 room. The work happened in the other room, and it would get very hot in that room. This was in Vermont, so I opened the windows, and then it would get very cold—so cold, in fact, that by noon, if you had a glass of water on your desk, it would ice over.

You may think this isn't so bad, but this was also my apartment. This was the bed, and you may look at that and think, "You've got a mattress and a nice pillow, and look at that nice blanket." But this is a rug I got from Home Depot to keep myself warm on those nights.

One day, there was a thunderstorm. The roof started to leak, and I knew I needed to do something because water and computers and servers don't mix well. I called the landlord and said, "The roof's leaking." The landlord said, "That happens sometimes."

But I knew that I needed to do something. When you don't know what to do, you go to Home Depot. I got a bucket of tar and a mop, and I went up on the roof. There was lightning, and there was rain, and I went up there and tarred the roof.

I did not glimpse the future in that case because I didn't know that when you're tarring the roof, you should start at the far corner and work toward the door rather than the reverse. I tarred myself into a corner, but the choice I faced was either the servers got electrocuted or perhaps I got electrocuted. As an entrepreneur, I was willing to take that risk—which, news flash, I survived.

My shoes, though, are still stuck on that roof in Vermont. That takes me to lesson 2: To see the future, sometimes you need to be a little bit insane.

It may appear to those around you that you were tarring the roof in a thunderstorm. To that point, I'm going to share a few slides here that a friend named Stewart Butterfield was kind enough to share with me.

Here's the inauguration in 1989, and there's someone taking a picture. That makes sense—probably a film camera. Then, in 2005, it's not very different. There's still someone back there taking a picture.

Let's go just 4 years later, to another inauguration. If we look closely, it's quite a bit different because now everybody's got a camera. Everybody's got a camera, and this was before cameras were merged into cell phones. It was around that time that it was starting to happen.

But that's not the most interesting thing about this photo, because in this crowd is someone who, to his friends, I'm sure seemed insane, and who also glimpsed the future. If we look closely, this gentleman has decided to—I don't know—livestream or record the inauguration on his laptop.

He knew something that those around him didn't know, which is one of the things that I've always looked for in entrepreneurs: They know a secret about the future that most of us don't believe.

Let's fast-forward to 2007. I find myself somehow at Google, and a challenge was given to me. The challenge was, "Google needs a venture fund." We were starting to make some investments, but we didn't have a coherent strategy or any budgets. I had to figure out what to do.

I first found a friend, Rich Miner, who's the co-founder of Android, and he became my partner in crime as we conceptualized what Google Ventures could be. We went up and down Sand Hill Road, and we talked to everyone. Anyone who was willing to talk to us and have a conversation, we were willing to talk to and see what we could learn.

We came up with a plan. Our plan was to obtain all the venture data that we could find. Being Google, you can imagine it was a lot of data: historical data, you name it.

Then we decided that, as step 2, we would use AI. But at that time, Google would not let us use the term "AI." This persisted for many years: "Bill, AI is science fiction. It is 100 years away, if it's ever going to happen. Let's stick to machine learning. By the way, when you say AI, it freaks people out. So stop freaking people out."

We had to call it machine learning, and we used machine learning to do 2 things: design the ideal portfolio construction by running millions and millions of simulations and back-testing, along with all the things you can imagine that data scientists would do; and determine what the ideal fund size would be.

People were excited. Here's a headline from TechCrunch at the time. People inside Google were also pretty excited. This is something one of the senior executives later told me.

I have to admit, it seemed crazy. The plan seemed crazy at the time, but let's look at how it turned out. Over this time period, from 2009 to 2018, top-quartile VC returns looked like this, and top-decile returns looked like this.

Using publicly available information, and not sharing any nonpublic or proprietary Google information, we would estimate Google Ventures' returns at about 4.1x. I adhered more closely to the strategy, and the investments that I led turned out like this.

That takes me to lesson 3: Don't bet against computer science. I've seen it happen many times in many fields. If you apply the right kind of computer science at the right time to the right problem, you will get to the right answers. I would not bet against it, even if it looks like you're tarring the roof in a thunderstorm.

Let's fast-forward to 2017. I decided to start my own fund. Again, those around me said, "You're insane. Why would you do that? You're in the warm womb of Google. Lunch is free, the massages are plentiful, and so forth."

After the idea sunk in, the advice turned into, "Raise as much money as possible. That's the right way to run a fund. You'll get a big management fee. You'll be happy. Things are going to work out really well for you."

I thought about that relative to everything I had done up to that point, and I decided not to take that advice. Over the course of my time at Section 32, we've had 6 funds. We've invested in companies like CrowdStrike, Cohere, and Coinbase. All 6 of those funds have averaged about $400 million in size, and all are performing in their top decile.

To the extent that there is DPI to measure, that's the only measure in venture that counts, as far as I'm concerned. That takes me to lesson 4. This will be heresy to some, but small funds outperform large funds.

This is simply the math. This is not an opinion I'm trying to convince you of, but there are many reasons for this. With smaller funds, you can have more focus. I've already managed a multibillion-dollar fund with hundreds of employees. It's distracting. You cannot give the attention to founders that I would like to give.

There are many reasons for this. If we look at top-decile performance by DPI, funds smaller than $750 million had an average return of 4.76x, while funds larger than $1 billion had an average return of 2.42x. Funds below $750 million across that time period represented 95% of top-decile performers, with discontinuous return compression above $750 million.

Why is this? There are a lot of reasons. You can use your own numbers, but I'll just do a little thought experiment. If you have a $500 million fund and, on average, these days you can own 10% of a company, you need $5 billion of exits to get your money back.

Let’s remind ourselves that the 75th percentile of venture loses money and that there is persistence of performance in the top quartile. If you need $5 billion to get your money back and you want to be in this business for the long term, let’s say you set your goal at 3x, you need to return $15 billion of exit value in your companies.

Now, if you have a $7 billion fund and we do the same math, you’ve got to return $210 billion. $7 billion to $70 billion, times 3x, is $210 billion, which exceeds the total venture-backed M&A and IPO exit value in most years. This year may be an exception, but that is something I’m looking forward to talking about when we sit down.

For those of you who want the numbers, we’ve crunched them; we’ve done all the math. Those are Bill’s 4 lessons for today. I hope that they’re somewhat useful. There are a lot of stories behind all this, and I’m looking forward to talking about them for a few minutes with the guys. Thanks so much.

You guys are old friends.

Yes, we are.

We go way back. Well, Bill, when he started Google Ventures, I was the first ex-Google company you invested in.

That’s correct.

How did it go?

Climate Corp.—a $1 billion exit to Monsanto.

What was your multiple? What was the return?

Oof, I don’t know.

It was actually good for you guys.

It was quite good, yes.

Back then, that was a good deal.

That would have been the seed round.

Like an A round?

Yeah, that would have been your A round.

Now we’re going to do it again with A Halo. I just want to juxtapose what you said with what Thomas shared. They’ve got a very large capital base that they invest, and they’re investing significantly in these later-stage rounds of these well-proven companies where, as the data he shared shows, you can get significant multiples to get to that next phase.

You’re more likely to go from $1 billion to $10 billion, and then you’re more likely to go from $10 billion to $100 billion, $100 billion to $1 trillion, $1 trillion to whatever. Doesn’t that justify an alternative strategy to what you’re saying, of having smaller funds focused on venture that you can maybe barbell? Have smaller vehicles focused on venture, and then very large vehicles that bet on the sure things that have that durability and that compounding advantage. You can kind of have the 2 together both be a 3x return.

My observation on that would be, 1, I haven’t seen the data science to support that second conclusion—that late-stage companies can be an ongoing trend—other than this one moment, this weird moment in time with these multitrillion-dollar exits that are coming. That would be observation 1.

2 would be that, at a certain point—and this is not a negative; it’s just an observation—if you’re an RIA and you’re collecting assets, that is not venture. Venture, as I practice it at least, is a different craft, where you are making concentrated bets of your time and capital on entrepreneurs and helping them build a business. There’s nothing wrong with late-stage investing.

However, I also have an observation that I have a bit of an objection to companies that wrap themselves up in public-benefit language and then keep the value creation to themselves and an elite group of investors through a big part of the curve, and then say, “Well, we’re here to benefit humanity.” Well, what humanity needs is money.

So it might be better to go public sooner, because we’ll see how these multitrillion-dollar IPOs go. However, if I’m Google—and I don’t speak for Google—and I decide to arbitrarily cut the cost of tokens to 80%, what happens to the business models of Open AI and Anthropic at that point?

What happens? Tell us. Actually, what does happen?

Well, if you’re a company and you can go to Google and Gemini and pay 80% less for that basically identical product, why wouldn’t you do that? Then the compression and the pressure on those other businesses goes supercritical.

What are the chances that the other shoe has fallen?

That might happen. If I were Google, that’s what I’d do.

Walk us through the scenario where Google decides, with its war chest and its money-printing machine, “You know what? Their margin is my opportunity. I’m going to give tokens out for 20¢ on the dollar.” Every time they lower their price, I lower our price. What happens on the playing field?

Would that not be the rational thing for Google?

It’s clear they’re going to do it.

Well, it may not be a margin, though, to the—

They may be burning investor cash, sort of like an Uber-type model, to grab market share.

Capital as a weapon, tokens as a weapon.

Token as a weapon, grab market share, grab an install base in consumer and enterprise. But fundamentally, at some point, you’ve got to have cash generation. So that’s 100% possible. It’s 100% probable.

Look, I’ll just say it’s been said before: $1 trillion in spend commitments on $60 billion of revenue. And now you’re going to go to the public and hope that retail is going to pick that up.

Yeah, tell us about companies staying private longer and how unfair that is to the bottom half of society who don’t get to participate in it.

For those 99% who are mostly not us, right? Your 401(k)s—those retirement plans—can’t get into those companies now, which are getting bizarre exceptions to S&P 500 rules. All of the rules are being broken. The passive funds, the ETFs, are going to have to pick them up.

Where do you think we are on that curve of value creation? Could they go 3x from here? Sure, but—

So, just to say it as plainly as possible, we’re going to force overpriced products on the 401(k) holders of America who didn’t get to participate early. This is your position: that this is profoundly unfair, creates more wealth for the people who don’t need it, and makes people’s retirement accounts the bag holders.

There’s a lot of risk in that, and my objection is: don’t say you’re doing this for the benefit of humanity and do the other thing.

Make the public’s retirement accounts the bag holders.

Or just say, “This is how we’re running our business, and this isn’t for the benefit of humanity.”

Bill, do you think that what happens to venture—I asked Thomas this question—is that when these dollars get distributed, there’s going to be a handful of funds that have ginormous returns, I mean just unbelievably excessive? Founders two is going to print a $100 billion return on $200 million of invested capital. But that’s 1 fund in isolation.

Right.

Right. And there’ll be a few. Your funds when you were at GV are going to print an enormous upside. If you don’t look closely beyond the averages, venture’s going to look incredible. If you look past the averages, you’re still going to look extremely bimodal: a handful of winners and a ton of losers. How does that play out?

1, that’s how venture is, right? 75% of funds lose money. But 2, in order for Founders Fund—or pick any fund—to get that $100 billion out, they have to sell that stock to someone else. Otherwise, it’s just on paper.

So who’s the buyer for that? Is it retail? You’ve got to make a business case in the public market that can show that this business is worth a discounted value of its future cash flows. Whether it’s SpaceX or Anthropic or so forth, can that case be made? We’ll see 6 months after or so. I know they’re playing with the lockups to kind of drag that out, but we’ll see what the public market thinks of that.

Okay, so we have this 1 set of companies, and then there’s everything else. What do you like in the everything-else bucket as a venture investor?

I’m going to make an analogy to the gaming industry. We all think about, “What does the future look like when AI is everywhere?” There are doomers on 1 side and utopians on the other.

Zork?

That’s Zork. I’m going to get to that. Just bear with me for 30 seconds. It’s probably not as bad or as great as everyone says.

So, let’s look at the gaming industry. I used to play this game, Zork. There was 1 called Planetfall back in the ’80s, and it was very brittle. It was turn response, turn response: “Grab the lamp.” “Oh, I didn’t—it’s a lantern. I should have said ‘lantern.’” “Go north.” And you wait for the computer to respond.

Let’s show the most sophisticated retail-available AI system out there today on the next slide and tell me how different it looks. What’s happened to the gaming industry from the ’80s to today is going to happen in AI, but in the next 5 years. That will be compressed in terms of how quickly that change happens.

We would all agree that games are better today than they were then. They’re photorealistic. You can inhabit them, and they’re moving very quickly. On the AI side, there’ll be ambient computing. The problems that Zork had will be solved for AI: lack of memory, lack of consistency, session resets, and so forth.

To answer your question, I don’t plan on investing in larger models, right? Just like it wasn’t better stories that made better games.

It was controllers and physics engines and GPUs, and those are the parts of the AI cycle that I’m interested in—the platforms that need to be built to—

Machinery.

You’re correct. That is going to make this reality real in the next 5 years. It’s not just bigger models. I think we’re at the Atari command-line stage of AI, and we’re going to get to the PlayStation 10 stage in the next 5 years.

You also used to do a lot of stuff in life sciences.

Yeah. Not as much anymore. My interest in life sciences—I founded Calico and have been very interested in that space. We were investors in Flatiron, Veer, and lots of other companies. I’m very interested in that space because it has a dual benefit of helping people and also doing good and doing well.

Correct. However, the therapeutic space that requires human clinical trials is a specialist investment area that we’re not spending a lot of time on. I’m very interested in computational biology and those areas.

If you just look on X, there’s a renaissance happening in human health. I don’t know if that’s true, whether it’s cures for pancreatic cancer, cancer vaccines, or peptides. Obviously, there’s just an explosion, and a lot of it seems to come back to computation. But this class of winners so far is not really computationally driven. It was just really good science 10 years ago.

Yeah.

And so, do you think we’re about to see this massive—

I hope so. I started Calico, and again, it was fringe science—longevity—at the time. Now we’re investors in New Limit, which is Blake Byers and Brian Armstrong’s company, and a number of other companies in that space, which doesn’t seem so crazy anymore.

However, because of human biology and the FDA, if you find a compound and you think you’ve got something, that’s like 5% of the work. There’s still all kinds of titrating and safety testing that needs to go on, so I don’t think it’s going to go quite as exponential as we would all like it to. However, if we can achieve a realistic simulation of a human cell in silico, then you will see that accelerate as well. We’re not quite there yet.

But generally, we’re seeing what some might say is a flight of capital to India and China right now. Are you seeing that their biotech path to market is faster if you invest in firms that are based offshore versus the US?

The FDA has always indexed on human safety over speed to market, and that has cost us in some ways. However, some other countries are indexed in the opposite direction, which costs lives. So there’s a balance there, but there is certainly research going on in China and other places—experiments in cloning and all sorts of things that, as far as I know, aren’t happening here.

So, yes. I think the gutting of the CDC and the NIH, and the anti-science vibe that has now pervaded this country, has driven a lot of mind share elsewhere as funding is drying up for basic research.

China’s got their own paper clip model now. They’re recruiting some of the best scientists from Europe and India, and they’re all emigrating to China.

Yeah.

They go to do work, and that used to be a scientific pool that we used to access and recruit.

And we’re losing—we really need the neurological reserves here. And this business with—

Or brain trust would be another way to say that, but yeah.

Well, the pushing out of H-1B holders—there’s so much happening now that it’s causing people to go elsewhere. It’s just easier to go elsewhere. That’s not good for science.

What’s your view on what’s been called deep tech for the last decade? These are traditionally long-investment-cycle, capital-intensive, high-risk businesses. Elon is one of the few entrepreneurs who has successfully tackled a deep-tech business model with SpaceX and Tesla. Is this becoming a more tractable area for entrepreneurs to activate and for investors to invest in because of AI enablement, physics engines, and—

Absolutely, because things are moving so much faster.

What kinds of things like that are you focused on investing in?

Human biology and healthcare—that’s probably the largest TAM in the world. So I’m super interested in that. And then all the others I mentioned that underlay the AI revolution, which are the physics engines, the controllers, the GPUs, and everything that it’s going to take to get us there.

I want to bring in Sax and Freeburg before we run out of time, if that’s possible. Sacks, I’m curious about your thoughts on the venture capital business. I think you’ve done 5 Craft funds or 4?

Well, we’ve done 4 venture and 2 growth.

I’m assuming you’re going to be going back into the venture business. But I’m curious about your take. When you started in venture and when we started as entrepreneurs 25 or 30 years ago, this was a much different playing field. What are your plans based on Bill’s look at this? And do you believe in the $500 million fund sweet spot, or do you think you need to become Andreessen Horowitz when you go back to the private sector?

Well, I don’t think we need to become Andreessen Horowitz. But I think fund size determines fund strategy. The size of your fund—because you’re going to divide your fund size by 20 to 25 names to achieve some portfolio diversification and construction—will determine your check size, and that sort of determines where you play in the market.

The thing that’s spinning through my head after Tom’s presentation is: Are you better off just focusing on, let’s call it what used to be called late venture, early growth? You’re writing $50 million checks. You just kind of wait for the breakouts, as opposed to playing in this really noisy, super-early-stage game.

Well, I think the problem with that is we have to look at the incentive structure of venture. So, a $5 billion venture fund that returns 1.01× gets to say that they are in the 75th percentile and can raise their next fund, and no one at the Stanford endowment is going to get in trouble for writing that check. They need to put two or 500 million into a fund multiple times. So I understand that dynamic.

So now let’s look at the GP dynamic. If I have a $5 billion fund and I return 1.01×, I’m going to make more money than Bill with his $500 million fund that returns 3×. Okay? That’s also a strange incentive.

So now let’s look at the entrepreneur side. I am Researcher X from Open AI, and I’m going to start a company. Bill says, “I’ll give you $20 million at a $100 million valuation. I want to buy 20% of your company.” Giant Fund Y—we’re friends; it’s a different model—but Giant Fund Y says, “Well, we have this giant fund. We need to put $250 million in.”

Then an entrepreneur says, “Well, but my company’s valuation is $100 million.” “No, your valuation is now $4 billion, and we’ll give you $250 million for 1% of your company.” They’re going to take that deal every day, unless you’re a seasoned entrepreneur who has been down the road and knows the pitfalls of that.

And so, the incentives are broken in all those ways, and the pendulum will swing back. I don’t think just staying late-stage and waiting to snipe at larger companies will be a long-term strategy. The data would suggest that’s not going to work in the long term.