Bill Gates foresees massive AI job loss: these VCs disagree | E2330

Bill Gates published a 6,000-word essay today called The Turbulent AI Era Is Here. AI will, quote, either be the greatest equalizer ever invented or the worst source of injustice. And that right now, quote, there is no plan.

Some jobs are probably gonna get— go permanently go away.

Humans just kind of sort of get in the way. It is probably like a real thing.

I trust tech more than I trust government.

Every time somebody really adopts these tools inside of my firm, 3 or 4 new opportunities open up for our company.

This Week in Startups is brought to you Rippling. Don't settle for AI that's all talk. Head to rippling.ai/twist and get the only AI built to give you full visibility across your startup and take complex actions across your entire organization. That's R-I-P-P-L-I-N-G.ai/twist. Sign up for exclusive access today. Northwest Registered Agent. Got a new business idea? Northwest Registered Agent helps you bring it to life. Get a free domain, email, phone number, and more with no purchase required. Learn more at www.northwestregisteredagent.com/twistomaine. Lightfield. Name one person who's ever enjoyed updating a CRM. Exactly. Lightfield's AI agent does it for you. It even prospects and books your meetings. Used by thousands of startups. Free at lightfield.app.

All right, everybody, welcome back to This Week in Startups slash This Week in VC. Every Wednesday we do a VC roundtable, uh, 10 AM on the left coast. 1 PM on the right coast, and here in the great state of Texas, we go live at noon on Wednesdays. With me today, what a lineup, what an amazing amount of news. The docket runneth over. Shiloh Monat is with us. How are you doing, Shiloh?

Great.

Good to see you. You and I getting into it— minimum wage, jobs.

I love it.

Great debates on—

I love it.

Dave McClure, founder of 500 Startups, now 500 Global. He's got a great podcast About strip mining the VC industry called Trading Places. I just came up with that.

Trading Places?

Yes, he's using my podcast to promote his podcast, but I give him a promo for his podcast instead of paying him the $400 minimum wage for his SAG-AFTRA appearance here. He's with Practical Venture Capital.

We launched, we launched on your show a year ago.

Yes. And I get your shorts. Your shorts are great. TikTok Shorts are great.

He's talking about my video clips, not my pants.

No, he— Dave is known for wearing shorts, uh, in completely inappropriate settings like conferences, his office, etc. And with us, I think for the first time, Hussein Kanji is here. Hussein, it's the first time you're on the program.

Yeah, first time dialing in from London, so making it global.

Great.

And Dave, where are you dialing in from?

I'm in Sunnyvale, which is very sunny today.

Awesome.

Shield, I assume.

San Francisco.

San Francisco.

Right. And I'm in Austin at the ranch. Uh, lots to talk about today. 3 great investors on the program.

And I think we start out with, uh, Bill Gates.

Bill Gates published a 6,000-word essay today called The Turbulent AI Era Is Here: The Choices We Make Now Are Critical. Uh, obviously you have to hit exactly 6,000 words. That's what Jensen and Zuckerberg did recently. If you don't hit 6,000 words, we don't know if you wrote it yourself. I didn't run this one through an AI so we'll see if it's a drunken Miller or not. But he argues AI will, quote, either be the greatest equalizer ever invented or the worst source of injustice, and that right now, quote, there is no plan. Gates warns AI risks are outpacing its benefits. Interesting. Threatening jobs across law, medicine, and software. He's calling for a new national institution of some type and AI usage taxes. And making some jobs reserve jobs that can only be done by humans.

Dave, uh, you read the manifesto, or at least the summary of it. Uh, your thoughts on Gates coming in at this moment now? So from a meta-commentary perspective, why is Gates commenting now? And what resonated with you as true, and what resonated to you as false? Not true, or God forbid, a worse take.

I don't know particularly why now, although maybe Gates had some bad press that he's trying to get past in the last—

Okay, I like that theory.

Some reason. I think the 3 main points I kind of agree with. Some jobs are probably going to permanently go away. Bad actors get superpowers. That's probably true. Not just due to AI. I mean, drones are already seeing bad actors get superpowers with not too much money. And then kids and human connection. Uh, and I guess that might be topical today because Meta is just paying $16 billion in fines for, uh, I guess potentially getting kids addicted to, uh, social media.

Yeah, well, definitely getting kids addicted. And that'll be our second story. Shiel, when you read the missive, any thoughts on why now? Uh, or any thoughts on what resonated as true?

I think everything he says is broadly true. There will be a lot of job losses. I still in my head don't have an idea of, is this like, is this actually going to change everything or is it some subset of the population that, you know, gets job, gets their job displaced, jobs displaced? And it starts out with like the jobs he's talking about. Many of those were already displaced when we started outsourcing, whatever, 20+ years ago. In terms of why now? I don't know why now. I think now is as good a time as any. I think he talks about the enormous potential that AI has, which is awesome. I can't say I understand enough about what we should do about it to have a strong, fully formed thought. He proposes that the AI companies should contribute, and there's a democratic participation where everyone Everyone has a say in how this goes, which I think is a reasonable idea. I think we've been operating as if the AI companies can just do anything, and clearly the public sentiment is very bad on that. So I think there probably does need to be some change, and I think his proposal makes sense.

I think the public sentiment— I just saw something recently where it was mostly coming from environmental sort of impact, not so much the economic impact, which I think maybe is a little surprising. Probably energy and power usage are primarily on people's concerns. I still wonder whether there's job displacement or not, shouldn't we be thinking about AI kind of like the new oil? Shouldn't we be creating sovereign wealth funds with our surplus coming from AI and sharing that broadly? I don't think Bernie Sanders taking 50% is the right story, but Sam was trying to get in front of this and offer like 5% to 10%, I think.

So a tribute, as the world's jobs evaporate, Hussain. Any idea of how you would get a coordinated effort? Because Gates points this out in his missive, how we get the world coordinated in either reserving certain jobs, not using robots for certain jobs, which I think would be factories maybe, or healthcare he brings up over and over again, and then how to tax it. Like, this is a tragedy of the commons, I guess, where you would need, Hussein, everybody to agree, okay, yeah, we won't give these jobs over, but if my company decides on the slide and sneakily uses AI to solve the problem, now I've got a better profit margin, right? So is there any way to coordinate the world if we can't get rid of nuclear bombs or can't stop global warming, can't stop burning fossil fuels, to get people to stop using AI tools in some coordinated fashion? Is that possible?

I mean, the idea of coordinating global, like, like world, different world countries since like post-World War II, I mean, is there any evidence that we are good at kind of figuring out how to tackle any of the big problems? I mean, so like, you know, this is a human problem, not an AI problem. And maybe you give the keys over to the AI, maybe they might do a better job, like it might do a better job. But the general idea that automation's going to happen, AI's going to get better at the automation, human jobs are going to get deprecated, and as a result, well, like society, like the economy will be so much better run by automation that you actually, humans just kind of sort of get in the way. It is probably like a real thing. Like, I mean, I feel like that's kind of sort of playing out. And if you really believe this idea that like a dollar of compute equals, you know, $10 of intelligence. It's, I think it's only a matter of time before even the higher cognition things in society kind of get replaced. And then there is like a genuine problem. And this is not a new topic. I mean, like science fiction authors are usually like, like decades ahead of everybody else on this stuff. Like Kurt Vonnegut wrote about this stuff, Player Piano. It's like the first book. Cory Doctorow wrote about this stuff. And in his book it was Walk Away. Like the old idea, like in, in terms of coming back, but it's, It's manifesting. And I think almost everyone who's in AI has thought about some variation of this. Like, you know, Sam had the UBI, you know, initiative, like a decade, like 5 years. It feels like an eternity ago in our industry, but like, this is, I think, a very important—

Which notably didn't work.

Which did not work.

I think that's just implementation details. I think, you know, not giving people money, but giving people access to capital with some amount of debt repayment is probably a better strategy.

You think microloans like we saw, what was the microloan site that we all lost our minds over 10 years ago?

Kiva.

Kiva. That actually It worked, right?

Like, we— do we need to double-click on that again and go back to like microloans for microentrepreneurs?

Micro lending. I was, I was on the board of a company called Unitus that basically helped solve micro lending in India 20 years ago and was successful enough that now micro lending is a commercial asset class in India.

All right, I'm here with Keith Peres. He is the founder of Lightfield, and they are a CRM system built natively in AI that takes all your emails, calendars, Slack messages, all the stuff you keep losing puts it in one place and creates organization out of chaos so you don't lose any customers. Keith, welcome to the program.

Thanks for having me.

I got to ask you, I mean, it's 2026. I've used 3 different CRM systems in my life. The team hates them. Everybody hates CRM. How does Lightfield help me as the CEO founder get people to use a CRM system?

The first thing is, uh, your, your team won't have any CRM admin to do. They don't even have to log in if they don't want. Um, it'll update itself. They'll have to occasionally approve some updates. So all of the clerical work is gone. Second, it's gonna do work for your team on their behalf. So they're gonna spend way less time doing follow-ups, outbound prospecting, account health scores. Lightfield will do all of that for them. And then third, for you as CEO, it'll help you better forecast because now your forecasts are based off of sort of real unstructured customer truth, not just pipeline numbers. And it'll uncover new opportunities. for you around what product areas you should be in, what industries you should be in, and proactively give you feedback from your customers on where to take your business.

So everybody go check out heightfield.app.

Let's be realistic. I don't think anyone trusts us as techies to kind of figure this problem out. And then the problem is you look around, you look across the aisle, right? Everyone in government, like, I don't think any one of us would really trust And you— and I don't even think you can converge on decisions in government to figure this out and then make the problem like a global problem.

I trust tech more than I trust government.

Well, if you trust capitalism, if we look at— if we parallel AI and intelligence, job loss, and this technical wave, and we compare that to, say, energy, um, capitalism and demand and supply Innovation in the form of solar becoming cheaper and cheaper and cheaper and batteries becoming denser and denser and cheaper seem to have solved the problem. It's now more expensive to either retrofit a coal mine or build a coal mine than it is to build solar. Great state of Texas. We're the largest provider of solar in the United States. And I can tell you, the people here are not precious about their oil. They're not like, we're team oil, we'll never touch solar. They're, what is the best most profitable, you know, next energy supply I can create. And in, I think, 4 out of 5 times, it's going to be solar unless you can build the data center right on top of a NatGas field. And so that would lead me to argue, well, maybe there's going to be a lot of displacement in jobs. I think we all know that's going to happen, displacement.

Sure.

But what I'm seeing is the people with these tools, every time somebody really adopts these tools inside of my firm, 3 or 4 new opportunities open up for our company, and the ability to go faster and solve more problems happens. So this harkens back, Shiel, to the quote that I think Jensen got right. You're not gonna have your job replaced by AI. You're gonna have your job replaced by somebody using AI.

Right.

That's what I see in the field in white-collar work. Now, for blue-collar work, Shiel, I think it's distinctly different. Watching what I'm watching, and I have only a matter of time. I have a lot of investments in self-driving car companies, in all of them essentially, from Zipline to Uber, you know, anything that's public, I, I, I dipped in. I don't know that a person who learns AI is suddenly going to deliver more burritos or do more airport jobs. I think they just lose their job, right, Sheel?

I think that's probably right. Now, that also was true like 120 years ago in the United States. Everyone was a farmer pretty much, and now we're not. There have been other jobs where things have been displaced. I think the timeline— we talk about it, and even Bill Gates talks about it as if it's instant, but it's really not. Like, it actually— these things will take time. And, uh, your DoorDash driver is probably safe for the next 5, 10 years. And in that time, by the way, like—

I don't think 10 years. I don't think—

Okay, but 5, you think? So we're between— I said 5 to 10. You take the lower end of the range?

Yeah, I think, you know, we're seeing job displacement in knowledge industries very fast, but like physical industries—

Where are you—

you're not seeing it in the data?

Yeah.

Not yet.

Not yet. I think—

I think—

You don't think like the jobs that we've been doing knowledge-wise are much easier now with AI?

Definitely easier, but we're not seeing the job loss.

Definitely.

Yeah. That's the question.

But we have data on this and it's not happening yet. I'm sure it will. I agree with you, but it's not—

We're still, we're still in the early innings of a lot of this. Oh, absolutely. I think the thesis question is not that this is, I mean, you could argue this is not going to happen and jobs will get created, or you could argue that this stuff compounds so well that it becomes so much more competent than any of us needing the jobs. And then the question is what to do with us. Like, yeah.

I mean, I think hiring in tech industries has definitely been like slimmed down a little bit.

It's flat.

We're not hiring.

Yeah.

We're not hiring as fast as we are, like, increasing productivity, which is arguably a good thing.

But, like, I think we're going to see— Well, I think that's the nuance between both of your positions. When Shiel says we haven't seen the job loss and you say we're seeing massive productivity, both of those are true. And if you look at a company like Microsoft or Google or Uber or Facebook, in many cases, they have the same relative number of employees over the last 5 years, plus or minus, let's say, 10%. And if you were to do that, uh, Claude search, or you did a, you did a search on like employee count per year, you'd probably see flat across them, and you would see earnings growing 30, 40%, and top line growing 10 to 20%, which is extraordinary. And that actually is AI making the people working in companies more efficient.

You're gonna see massive net positive benefits for small business and small business creation due to AI. Like, I think actually the biggest beneficiary of AI will be A whole bunch of sole proprietors and small businesses.

Totally. All across.

Why? Unpack it. Why, why will we see a lot more? I, I think I have a guess.

Because they will move fastest. They can move fastest. They can be creative. You know, the benefits of AI tools to someone who's running a small business is probably like pretty immediate. Like if you learn how to like get customers better, do your job better, faster, you know, you get the gains immediately.

They accrue immediately.

Sure. There will be AI software companies That help people run their sole props much better. So like, if you're a therapist, you used to have to have a receptionist, you had to have somebody schedule, like all this stuff, billing. Now an AI can do all that stuff for you, which does speak to job loss, but also speaks to sole props being able to function much easier than they could before.

If the virtual— One person doing each of 5 jobs and now they're doing like 10 jobs with AI.

Yeah.

If the, yeah, I mean, simplistically, if the virtual humans are better at doing the jobs than the actual humans and they're a lot cheaper and you can put, you can, you can deploy it thousand of them like overnight versus having to deal with interviewing, hiring, et cetera. Anyone who hires people is going to probably benefit from this thing. That's a simplistic way. The big question to me though, going back to this, the big question, like what happens if humans are deprecated, right? Then how do you build a market-oriented economy to figure out what the heck to do with the humans? Because you probably can't get an autocrat to figure it out. You probably can't get government institutions or global institutions to do it. So how do you build a market-oriented economy to figure out how to solve this problem. Like, this is— this— the problem stuff is like kind of trivial to me. Like, the solution stuff is like the real intellectual— that's the hard—

Let's, let's pull that string. I, I'm going to agree with Dave and Shiel on sole proprietorships booming, but not just because the existing ones are going to be run better and then they maybe make more profits. So hey, the person who's the therapist is like, you know what, I just want to have an assistant to manage all the AIs and keep that train going because I'm so much more profitable, but I don't I don't need the second and third one, right? I don't need the outsourced accounting firm. We can just do that in-house. So there could be less people per sole proprietorship. But if you can't get a job coming out of school at Uber, Meta, Facebook, and Microsoft 'cause they're not hiring, but you have been using AI to cheat your way through school and you're an expert at it and you know how to get, you know, straight A's.

Wait, you're talking about my kids right now.

I've got to stop going right here.

It is what it is.

It is what it is.

Like, I— as far as I'm concerned, if you learn how to cheat on your way through Harvard, and then you're perfectly positioned to use AI to actually create a company because it's the same tools. How do I get this task? Here's the goal. Go get the goal. And you just motivate it, set the loops up, et cetera. So I think we'll see a Cambrian explosion in sole proprietorships. Because you're just not gonna have 4 job offers. People graduating from college are not gonna have, like, my— unless they happen to be a PhD in AI, but they're not gonna have people competing for it. But let's talk about your question, Hussein, which is, what are practical solutions?

I have one that I've been talking about a bit. Um, people tell me I'm a libtard when I bring it up, uh, because God forbid we even consider humans and employment in this occasion. And, and and giving them— you know, it's just complete weakness to have empathy for somebody who's an Uber driver or a DoorDasher. But in China, um, they are now giving out licenses for self-driving vehicles, and they have stopped giving them out. So they have a cap essentially right now, uh, because in Wuhan, in Beijing, in Shanghai, and the markets where you have 3 or 4 different people flooding the zone with these, Young men who are not making as much as drivers or can't get a driver job are getting restless. My prediction is the socialists in the country who are rising up right now and saying, hey, we're going to defend workers, we're going to make sure you have a more affordable life. 2 ways to do that. Cut prices, which they can't do. That doesn't work. Second way, give workers protections through unions and minimum wage, et cetera. they can successfully do in all likelihood, or it's an easier path. I bet you we see AOC and Mondame and Boston and DC, LA just say, you know what, if you want a self-driving car license and you want to deliver burritos, deliver burritos, and I'm talking against my book right now, you just need a license. The licenses are auctioned off at $30,000 starting price.

I think that's literally what we're going to have in the next 2 years as people respond to it. Do you have solutions, Shiel?

In the case that we had a quick job displacement occurring, a brisk one as opposed to a manageable one. Okay, so you've identified a real problem and you put together a solid solution and a business model that you believe in. So you're all set to launch your new company, right? Not so fast. If you want investors and potential customers to take your new business seriously, you need to consider forming a Delaware C-Corp, and that's where Northwest Registered Agent comes in. They're going to give your new company a real identity. That means an address for your public filings, a domain, a custom website, a business email, and of course, a phone number. And that's going to take just 10 minutes and 10 clicks. They don't charge hidden fees. Customer service is available around the clock. They're not overwhelming your inbox with spam, and they make it easy to cancel at any time.

So get all the advantages of a Delaware C-Corp independent, regardless of where in the US you're operating from.

Visit northwestregisteredagent.com/twist for more details, and the links are in the show notes.

I think that's a reasonable solution to say. So there's a, there's a license for a self-driving car, and, um, there are externalities that, that develop from that. Now, there are huge positives that come from that, like you have more trust, you, um, you have fewer accidents. Self-driving cars are amazing. But, um, I do think that it's totally fair to say, uh, to society that we need to tax these. And I, I, I don't see immediately a bad, uh, that, that's a bad plan.

Dave, any thoughts here for if we do see any kind of acute job loss, as Bill Gates is predicting, ways society can manage it? Do you think that we should be taxing consumption of tokens in corporations in some way? Should we be looking at the people providing the tokens and saying, hey, you know, when you sell hotel rooms, we put a 12% tax on it in New York City, whatever it is, and that goes towards, you know, whatever infrastructure.

Do you think there's a— there could be a time where there's a federal token tax of 5% or 10% that goes into the unemployment pool?

I think that's a great idea because token costs are going down to zero, so you're going to be taxing a diminishing return asset. I do think Gates is correct that right now incentives are probably, you know, because you can depreciate infrastructure and CapEx spend, you're more likely to hire, you know, agents and things that you can, you know, depreciate. So we probably got to figure out some way to make that a little bit more level playing field. But I think what we're really trying to do is how do we recreate the New Deal program in an AI era where, like before, you know, Tennessee Valley Authority, building the highways, doing all that stuff back in the '30s and '40s and '50s made sense. But these days, I would say probably got to think about more human-oriented stuff. So maybe some kind of care economy for seniors and people, human services. Maybe we have an advantage in providing emotional support and care for people over agents, at least for the short term. Probably not to sound too liberal, but hey, arts and culture.

Yeah.

If everything is— jobs are being taken away in productive industries, maybe we create cultural art that people actually enjoy.

I don't know.

I think it's going to be hard to create hundreds of thousands, millions of jobs if those get displaced. But I do think we need to start planning for The equivalent of, you know, retraining budget and/or support budget for people who are going to be displaced. Like, it's not going to go well if those people feel like there's a small percentage of humanity which is empowered and rich and a large percentage which is displaced and poor. That's going to end up looking like France in the 1800s.

Dave, the China model feels like it's trying to stymie progress, right? You're trying to put a barrier in.

China has some big problems. Like, we're taking a group of folks to China in October. About 50 people are going to go to Shenzhen and Shanghai and take a look at what's going on. But China has significant unemployment problems for the youth.

Like, but, but if you—

let's, let's assume if this thesis is correct, you don't want to stymie it. You actually want to accelerate it.

I agree.

I don't— Get the automation to be, to be as fast as possible. Because remember, since the Industrial Revolution, we have largely, like, lived to work. You know, there is a theory where we may not have to do that anymore and we could go back to kind of purposefully—

Completely agree.

Especially if, and then like, it's very Star Trekky, right? You know, the stuff just happens, but you want to, then there's going to be a 20-year period, maybe a 10-year period, like a decent amount of time where it's going to be awkward as heck because the jobs are going to dissipate. It won't be as offset. But once you get past that funding trap, like—

I don't think it'll take 20 years.

You think it's longer?

I think it'll take 20 years. I think the big benefits of AI, like 3 big benefits that I do think will happen is we will get to free energy, we will get to free food, we will cure cancer and major disease. Those things are definitely on the horizon. But the distribution of those benefits to all of humanity will be incredibly uneven. And that's where I have concerns. Like, if we solve all those problems, but like 2 billion people on the planet are still, you know, hungry and don't have access to power, That's not going to be pretty.

I, I agree with you on the power one. Um, I recently had somebody on This Week in Startups who has robots taking panels, driving them out and installing them, and then goes back to the factory, grabs another panel, installs them, and there's a human like watching it occur and just making sure, yep, that bolt's good. And I just said, wait a second, we, we're gonna have a 24-hour factory. Like, Elon's gonna pop up a solar factory. in the middle of a desert, and it's just gonna go out in all 360 degrees and just lay solar panels. And then we just have to obviously get the hookups to go to where it's needed, um, which then would— of course, Dave, from our time in the Middle East—

would make for free, uh, unlimited water if lunatics don't stop desalination because they're concerned about like the guppies.

So you have free energy, You have free water, free water plus free energy, and a bunch of excess Optimuses.

And Elon showed me one of the latest Optimuses. He's not showing them publicly anymore, but he just happened to show me on his phone, and I was like, whoa, that's moving faster than what he showed me 6 months ago when I visited the Optimus lab on a Sunday and it was packed. He's going to make a billion of those. If a million of those were out there, you know, planting strawberries. Like, the price of strawberries is going to be a dollar a pound, 25 cents a pound.

It's going to be nuts.

Shil, your thoughts on the benefits?

I agree.

I, I mean, I feel like we don't talk about the benefits enough.

Yeah.

And all we talk about, sort of like what Dario says, everyone's gonna lose their job or whatever. Um, I think taxing tokens doesn't really make sense because you actually want to promote the use of tokens when it makes sense. But I think it is fair to say, hey, if these companies have enormous excess profits that concentrates wealth, then tax the excess profits. It also makes sense to say, like, hey, if we believe that there's a particular type of harm that's happening and it creates costs, we should tax that. So, like, you know, the robotaxi, maybe there's a per-mile tax, totally reasonable. And then I think if people talk about the energy usage or water usage, I don't think is real, but people talk about energy usage, which is real. And I think if it's consuming a scarce resource, then we have the right to charge for that scarce resource.

So I think you're right about the water usage. There's more water usage by pistachio and almond farmers than there are.

Totally.

It's all vultures. But people still People— I don't understand.

But the power usage is real. Power usage is going to raise bills for people unless we subsidize and take care of power.

Which anybody building a data center has the brains to know that they should just go to the community and say, not only are we not going to increase your electricity, how does free sound? Free for 5 years. Sound good?

But that's basically like we're doing redistribution on a piecemeal basis. And I think it's probably just better to get ahead of that and say, look, Let's take 10% of Anthropic and OpenAI and solve our national debt problem.

Wow.

I mean, they probably will be able to.

So you're proposing just the government takes 10% of the companies? Is that what you're proposing?

I know that sounds anti-capitalist, but I actually think it's a reasonably good—

I mean, Trump is doing that all over the place. We have— we have positions.

Not in a very normalized way either. He's putting it into a pocket.

He's doing it with threats is what you're saying, Dave.

Like, literally, he just threatens people. And gives them the carrot, they give their shares. I mean, how is it different though?

Norway—

let me ask you guys a question though, Dave.

How is it different the way Trump uses his carrot-stick method to extract shares for the American people because he likes doing it versus the socialists saying, hey, you got to give us 20% of OpenAI and Claude to pay down the debt? How is it different? I, I just—

Let me give you 2 extreme examples of how that can go right and how that can go wrong. So Venezuela and Norway both have extreme oil reserves. One country has done an extremely good job of providing for their citizens and their wealth and preparing for the future. And one nation has not.

Rule of law?

I don't know what the implementation details were that went different directions.

Rule of law.

It happens to every business at some point. You're nearing the end of the month, or maybe it's the quarter, and your sales team is at risk of missing their quota. You could waste days pulling together data from across the entire organization, Or you could just ask Rippling AI. Rippling's already built on your real-time people and business data, obviously. So Rippling AI can pull metrics from both Rippling and your CRM into a meeting-ready dashboard. See exactly how close you are to hitting that quota, break down your revenue by region or sales rep or more, all in just seconds with Rippling AI. Look, you're running a company, you're busy, you have critical questions that you need immediate answers to. And you can't just file a ticket and wait a week for your team to get around to sending you a report. And you don't wanna sit around waiting for reports. You need an actionable insight right now. So if you're ready to rule your business, head to rippling.ai/twist to get the only AI built to give you full visibility and let you take complex actions across your startup.

There's some really interesting, you know, historical reasons why Norway went that way that actually had to do with the shape of the country and how the timber industry developed. Actually, they developed hydropower in Norway before oil was discovered, and that was an actual resource. Like, Norway is a country of like 4 or 5 million people that became a timber and shipbuilding powerhouse across Europe. And I think the things that they did early to like develop timber and hydropower actually went very well once they discovered oil.

Fascinating.

What do you think, Shiel, about this?

If the government owns, let's say, 10 or 20% of Anthropic and OpenAI, first of all, how do you decide which companies they own a part of? And then is there a regulatory challenge? Is the government going to then be easier on the regulations? And I think that's probably why Sam Altman is proposing it in part. The government would have a financial incentive to protect their valuations and, and go easy on them. Um, so I'm not sure that that makes sense to me.

Uh, all right, let's pivot to another technology that caused a lot of social harm and a lot of hand-wringing, finally coming to what it seems to be closure for parents and Americans. Meta is agreeing to a $17.1 billion settlement with 29 states over claims it hooked kids on Instagram and Facebook. The remedies are well beyond just writing a check. This is the most intense settlement I have ever seen for a technology company, and I think the previous record was set by our friends at Meta. So you get this huge settlement of $17.1 billion, but In addition to this, they are going to put hard limits, uh, similar to what China has done and other countries are considering. Productive pauses interrupt scrolling after 15 minutes of continuous use, then again at 60 and 90. This is a self-imposed, uh, feature that Instagram will give to kids. They will also have a nighttime block—

no access from midnight to 6 AM— to make sure kids get, uh, sleep. And then no push notifications during the school day, 8 AM to 3 PM.

They're adding age verification.

They're redesigning the features that they believe, or that studies have shown, will cause harm to kids, like beauty filters. Oh my God, the fact that they even added those. Visible like counts. And then add stronger safeguards against bullying. And they'll obviously be, as there typically are with these, independent auditors. Uh, It would have been nice, Hussein, if Mark Zuckerberg and— who's our guy running Instagram?

Mosseri.

Adam Mosseri.

Adam Mosseri, as I believe both parents, looked at this data 5 or 10 years ago and said, we'll do this without 29 attorney generals bringing us to court and embarrassing us and shaming us. But here we are. It's getting done. Hussein, what do you think of this unprecedented settlement?

Long time coming. I mean, I think the harms are pretty clear in the academic literature, and your parents definitely notice this thing if you have, if you have kids. And you're starting to see other governments actually crack down. Australia was, was a big one. The real question is, is it— is all of— are all of these changes really enough? Like, I, I'm in the view, like, you might actually not want like prepubescent teens and maybe teens to be using some of these things. You might really want to gate these things like this might, and this is a reasonable, and this is a big number in terms of a fine. That's a small, small, small portion of Meta's market cap. Like this is, this is a drop. It's a small portion of like Meta's free cash flow. Like it's, it's not that big of a fine.

Probably 10% of free cash flow. I'm guessing they have $200 billion.

I mean, Meta stock is up.

And Meta stock's up probably larger than $17 billion is if I'm going to guess it went up more than 1% on settlement. So it paid for itself.

And you've tabled the problem.

Jason, first of all, so my wife works at Instagram as a lawyer, but, um, disclosure, um, you're wrong actually. Instagram instituted this stuff 2 years ago.

Okay.

So they actually have teens, uh, have sleep mode from 10 PM to 7 AM. There's an automatic notification if you spend 60 minutes on the app. They did a lot of this stuff. The, the difference in the settlement is it's now across apps. So it's Facebook and Instagram and I believe WhatsApp, but, um, they actually self-instituted this, this policy a couple years ago.

Got it. So we can give them a little bit of credit. Um, give them some credit.

Um, I, I think, I think that they have— and like the AI world, I think social media has totally bungled the PR. I actually think there are so many benefits from social media that are never talked about. During COVID I was depressed. I'm an extrovert. I love talking to people.

You are.

Social media cured me, I think, like, from depression. I, I, um, there's so many outlets for creativity that teens have that they did not have before. So I think, I think there's a huge positive that we never talk about. The studies show some negatives, um, which I think are real.

Curious, do you have kids?

I don't.

Got it. Okay.

Uh, Dave, you have kids. I have kids. Our kids grew up on this. What's your take as somebody with kids and watching up close and personal what that generation experienced with these tools?

I have a really funny story about this because one time I was in Palo Alto having brunch with some friends and my kids were there. I think they were probably like 3 and 5 at the time. And one of them was playing with an iPhone, one of them playing with an iPad. I looked over at the table next to us And Steve Jobs was sitting at the fucking table right next to us. And at the time I went over and said, hey, you know, these are actually really fun tools. My kids are learning how to like, you know, draw and read and do some other stuff. Thanks. And he sort of said, you know, you know, you're welcome, whatever. Later I found out he didn't allow his own kids to have access to phones at that age. I would say we've had a mixed experience, I think.

Unpack that.

At least with one of my kids.

Oh yeah, sure, sure.

Um, privacy is—

Tell me about average kids, you know, that you—

in Silicon Valley.

Tell me about average kids.

I live in Silicon Valley. I don't know what average means here because it's like completely weird and different. But, you know, I think there are some great things about Silicon Valley. We live in a very diverse community. Like, my kids went, you know, to school with people from all over the world. They got to experience a lot of other cultures. But, you know, everybody had technology all over them. Everybody had very high expectations, you know, One of my kids went to a pretty high-pressure, you know, private school where, you know, a lot of negative things happened to kids. Some kids actually committed suicide. Many of them had stress.

Yes.

Related from tests and things. So like, it's a mixed bag living in Silicon Valley. I kind of agree with Sheila. I think there are both pluses and minuses, but there are definitely some minuses. And particularly, I think when kids are, let's say, between 12 and 17, there's a lot of social pressures and challenges. I mean, without technology, kids can be assholes to each other. Um, and, you know, sometimes that's amplified not because of the technology, but because of the people who are on the technology. Uh, and so bullying and shaming and like—

Social pressure, body issues. Yeah, it's all contained in that window.

That exists in real life. It's not like the technology is what's causing it.

Well, it amplifies it, I think, to a level that is superhuman. So the same way, you know, you could become famous and go viral and help your business or, you know, get people to read your novel, whatever it is, all those positive things. Or if you're feeling melancholy— I can't believe you would be depressed, Sheil. I'll just go with melancholy. If Sheil hit a level of melancholy and, you know, talking to people online and, you know, having never-ending threads debating social topics got you to be less melancholy, I'm for it. But I just had a hard rule. The kids were allowed to have a certain number of hours with the controls Apple built, but no social media until 17.

And my oldest daughter's 16, and we're just— she'll be 17 shortly, and we're just doing read-only accounts, read-only accounts, because she's really given me the full-court press. She's literally saying, I'm the only one, Dad. I'm the only one without this now.

Yeah.

And I look at—

What are her friends posting?

Um, most of her friends are great kids, um, and they're posting pictures of their outfits and their friends goofing off and nature and food and, you know, anything that we would all post. But it does start you down a, how many followers do I have? How many likes do I have? And already kids are coming back to the phones and back to the iPads, and this creates the same addiction you and I have, Sheil, with Twitter. Yeah, you and I get pulled into it.

It's not just kids, it's us. Like, we have these problems too. Yeah, it doesn't stop at 18 or 21.

Yeah, but the kids—

I bet you median time in the bathroom has doubled or tripled.

100%. 100%. Yes. Hussein, go ahead. You want to add to this? Do you have kids, Hussein? And do you have a—

Do I have a 4-year-old? Uh, so much younger. But these kids are just so much more sensitive to it, right? And they're in like these growth, the, the, these development cycles where the number of likes, and if someone leaves a mean comment, we're more resilient than they are. Like we've lived through more and we've had to deal with this stuff. And maybe the argument is kids have to go through this stuff, but I think shielding kids from this stuff is actually a good thing versus a bad thing. I would, despite the—

Yeah, I, I would love to see Apple and Google put this into the hardware level of the products in America. So if you're a parent and you buy an Apple or an Android-based phone, you're kind of bought into the same— you know, both of those ecosystems allow protection of kids at a very root level. And the controls in Apple's products were extremely confusing and complicated.

And after hours and hours of managing them, I— after like a year, and I'm a technologist, you know, I've kind of got a grip on what I want to do. As an example, I'm into audiobooks. I like to listen to them. I'm an audio listener. I like to hike around the ranch. I got the girls into Audible. They're addicted to Audible. I put Audible— you can listen to Audible 10 hours a day. Now we were having like the sleeping thing. I let them put a sleep timer on. They can listen to an audiobook. They put a 30-minute sleep timer on. Of course, we'll read them stories, but they're kind of getting to the age where that's not cool anymore, 10 years old. They're like, I don't want you to read me a story, which is sad, but they do want to listen to something when they go to bed. And I'm like, hey, if I can get them addicted to audiobooks, that would be fantastic.

Yeah.

And so I'm I'm starting to have the controls work in my favor. And if you want a child to behave well, if they've got a device and you can say, after, after these chores are done, go to the— clean the chicken coop, take the dogs for a walk, yada yada, then you can earn more time on the device. You, you'll be amazed, Sheila, when you have your kids, how quickly chores get done when the iPad does not work. Because you remotely turned it off. Oh my God, they just have to come to Dad. And I'm like, okay, you were arguing with Mom? You were disrespectful to your mother? And you didn't— oh, she asked you 3 times to empty the dishwasher? Okay, great. Let me know when the dishwasher's free. I'll turn on your iPhone again. Problem solved. All right, let's get into venture capital, unless anybody wants to add anything there, but I think we beat it to death.

One funny thing I saw, Taylor Lorenz posted a thread of all of the past moral panics, and there were studies done around novels that said they were bad for teens. There were studies done around jazz music saying they were bad for teens. Radio, television, of course, bad for teens. And actually—

Dungeons and Dragons.

They're all remarkably similar to the stuff that we talk about social media today.

D&D is definitely detrimental to your dating life for sure.

Yeah, dating life. But it's also good for your Creativity. You're definitely not going to get a lot of dates for that. Yeah, she's got a point, except I don't think we've ever seen anything in humanity as addicting as social media. I'm trying to think of— cigarettes and social media seem to be the two, and I guess sugar and fatty foods, right? And so if you were to think of those 3 that just came off the top of my head, The damage they've done, the damage that, you know, uh, processed foods have done to America.

Tobacco, alcohol, and carbs.

Tobacco, alcohol, carbs.

And then you add to it the mental health issues with screen addiction. That's a pretty big, uh—

those are the 4 horses of the apocalypse for humanity. Um, hey, question for you guys, uh, is there anything left to fund outside of AI right now for us?

Has anybody Looking through your last couple of investments, was the company— 'cause there's 2 things: a company that uses AI to build a company, you would be foolish not to, that would be a sure sign that your company's gonna fail. But then a company based on AI providing services that are AI-related and/or powered by AI. Is there anything outside of AI that anybody is funding?

Uh, I think that's an extreme statement, but it is probably at least half of the VC brain, if not 2/3.

Yeah.

But, you know, let's, let's say there's all these new sexy categories— space tech, defense tech, robotics, AI— are probably like the hot, hot and sexy categories. I still think fintech is like a big category.

Yeah.

And it's not, you know, overfunded, particularly if you get outside the US and a lot of emerging markets. I would say fintech and e-commerce are still big categories and are still growing. They're not sexy in the venture world. At least not in the US.

I was an investor in a fund and I got a distribution for this Nubank, N-U-Bank. And I'm like, what's a Nubank? And it was like this tiny little amount of stock. And I'm like, well, if this legendary venture firm found this company, I'll just hold onto the stock and see what happens. And then all of a sudden it became a very large amount of stock over a period of time. Um, yeah, for sure, still much room to grow. What do we think, Hussein, of Stripe buying OpenRouter? Those are 2 different things— tokens and web hosting. And—

Oh, it's more than web hosting. It's like the— it's the information layer. You kind of know exactly how the token spend's kind of going through the economy. And I mean, I think these are— these— it's an interesting strategic purchase. We all kind of scratched our heads with YouTube. Well, we all scratch our heads when YouTube got bought. We all scratch our head when WhatsApp got bought, like maybe a little bit less so for Instagram. I think if you're leaning into AI and you're the plumbing system for the payments, being leaning into AI and knowing where the spend is actually happening, how these models, like it's actually strategic. The why I think we're going to figure out probably 5 years from now, we're going to kick our, like, look at it and be like, that was such an obvious big purchase. Like, yeah, I think you give a lot of credit to founder-led businesses who can kind of think 1 or 2 steps ahead of this.

I like how you're framing it as like Stripe's infrastructure, OpenRouter's infrastructure. Infrastructure respects infrastructure. That, that could be the beginning. I never thought of that. Um, that makes sense to me. They— what do you think, Shiel, when you saw that purchase?

I mean, yes, Stripe routes money, payment methods, banks, et cetera. OpenRouter routes inference, basically. And, um, I think broadly it makes sense. I think the price Is wild.

Um, and what did it go for? $12 billion?

$7 billion?

Oh, $7 billion.

$7 to $8 billion. $7 to $8 billion something.

And it was making $100 or $200 million at the time?

I think that's—

that was rumored. I don't know. I don't—

I don't think it was more than that. I thought it was a little—

it was still 50 times revenue, 100 times revenue, something crazy.

Yeah, because they had just raised at $1.3 billion earlier in the year.

Got it.

70 times revenue.

70 times revenue.

Okay.

Well, sorry, that was based on the $10 billion number, so I guess less than that if it was something in that range.

Okay, so between 50 and 100 times. It's unbelievable.

It's clearly a bet that neutrality ends up— like, that you need to route, which I think I see no evidence that that isn't true, but people are building their own. Ramp launched their own router.com the same day that the deal was announced, which is crazy.

Uh, yeah, I mean, hey, it's been a fucking great week for Martin Casado at, uh, Andreessen, man, between— yeah, or in Cursor.

Uh, well, I mean, a $60 billion Cursor deal. I mean, explain how unprecedented that is, Dave, in terms of the last 20 years you've been a venture capitalist and 12 I've been one.

Uh, I think Cursor was the biggest M&A Ever, $60 billion.

Yeah.

Andreessen Horowitz and Thrive, I think both invested roughly $60 million, $50 million. They both made— Andreessen got $6.6 billion back on $44 million in, 150x return in a couple years.

And a big cash-on-cash number. So 150 times multiple might be small compared to some angel deals, but it's on a big number.

Right.

Exactly.

Some of the angel investors even better. Like I think Neo, Ali Partovi, I think put in $1.5 million and is getting, I don't know, billions back, something like—

Yeah, nobody could ever challenge Y Combinator, could ever compete with them. And the Y Combinator people attacked him savagely. And I wondered why they did that.

Because he was a legit threat, probably.

He was good at what he does.

But notably, Cursor didn't go through the accelerator. It was an investment outside the accelerator.

Right.

But I agree with your point though.

Emma, what did you think when you saw that?

There are also investors in Replit. I think they did go through.

What did you think, Sheila, about that? Like the Y Combinator trying to put him out of business?

I thought that was— that whole thing was weird. Why come out with this personal stuff? And you, you make claims that like, oh, he's a bad person. We know the story, but we're not going to tell you. Yeah, I don't like all that stuff.

That's the YC playbook. They did it to me too.

At one point, I can tell you from experience, it's a very different thing when you are not competing with YC and are competing.

They did it to you too, Dave. They went after you.

I was pretty close with a lot of people before I started 500. I even emailed PG and Jessica before I started 500. And then as soon as I left 500, things got very warm again. It was an interesting experience.

Yeah, they're circle the wagons. It's so lame though to be swinging your elbow like that and trying to— it's like there's plenty of— if you accept less than 1% of startups, Then you should be happy that the other, let's say, 9% in the top 10% have a place to go. Because I can tell you, Dave, you and I both know the business pretty well. Do we know the difference between top 1% or 2% or 3%? And does YC?

Obviously not.

Not until 7 years later.

Correct. So what are we talking about here? Like, you could be kind to each other and supportive. I'm super happy for him.

I think you know entrepreneurs of all kinds are competitive, and you know you're probably very familiar with you know how Elon looks at people. How about before a year ago how Elon thought about Dario, and this year now that Dario is a big customer?

I think he re-underwrote his position on them after talking to Chamath. Re-underwrote his position on Trump. It's a lot of re-underwriting that occurs in our industry. All right, let me just take a pause. for the cause right now. I love using AI. Obviously, we all do. And then I found out about this great product called Harmonic. What is Harmonic? It's a database of every transaction, every investment in venture capital, every profile of every company. And I've been getting into robotics because when I was in Japan, I started seeing these robots fighting each other, and I was like, tell me about these. And it very quickly, as you can see on my screen here, told me who— which ones are the ones being used at all these different fighting competitions. But then I was like, you know what? I want to have this type of search going on, like an associate or a researcher at my firm might do. And so I've empowered my team to create scouts, just like the Sequoia Scouts program. I was the first Sequoia Scout. And they named it Scouts. So I said, hey, find me any robotics companies that are pre-Series A. And if they went to Carnegie Mellon, MIT, Harvard, Stanford, get me that list as well. And now it runs it every Monday for me. That's the power of Harmonic AI. If you are in our business and you don't have Harmonic AI, you might miss the next Uber, the next Micro One, the next Cursor. Give it a shot, folks. Harmonic.ai. Lots more to get to here, but let's talk about physical AI, robotics, defense industry. Uh, Hussein, have you jumped into it yet? Are you monitoring it? Because it feels like frontier models— obviously that investment opportunity, uh, seems to have manifested itself fully. There might be less opportunities in backing a frontier model, but feels like we're in that moment for robotics. And still for the— for military tech, still seems pretty nascent to me, especially on the, uh, implementation side.

These things aren't in our homes yet, aren't really in the factories, and if they are, they're kind of kludgy.

So your thoughts on physical AI and just physical real-world stuff being the moat now?

That's what we're hearing from a lot of investors.

Hey, there's no moat, but hardware used to be hard, so we didn't do it. Now hardware is still hard, so we do it so we have some kind of defensibility.

It's amazing how this industry has transformed from being anti-hardware 5 years ago to very, very, very pro-hardware across the board. You know, I am convinced that the future is like, like it's the everything language related is done, like, and every derivative of the language stuff. It's the frontier models that we all kind of know, OpenAI, Anthropic, and then, you know, people building on top of the stuff. I think it is still up for grabs who's going to win transformers AI for physical and transformers AI for biology. And it's no surprise that the 2 spaces have kind of gotten hot. And there's a bunch of that stuff in Europe, which is kind of good news for us because that's where we kind of invest. And we've done one stealth one that came out of Imperial College and a bunch of people are citing their papers. I can't talk about it yet because they're going to launch in about like 2 or 3 months, but it's super exciting. It's like a new frontier model for robotics that kind of upends the figures and the generalists, et cetera, of the world with a different technique that requires a lot less training and makes it much more efficient. So big believers in this stuff.

Um, Shil, any, um, investments in the area so far? Are you double-clicking on it? What are your thoughts? I mean, you're early stage, so it's a little bit scary to put— I'm assuming your average check size, $250,000 to $1 million, am I correct? Or do I have all the different—

Um, we're probably like $2 million average, $2, $2.5 million.

Okay, so that would burn through like the prototype in 9 months.

Yeah, in a hard way.

It is hard.

In general, these AI companies take so much money. We have not invested in anything in physical AI. I am super excited about what I've seen. I don't know if you saw the demo yesterday from Scaled AI. They announced something. They have a robot that you show it flipping a pancake and it'll flip pancakes perfectly every time. They did a really cool thing where the demo had The— they cut off a dog's leg and like a robot, a robot dog's leg. Oh, sorry.

And this is a robotic dog.

It's learned.

No dogs have been harmed in the creation.

No dogs have been harmed.

So we're clear.

They cut off a robot dog's leg and show that it just learns how to walk on fewer legs, or they cut off part of its leg and it shows learning how to walk The demos are amazing, and I can't freaking wait to have a robot in my house doing my chores.

Uh, and this is from Skilled AI.

Um, Skilled AI. Yeah, here is the Carnegie Mellon spin-out.

Yeah, I mean, that's why I put that into my search. Here it is, folks. Let's take a look. Skilled AI person has, um, some gloves or What, a harness? I guess a physical harness. Here's the pancake being made. Here's a plant being planted, and they— you can see here the human has devices, cameras on each wrist, and it's studying the human doing planting, you know, moving things from pots and flipping pancakes. And lo and behold, this is like the scene from Planet of the Apes. Dave, you remember that famous scene where it was like, do, and they were like putting the forks and the knives into the, into the proper trays? And, uh, it does look like this is going to be the end of chores and human—

Or your daughter is never— always is always going to have her iPad.

I mean, basically she's going to be like— she's going to be doing to the Optimus or the skilled robot what I do to her?

She's gonna be like, I'm turning off your internet access until that dishwasher's empty. And the robot's gonna be like, okay, I'll empty the dishwasher. But this feels like it's getting close. Yeah, it's getting close.

Yeah, it's compounding at a rate which is like exponential. And like, I, I think this is— I mean, I think if you're— you can take the long view and be like, it's 10 years away, but I think it's a lot sooner.

I mean, watching this, I think the training data is happening at such a fast rate. So, Shiloh, your thoughts on the compounding nature of what we're seeing right now? And when—

I mean, as these things get better and better, it's— I just can't wait to have a robot in my house. And I would easily pay way more than it costs to make one of these things. Like, is a robot worth $100,000 to me?

Easily.

Like, if it's cleaning and cooking, it's worth $100,000. And what are these things going to cost to make? Not that much.

$30,000 is, uh, I think what the first Yeah, what do they say, bill of materials or, um, BOM?

Bill of materials.

Bill of materials. I always say build. Bill of materials.

BOM.

The BOM, I think, is like probably $15,000, $20,000. You put in some extra pricing, I think they got to be $20,000 to $30,000. So I think they're the cost of a Toyota Prius.

So think about what that unlocks for humans, and it's just incredible. Like, at that price point, it's incredible. Yeah, it's, it's going to be like Everyone— not, you know, unfortunately not everyone's gonna be able to have one immediately, but like, even for the average person, at $20,000, $30,000, it actually pays for itself pretty quickly. We have Rosie the robot.

Unbelievable. Unbelievable. All right, listen, while we wrap up here, SHIELD is a beast on Twitter, and we're gonna play a new segment. It's a new segment. I was saying it's called SHIELD's Hottest Takes. Take number 1.

Let's see.

You disagreed with Chamath's all-in that Silicon Valley has lost the idealism that made it great and that people only care about money now. You suggest, quote, most likely a function of who Chamath surrounds him with.

Am I supposed to take this personal?

What do you— what do you think of this take, Dave? And then you'll end it up. We'll end it with you, Sheil, defending your take.

All right.

That is below my line.

It's below your line.

Okay.

Just saying your take on the. I mean, is it too much about money right now and a little less Steve Jobs and you know idealistic? We're going to index the web, Larry Page. You know, you know, it did feel.

Valley is always about the money, but it's also always about the innovation. Jason, you were probably around. Well, you may not have been. Were you here in the Valley in the late 90s?

You can.

No, I was in New York still.

Yeah.

But I mean, I would come out.

out 2 or 3 times a year. But, um, yeah, I mean, it did get—

There were a lot of people who thought in the late '90s it was just people here for the money, and a lot of them were just here for the money. And then it blew up in 2001 and a lot of them left town. And then again, yes, you can— that happened probably 2 or 3 times. So like, I don't, I don't know that it's wrong, it's just not the whole story.

Um, the true believers versus the get-rich-quick people is, I think, a trend you are correctly identifying, Dave. And we are, I think, tipping into the get-rich-quick kind of like, hey, can I get in here?

Go ahead, Hussein.

And I was going to say, I mean, like in the '90s, there were a bunch of tourists who came in. They were definitely there to make a buck and they left. But like, we did not see seed rounds and Series A rounds of secondaries where founders were taking a ton of money off the table. Pretty much like, like life-altering, like not life-altering money by today's like trillion-dollar company standards, but life-altering money. And it's like, Like millions of dollars.

I mean, if you break 10, it's life-altering.

Yes.

5 or 10, it starts to become much more coin-operated than it used to be in the past. Like there's a, and I think there's a bit of a negative halo around some of this stuff because I think that voice has become much more amplified even among the most technical of founders and the most innovative of founders. There's like a, it's become a business in a way where it wasn't as much of a business 20 years ago, 30 years ago.

It felt more clubby, that's for sure, 10 or 20 years ago. So, Sheel, um, you defend your take? You still defend your take?

Yes.

Or were you ass-posting? Be honest, were you ass-posting?

No, no, no, no, absolutely not. I think, I think that— so Chamath said Silicon Valley lost the idealism that made it great and only cares about money. Are there people that only care about money? Absolutely. Has always been the case, and it ebbs and flows. On the whole, The people doing the great work, like my friends at Anthropic and OpenAI, could give two shits about the money. Like, they're doing it for the mission. I think like this—

Easy to say when you pass a $100 billion valuation. You're—

yeah, that's right. But, but also like their lives haven't changed. Like, they haven't changed the way they live their lives. And what are they— they could, they could easily just say, hey, I'm done, I'm retiring. But if you look at the, the number of people that continue to do stuff, it's not strictly money motivated. It's because they're excited about the mission. And I think I've actually— I didn't see that idealism 5 years ago that I do see now.

It is a peculiar group of people. I will say Anthropic, because of the leadership having all self-selected out of the Sam Altman experience, which is the cutthroat experience in their minds and the Like, the money people are gonna make the decisions. You know, code monkeys, code monkey, but leave it to Sam to, you know, take the credit. This is not my interpretation. This is what people say. Like, Sam's taking the credit, Sam's doing the deals, Sam's the frontman. That got a little tired, I think, for that group of people. Anthropic is like, oh, we're a bunch of monks. We're a bunch of pious, like, we're doing this and we're giving all our money to the Anthropic Foundation and we're gonna cure cancer and— But you're gonna lose your jobs. They seem like headcases to me, a bit on the margins, like really delusionally, uh, committed to being the last company on Earth. It is a little weird.

All right.

What about Sam not having any equity in OpenAI?

I don't know if it's true.

I mean, I think we're missing the— we're missing the point. He's got a lot of money probably in nonprofits and philanthropic entities that he controls. And like, who's to say whether that's, you know, money in his personal pocket versus money in his philanthropic pocket? It's fuzzy. People need liquidity. And, you know, obviously talking my own book here, but I just wanted to share a few interesting charts.

Oh, sure.

That might be relevant. I've been doing this presentation for another talk. This is like comparing $100 million private rounds versus $100 million IPOs that crossed over around 2017 when SoftBank Vision Fund came out.

Got a little nuts in 2016 is when the flip happened. The blue line is $100 million private rounds. The red line is $100 million IPOs. And we see them bifurcate in 2016, just so people listening understand.

And now it's 10 to 1. Last couple of years, the number of private $100 million rounds, 400 per year. Looks like we're going to continue on that track. Number of IPOs of $100 million, only like 40 or 50, right? And I think that's what we've seen is the IPO market now is requiring half a billion to a billion dollars to go public. It's not that it's closed, you just have to be a lot bigger. That results in companies being private a lot longer. The other graph I will show is this one, which is the tender offer. And this is corporate tenders or internal secondary offers for their employees. That is a booming business, somewhere between $30 to $40 billion a year now. And, you know, growing probably north of 50%, at least north of 30, 40, 50% per year.

Yeah. And here we're seeing 2023, there were $6.5, 2024, $12, major jump from '24 to '25, $12 to $27 billion. So just over 2x, and then $27 billion to $37 billion into the estimated 2026 Tender offers, which for people who don't know, that's, uh, some investors coming in and buying common shares from employees.

That it's, it's corporate organized secondaries, primarily for the employees, sometimes also for their investors.

Early investors who have been patiently—

For a lot of companies, for a lot of companies like Databricks and Stripe and Kem and others, these are happening on almost an annual basis now. Those numbers are probably distorted a little bit. OpenAI did their own $7 billion secondary.

Who was the buy side on that $7 billion? Was that Kushner?

It's secondary firms. It's the primary round VCs who want more equity when they're oversubscribed rounds. It's retail market and family offices, at least for the popular names. But it's kind of interesting because outside the top 50 companies, there's not as much awareness of those companies. And I think this is what's going to start to happen is these companies are staying private for 5 years longer than they used to. Employees there want to buy a house, want to put their kids to college. They want liquidity. They're sitting on a ton of, you know, unrealized equity, but they can't buy a house in Silicon Valley. They can't, you know, pay for schools and everything.

Well, and their concentration, Dave, is the real issue, right? Like, you're basically 99.99% of Shiel's friends at Anthropic, their wealth is in one stock.

Well, okay, one more, one more graph and I'll quit here.

Um, oh, we love your graphs. Keep them coming.

Here's my segmentation of that market. In that top category used to be SpaceX, still is Anthropic, OpenAI.

Now it's like a trillion.

Stripe. Well, but as soon as OpenAI and Anthropic go out, then the remaining ones will be like Stripe, Databricks, Maybe Revolut, Anderol, a few others, I guess ByteDance if you consider—

Right around 100. Yeah, some of them just below, some just above.

So that will—

Huge amount of dollars there. But like, look at the rhinos and stallions and maybe the donkeys. These are like where the majority of the number of tenders are. They're not like huge numbers. These are like $10 to $50 million tenders. Um, amount of market cap that's being sold here is probably somewhere between 2 to 5%. Got it.

So they're trimming very little of their cap table for very large numbers.

Right.

And, you know, let's just assume these companies are probably growing, you know, north of 20%, 30%, 40% per year. It's not at all unreasonable to say you would, you know, do 5% every year in sort of secondaries.

Yeah.

Back to employees. This is really the new IPO market. This is like where liquidity is coming from for employees. It's why real estate agents in San Francisco make a ton of money and car dealerships. And, you know, private schools. But, but the challenge here is it's still a pretty concentrated story. And, and I think we're just going to see more and more of these corporate tenders happening on a regular basis, if nothing else, just to be competitive. You know, if you, if you have the choice to go work for a company that has an annual tender offer and one that doesn't, which one you're going to choose?

Uh, yeah. And who's the loser here, uh, Hussein? Who, who loses in this new private marker tender offer? In your mind.

And retail, retail investors, right? Retail investors have to wait until these things are super expensive. And you're— and in the old days, Dave, like in like the '90s, like you take these companies out like really, really early, like at market caps of $100 million, and employees would have liquidity almost immediately, like after the lockup.

And retail could buy those shares. More importantly, a civilian could own Microsoft at $100, $200, $300 million.

Well, they can still do that, but the difference is now it's a fuzzy story because when those companies go public, they report their numbers There's quarterly transparency in what's going on. You have a $2, $3, $4 trillion middle ground market where these companies exist before they go public, but their financials are not disclosed to anybody that's not an insider or a primary investor. And so all the secondary trades that are happening on markets, Hive, Forge, Equizend, Augment, all these other brokers, there's no underlying corporate financial data on most of them.

Right.

Everybody's flying blind and buying that on vibes.

Yeah, it's vibes.

Yeah, totally vibes.

Uh, and the buy— yeah, the buyers who are buying it, um, are hearing people talk about these names, and when something gets a certain amount of PR, then it looks great while SpaceX and Anthropic are going through the roof.

But I guarantee you, as soon as that market turns around, it could be like the SaaSpocalypse, right?

Like late-stage VCs, they suspended disbelief, they bought Airtable— no, no dig to Airtable— at $10 billion, And it didn't get there. They got their money back.

We could have a lot of those. We could have a lot of those. Like, if we— I mean, I don't know how many unicorns there were back in '21, like, you know, 600 plus. Like, how many of them are real, right?

Like, we were just talking about Headspace just got bought for $200, $300 million, but the last private mark was $3 billion. Headspace got bought down 90%.

Who bought them? I'm still a large shareholder in, uh, Calm.

I did take advantage of secondary twice during that opportunity on the way up for my LPs, But who bought Sword Health? Plans to acquire Headspace in a cash transaction. Wow.

And by the way, this same issue exists for a lot of seed funds. I think this is the big challenge for a whole bunch of fund managers is they're sitting on unrealized marks. They may have a fund that's at 3, 4, 5x TVPI, but DPI is 0 or barely 0.5. These funds are getting to 10 years and there hasn't been 1x return back to investors. Yet. And I think we need to start figuring out how we're going to get liquidity for these funds.

I, I really— if I wasn't doing what I'm doing right now and so busy, I would do a roll-up vehicle. I would raise $500 million and then just go straight down like, um, Bending Spoons is doing.

Bending Spoons.

But I would do a private Bending Spoons and then just hire a bunch of AI kids, developers, give them like really crazy bonuses for hitting crazy profitability and just buy every Headspace slash whatever is out there with— maybe Headspace has $50 million in revenue and they get sold for $100 or $200 million, and then you can get rid of 80% of the staff. You can do what Elon did at Twitter, you know, get rid of 80, 90% of the cost, and now you've got a wonderful business. Um, I remember meetup.com. My friend Scott Heiferman had started that company. Kevin Ryan wound up buying it, and then it went to Bending Spoons And I had heard a whisper that it was making $15 million a year, but then it got bought for $20 million. And I was like, wait, how's that possible? It doesn't make any sense to me. And I don't know if that's true or not. That was just a whisper I heard. But I think if it's not growing, VCs just don't want to waste their time. But what an incredible opportunity.

Well, there's some stuff that's growing, just not fast enough for venture returns.

Yeah, that was Airtable, right? Airtable was still growing 20%. And people were like, well, this is a waste of my time. It's like, really? But okay, uh, I guess we'll see.

It all depends on— it all depends on price, right? And at the end of the day, well, it, it's price.

And but I think more importantly, see, this is what I thought that too, Hussein. Then I realized it's not price, it's, it's opportunity plus price. You have a certain number of things mentally you can keep in your brain as an investor, probably 10, uh, you know, like as in board seats. And you're like, well, if I have—

I would say 3.

Okay, but you know, I'm being generous here. Let's say it's 8. Let's say it's 7. So you got 7 and you're like, hey, these 3 are filling my space with a 20% growth company, and it's hard, and every conversation is hard and a bummer, and we can't get out of the mud for 4 years. And I could give those 3 seats to, I don't know, this robotic company making pancakes.

You know, it's just you only have so many, um, spots on your dance card.

I think that's—

Yeah. Shiel, you're nodding, I think. Yeah.

No, I agree. I, I think there's a, there's a question of like, are these companies ever gonna go public? Like, why would Stripe go public ever? Like, they have their unlimited capital.

They want to grow, sell their capital, you know, why not? They don't have to.

They have liquidity. If they want to acquire PayPal, it seems like they can because somebody else will give them money to do it. So why would they ever go public?

Mike Bloomberg never took Bloomberg public.

Different. He owned it almost entirely.

Sure.

80, 90%.

Like, because he didn't have to solve for like stress Stripe gets this tender thing that Dave's talking about, solves for the liquidity question, and the private markets have grown so much that they don't need to go public to grow.

Bloomberg was a technology company. You know what he did in New York? He was famous for just paying 50% more than everybody else. If you worked at the New York Times, you made $100,000, you were making $200,000 working for him. If you worked at the Washington Post and you went to Bloomberg, you got a $50,000, $75,000 raise, and you didn't get any equity.

But right now, I think there's an arbitrage opportunity here where these companies have been priced on sort of venture-scale projections, and now some of them are profitable but not growing that fast, and they're going to be priced on free cash flow projections, you know, not just multiples of revenue or growth. And that's going to be a different number, very different.

I mean, if you have a currency like Elon did, I think Cursor might wind up being You know, this generation's YouTube acquisition. You've got a—

that's a great buy.

That was $2 trillion and they brought in $3 billion. And what was their runway?

$3 billion, $4 billion at the time?

It's, it's at $4 billion now and it's on its way to $10 by the end of the year. But like that, that's not even like the total value. It's like, you know, to be, to be kind, I would say xAI was a shell of a company that was mostly valuable because of Elon infrastructure and CapEx. That Elon bought. It wasn't because of the tech or even the people who largely walked out the door after that acquisition. But, you know, turns out that all that infrastructure was very useful to rent out to, you know, Anthropic and Google. And then Cursor was really the value that came in. And I think that combination— like, you got to give Elon credit, the massive pivot that he's done in the last 6, 9 months with that company, right? He's gonna have a line of business that is the majority of their revenue that did not exist.

Yuan Web Services. Yes.

That is just mind-blowing.

All right, let's end on agents. Uh, I have been obsessed with GrokBot. Just does what you tell it to do. Shiel, you have been obsessed with another agent which has a landing page that looks like it was built in 7 minutes by a Python developer. And I signed up for it, and people are losing their minds over it and the terms of service. Tell us about this, um, new OpenCLAW abstracted into just iMessage that people are losing their minds over. Tell us everything.

Yeah, so I, I've been using GrokBot too. Um, I— this, this company is called Instinct. Um, it was funded at seed by Conviction and A, I believe, by Kleiner Perkins. Not announced, but that's what I've heard. Um, also, I've also heard crazy valuation numbers like seed at $100, uh, A at $500, and I heard they have offers, uh, at the billion-plus range. Um, what it does is it's basically like your assistant. You can— yeah, that's the website.

Yeah, this is worth a billion dollars, folks. Here's your— yeah, I mean, could we even just put it in a No background?

No?

Okay, sounds good.

Um, you can text it and it just does stuff for you. Um, what's cool and I think valuable about it is it actually like goes the next step beyond what you would think to do, you would think a bot would do. So, um, I— if you ask, uh, ChatGPT Work to book a flight for you, it will take you to the landing page, Google Flights. Here's the flight to take, Instinct will actually go through and book it for you. And they're able to do that because they have a vault where they store your credit card information. Um, and they're probably violating all sorts of terms of service. Um, but hey, the thing works.

Um, I signed up for it.

I saw the terms of service debate. I paused. I was like, I'm not sending anything until I find out what's going on here. GrokBot has been amazing for me. A bunch of the things I wanted to do with OpenCLAW and Hermes, I call it Hermes 'cause I think it's funny. 'Cause it's a harness and they make bags.

But anyway, a lot of those agents in ClawCode, just, they had too many restrictions on what they would do and not do. I took a clip of a YouTube file and I said, analyze this into the 5 most important clips. Make me 5 clips. And it took this Mearsheimer, John Mearsheimer video about tactical nukes, whatever. He's been on the Olin Pod. It made me 5 clips. I was like, whoa. And it's like, would you like one of these vertical with captions? And I said, sure. And then I shared it on my Twitter. And I was like, well, that's a $50 an hour video editor, you know, freelancer—

Totally.

That I would have done that back and forth with. Instead of taking 3 hours, it took 3 minutes. And was essentially close to free. And then I asked it to do some things on LinkedIn and it just did it because it has a computer window that abstracts into cursor and it kind of works all the time. I think that product, if it goes multiplayer mode, will go parabolic. And by multiplayer mode, I mean, can I just add 3 team members to it? as comment only? Can I add 5 team members to it as read only? You know, like the Google Doc suite. If they add multiplayer mode to that, I'm gonna— I would have taken the docket creator for today and shared it with you guys, and we would have all been there talking with the docket. Really interesting product. Have you used—

anybody used GrokBot yet? And any impressions?

I'm using it too.

I use them all. Like, I mean, I mean, which one sticks?

Which one is sticking most?

Which one do you go to first?

OpenCLaw was the least durable. Like, I switched over to Hermes, and then GroqBot became the next one. Then Instinct, I just got— it has a signup list, and I'm not cool enough like you guys to get access to it.

Dude, I put my name in. It was instantly— it was like, OK, you're on the list. Boop, you're in. I was like, OK.

OK, somebody knows where the bread is buttered.

So I'm now working on that one. But it's also amazing. Like, a year ago, none of these things existed. Like, like OpenClaw was like a year ago, they were like notorious. Like, yeah, yeah, yeah. It's like the, the pace is like super fast again. These things are compounding.

I gotta say what Sam Altman— I mean, listen, I don't, I don't wanna beat up on Sam here, but what he did to OpenClaw was so sinister and cynical in my mind. He, he like buys the guy, brings him internal, and then, you know, Dave Morin like is trying to keep this foundation going. And literally, the second the, um— his name is Peter, I believe. I've never met him.

Yeah.

The second Peter went to OpenAI, OpenClaw just lost its voice, lost its momentum. Everybody was like, well, this is over. Whether it's true or not, I've never seen a project go from everybody embracing it, talking about it, to completely toxic and people being like, yeah, you know, I gotta, I gotta go find something else. I gotta find something else. And I don't know if that's because they thought it was part of OpenAI and they didn't want to— that, you know, they were open source people— or better options came out. I don't know. What do you think, Sheel?

Yeah, I'm not sure what happened there. I, I think to your point about GrokBot being like a— we talked about this— having 100 million users, I, I think that relies on the other guys not catching up, and they actually already have. Like, ChatGPT Work already yesterday announced new stuff that's doing A closer version of this.

Show you guys one more graph.

By the way, Shiel, is OpenAI work basically what Peter is probably working on there? And it's basically OpenClaw inside of OpenAI?

Totally. Yeah.

I mean, what did he get? I think he got $300 million, Shiel.

Yeah, I think it was—

I—

what is he worth?

You heard that? I don't—

I thought it was somewhere between $200 to $300.

That sounds right.

Oh man, but maybe we get a $300 million valuation. Or 400 million. That might be triple. It might be a billion dollars right now. What do you think, Hussein?

Just one guy.

Just one guy.

That was never disclosed.

I mean, the rumors were like a crazy number.

Sorry, you know what? That wasn't. That was the other company that was acquired.

The podcasting bros was two experiments.

No.

You remember the?

Oh, there was a social media experiment that they bought. Oh yeah.

What was that? No, no.

But I think Peter got the whisper number on Peter was two or three hundred million in equity.

Yeah.

And that was when it was a $300 or $400 million company. If it goes out at $2 trillion, that could be 5x, 6x.

Anyway, I just want to point this out.

Remember how crazy it was?

Go ahead.

The OpenClaw hype. The OpenClaw hype was so crazy that there was a, there was this Multbook, a Facebook—

That was the one that I was talking about. Multbook got acquired. I think that got acquired for like $250 million by Meta.

By Meta.

Yeah.

Right.

Yeah.

And there was like the OpenClaw religion, like, because someone planted that seed, like, it was like there was a euphoria around this thing and then it just like Died.

Like, it was about one month, the whole thing, start to finish.

Well, I think the, the hype was maybe overplayed, but the story was sort of for real. I think that was like original agentic sort of adoption. But wow, I just want to point out this graph. This is kind of like crazy because—

Explain the chart.

You got to explain the chart, Dave, because people listen.

Okay, so there's a little story here in that CFO, uh, Sarah Friar at OpenAI, 4 months ago, she was like, Sam's like, we're going to IPO. And she's like, we're not ready to IPO. And then just recently, this was leaked from their internal conversation. She's like, we're definitely going to IPO in 2027 and maybe sooner if our growth keeps inflecting. And I was like, what's she talking about inflecting? And I was like, this graph is what she was talking about. Look at those numbers sort of like go through the goddamn roof.

Yeah. And what we're seeing here is OpenAI active agent users in Back in January, $200,000. Breaks a million.

Just, just a month ago. Like, that growth went through.

But it hits 5 million in May, and then 6 million in July, 10 million later in July, 15 million in the beginning of August, 20 million towards the end of August. So you're talking about 4x-ing in 2 months.

Right.

Unbelievable.

And I don't have the graphs in front of me, but I know that, uh, it was reported that OpenAI revenue was growing faster than Anthropic this month.

This is after they had quarter-over-quarter slowdown. Hmm, interesting. All right, listen, another amazing episode. This Week in Venture Capital, a special edition of This Week in Startups, an offshoot, whatever. Hey guys, best investment. What's your biggest portfolio win these days? Sheel, tell us. When'd you get in? What is it? Which one is just gonna sprinkle all that beautiful DPI on your LPs? Tell us everything, Sheel.

Tell us where hope is.

Uh, where is your hope tied to? It's a hard job being—

I would say a real DPI or hope. Those are two different things.

Selling hope, baby.

Yes, hopium.

Tell us about your hopium.

I would say recently one is called Basis. It's an AI for accounting, really automates the work of a junior accountant. If you think about Harvey or Legora for law, um, and you've seen the growth in those companies This is that for accounting. We led a seed. Keith at Khosla led the A and Accel led a B recently. And the growth that I've seen is amazing. And companies, these accounting firms are really using the product and it's been amazing.

Fantastic. And you got in at seed. I think we are investors in the Uber to your Lyft or your DoorDash, whatever the analogy is, Tax GPT that we incubate. So yeah, I agree. This is going to be a fantastic space for both companies. What do you got, Dave? What's what's making you?

We talk about what we're going to get DPI and what we hope to get DPI.

Yes, yes. Which one do you when you talk about it put an eggplant emoji at the end of the sentence? Which one?

Well, I guess because I worked at Founders Fund sixteen years ago, I have carry in Founders Fund too and. And when we started our second secondary fund, I rolled in about $1 million of GP commit in that fund thinking, oh, this will perform pretty well.

Oh, yum yum.

Turns out that was all SpaceX.

Fantastic.

Well done. It was our best investment. So that went really well.

That's a career maker.

But the things I'm actually excited about probably are not generally on people's radar. We spend about 10% of our capital in Latin America. And one of our best performing investments, I may have mentioned this on the show before was a company called Motu, and they are a motorcycle manufacturing and small business lending company for delivery drivers in São Paulo, Brazil. And that business is on fire. It's doing over $300 million in revenue, profitable, growing 60%, 70% per year. And it's very, very reasonably priced. A lot of LATAM is actually on sale.

I'd like to meet him. I'd like to meet him. Yeah.

And as Sheel knows, you know, we used to do a lot of investments all over the world in 500 Startups. Now that I'm doing secondaries, most of it's in the US, but at least some of it we're doing emerging markets. A different deal that's not a typical secondary for us is in a company called Equitybee. And we recently put money into that because they are a platform sort of similar to secondary marketplaces like ForgeEquities then, but they do employee stock option forward contracts. And so again, getting back to those employees who are sort of locked up and want or need liquidity. before the shares are actually available to them. They can finance and exercise and hold those shares and get some liquidity. And really helpful for people who are leaving companies who might otherwise lose those options. equitybee.com, B-E-E.

Amazing. Well done. Hussein, who do you got? Who's making you feel optimistic these days in the portfolio?

This year, this month, we did one in the AI space for material science called Cusp AI. We were the inception check. We wrote them a $10 million check at a Great. Well, what at the time felt like a crazy number, it's not a crazy number in hindsight. It's worth $2.6 billion. Kleiner just marked it up 2 years in, like almost a fund returner for our fund 3, like in 2 years. Absolutely feels pretty good.

Congratulations.

Not our most interesting though. Like our sister company to that in many ways is like the same type of thing, but a drug discovery company. So AI, AI for biology called peptone.com, which like The AI, this stuff really works. And like, we have a, we have a drug against prostate cancer that might actually stop prostate cancer, almost entirely generated by the computer. And like, we're going into clinical trials. So like that, that to me is like the one that's probably going to generate the ton of DPI for us. But both, both are great.

Like, absolutely fantastic. Congratulations. I'll just give you 2 quick ones myself. Uh, we love all of our startups equally, but you know, sometimes we see, um, things get marked up and this is This Week in VC. So, um, just on the markup trail on-prem, uh, and Sovereign AI became like a big deal.

But we had incubated this company Apicus 18 months ago. And so for banking and healthcare, etc., they take this Go One server, put it into your facility, and they help you build your own large language models, harnesses, etc., and keep you from having to give all that data to a Frontier or risk leaking it, and they have a waitlist for their product, and that is always a great sign. Second one, Micro One. I met this crazy founder, just high energy, you know, reminded me a bit of Travis, and he was using AI to figure out who the top developer was— top developers were. It's a pretty interesting product. I think we invested when it was $10 or $15 million, and then he pivoted into AI training, This week he announced he had $500 million in training revenue, everything from legal, etc., and that they also hit a $4 billion valuation. So that is a fund returner times four, three or four, I guess. But of course, long way to go. Still more work to be done. Dave, Sheel, Hussein, great job on the show, and we'll see you all next time. Bye bye.