Perplexity CEO：Micron将在价值上超过Meta，以及出口管制为何帮助了中国而非伤害中国

1. From Lower-Middle-Class India to a $20B AI Company

Ready to go? Aravind, dude, I am so excited that we got to see this. We've done one remote, and then we did one at Founders Forum last year. So, thank you so much for joining me in person.

Thanks a lot, Harry.

Dude, it's a weird start, but just roll with me on it. I ask this of the best founders that I meet: are you motivated more by the fear of failing or by the thrill of winning?

The thrill of winning.

Why?

Because I have nothing to lose. I came from nothing. I never even imagined myself doing all this. My life has already been extraordinary, beyond any level of imagination.

I was just in India, doing my undergrad and training neural nets with graphics cards that people in the labs were using to play video games. It was all for fun, and my path led me all the way here. For my mom, just getting a job was success, because we were financially lower-middle-class in India, which is not even like lower-middle-class in the UK or the US.

From there, all we wanted to do was get a job at Google. Being an engineer at Google was considered a win. So, I'm already doing remarkably well compared to the ambition we had as a family. There's really nothing for me to lose.

2. “Attack, Attack, Attack”: Aravind’s Founder Mentality

That's why, anytime I try to act like I'm trying to avoid failure and be on the defense, I remind myself that that's the stupidest thing to do. It's better to go all in and try your best. Be on the offense all the time. Attack, attack, attack.

When you review, then, what are you not being aggressive enough on today?

Maybe in the early days, we'd be very, very loud on social media, talking about Perplexity versus Google, and I used to do that myself a lot. Some people don't like me for having done that.

Today, I'm a lot more measured in how I talk about our products, our competitors, and stuff like that. But it's not a lack of aggression. It's just that it's boring. People have already heard that enough from me.

Do you regret being so bold in your messaging?

No.

So, it's not a nuance and maturation of the message. It's that that's stale and you need something new?

Not just that. I don't think it's a relevant framing anymore. We worked on search. Perplexity started out as search. We built the first answer engine in the world that people know Perplexity for even today. If you mention the name Perplexity, people will think, “Oh, that's an answer engine.”

3. Why Perplexity Forced Google to Change Search Forever

We built a lot more things after that. We built a lot of agents, browser agents, deep research, and Computer. We built so many products after that, but we're still known for that first product. The mark has already been made.

We changed the roadmap of Google. You could argue that I, or the company Perplexity, changed google.com more than any product manager at Google has ever done.

Make that argument for me.

Well, nobody ever wanted to ship an answer engine at Google. Nobody wanted to tinker with anything on the interface that made them $250 billion a year. Now, you look at AI Mode, and it looks exactly like Perplexity.

There's not even any difference: the font, the citations, the specific bolding of inline text, inline hyperlinks, suggested follow-ups. The whole experience literally looks like Perplexity, except it's still not as good.

Is that bad or good for you, that they learn from you and adapt?

It's both good and bad. I knew, around the end of 2024, that this was going to happen, so it never caught me by surprise at all. It was just a matter of time.

I'm still surprised that the quality is not there, because I regularly test every product out there. But I'm happy that, honestly, they changed Google to be what it should be.

I believe that the frontier is where the money is. The frontier in AI is not about answering questions anymore. It's about actually going and doing work for you.

We still have the state-of-the-art deep research in the world, and that's actually where people subscribe to pay for our Pro or Max products. It's not for getting answers in the traditional way. They're asking for sophisticated research reports. They're asking for agents that go and do things for you.

We wouldn't have been able to do all that if we were sitting in 2024 thinking, “We have everything settled here. We're good and comfortable.” No. The answer engine was always lead generation for the frontier products we built.

You need something, right? Think about it. Every company needs to have one successful product to build the next set of products. In AI, nobody can sit comfortably thinking they have it all sorted out. That includes Anthropic.

If Anthropic thinks Claude Code is already a win, in 6 or 12 months from now, they won't even be around. It's an uncomfortable fact about the whole field.

Would you argue today—you just told me before we started that you think OpenAI isn't ready for an IPO—would you have believed you'd be in a position to say this 2 years ago, when nobody wanted to deal with any product other than ChatGPT?

Think about it. Anyone, even in such a massively advantageous position, can be in a position where they're no longer the kings. They're fighting from behind, right? So, that's the state of the field.

It's less about Perplexity, Anthropic, or OpenAI not having moats or having moats.

Can I push back on that?

Yeah.

I would have stood by it 2 years ago, even when they were dominant, and they are still a dominant consumer product. I would stand by it because I don't think they are financially ready. When you look at the balance sheet of that—

Okay. Maybe I'll decouple that. Let's decouple that: financial readiness for an IPO versus the perception of a dominant leader.

Do you perceive them as a dominant leader right now?

Yes.

In what?

Consumer search.

Well, except there's no money there, right? Because it's been commoditized. It's always a lead gen. For example, why are they going all in on Codex? Because that's where the money is. We're doing the same on Computer. Anthropic is doing the same on Claude Code.

4. OpenAI, Agents & Where the Money Actually Is

Google doesn't yet have a product in this category, but I'm sure they're going to come after that. Meta is trying to launch Hatch for $200 a month. You see what's happening, right? Nobody has—

But there has to be more money than just code, Codex, and Claude Code.

It's not about code. That's the main thing. The money, at least in non-advertising—I'm not talking about advertising revenue—is in subscription or usage-based revenue, is in whatever is at the frontier.

Today, the frontier is about going out there and doing things for you.

Do you not think, then, that that will be a $100–$200 billion advertising business for OpenAI?

Yet to be proven. Let's work through the categories of advertising. Who's the number-one advertiser on Google? Amazon. Who's number two? Booking.com. Number three or four, I think, is Expedia.

How much do you think Booking.com spends on Google? $16 billion, something like that. Some crazy amount like that.

How do you book your hotels or flights today? Do you book it on ChatGPT, or do you book it on Google?

Google.

Why is that?

For me, actually, I like discovery. I would like to see the options.

Exactly. The interface is less about conversations and more about exploration. When the decision-making is more subjective and vibes-based, you don't need an objective answer engine.

You think about the other category of advertising: direct-to-consumer products and fashion. Where is most of that advertising budget going? It's going to Meta and Instagram, because you're just browsing. You're just doom-scrolling, or whatever you call it.

The chat interface doesn't capture that user intent, that user behavior, right now, which is why it was never a great fit for advertising.

It also fundamentally corrupts the trust that people have when they go into a product and want the accurate answer, which is what Perplexity is known for. Then you're like, “Hey, by the way, you asked for the best protein shake, but these are good protein shakes that you can check out.” It kind of hurts the trust that people have in your platform, in your product.

That's another reason why, if you think about it, Meta—or I think some other companies in the past—have tried to put ads inside messaging apps and emails, and it's never really worked out. It works out in China, in WeChat, because there's no other way for them to fund the whole thing. The whole economy and user sentiment and user behavior have been optimized around gamifying. That's not how things work in America.

I'm bearish on advertising really taking off in the chat interface. I'm happy to be proven wrong there, but I'm bearish on that.

5. “The Model Is Not the Product”

There are 2 areas that I want to unpack. The first one, just taking them chronologically and as you said them, is that the money's in the frontier. The more I hear this, the more I question it, because I think we dramatically overestimate how important frontier models are to doing quite basic work.

Frontier doesn't mean frontier model. Frontier just means whatever the frontier outcome you can have right now with AI. Greg Brockman recently tweeted, “The model's no longer the product,” right? It's funny because, as a leader of a frontier lab, he has every incentive to say the model is the product. That's what Google people tell us. I think one of the Google people keeps tweeting that the model is the product. I forgot who.

The reason Greg's right is because, if you take Codex, Perplexity Computer, or Claude Code, what is that? It's an orchestration system, right? It takes a model and pairs it with an agent harness. Think of an agent harness in the simplest way: it's rules for how the agent loop should run. What are all the skills, sub-agents, connectors, tools, and accesses?

Without the harness, you don't necessarily capture and convert the intrinsic intelligence in the model into valuable output tokens. If you're literally just a reseller of model tokens, you have no business. The model will get commoditized, so even if you're a model builder, you don't have a business. As an infrastructure layer, you have some business serving those output tokens. But as an application layer or model builder, you don't really have a business if you're just a reseller of tokens that come directly out of the model.

You have a business if you know how to take the model, ground it in valuable context, orchestrate it with a really good agent harness connected to the right set of tools and connectors, whether they're personal connectors or business connectors, and provide the experience to people in 1 single, unified system. The way we differentiate ourselves at Perplexity is that we don't just orchestrate across tools, files, and connectors; we also orchestrate across models.

That is the differentiation that Anthropic and OpenAI cannot claim, because you wouldn't find GPT-5-5 inside the Claude Code harness. You wouldn't find Claude Opus 4.7 or 4.8 inside the Codex harness. These are competing with each other, right? Whereas you would find both of these models inside Perplexity Computer.

That way, we can increase the token value per watt per user. If you assume that the size of the prize in dollars is fundamentally the power in watts, that's the thing that nobody else can subsidize other than the government. Whoever provides the most valuable output tokens with the least amount of power expended to produce them generates the greatest value to the end user and has the most pricing power and the most value.

That's the orchestration problem to solve. The single most important metric in AI is token value per watt per user.

What does it mean for the value of OpenAI and Anthropic if the model is not the product and becomes a utility, something you can switch into and switch out of?

Everyone thinks we're all building the model layer of the race. We're not, actually. I would even argue that building models is a way to stay at the frontier, but you have to own an interface in which valuable AI output tokens are generated. The most valuable tokens. It doesn't have to be the product.

This is the single most important thing to unlearn for most founders. I had to do it, too. To be successful at the AI product layer, whether you're a model builder or not, it's not about building something that gets 1 billion users. That mentality has to completely shift. There are a few power users who are propelling this token economy right now.

6. “Perplexity Was Voted Most Likely to Fail”

If you look at all these crazy stories about how there's 1 engineer who got Amazon to spend $500 million a month because of some stupid way they set up an agent loop inside Claude Code, okay, maybe that's a mistake. But there are real engineers at Meta and other companies spending like $10 million a year per engineer on these coding tools.

There are users in Perplexity Computer. There's 1 user, I think, who spends upwards of $10,000 a month, something like that. Crazy. They're not wasting it. Their business runs using agent loops inside these harnesses, and they use these products in sophisticated ways that I couldn't even conceive of when we were building the product ourselves.

Even internally, inside our own company, there are some people who set up this kind of multi-agent hierarchy and agent loops that looks like its own software architecture. I often just ask these guys to come explain to the rest of the company, “Hey, what are you doing with these tools? You clearly are consuming them way beyond what we thought the average person in the company would do.”

The single biggest differentiation between those who use agents a lot and those who don't is whether they run repetitive cron jobs. Whether you use AI for one-off tasks—you just delegate a task, and then it gets done—that's kind of using it for deep research or whatever. Whereas the AI is continuously monitoring something for you. The AI is continuously triggering based on certain events and going and doing certain things, giving you alerts. You set up workflows that keep running all the time.

7. AI Agents Will Generate More Revenue Than Google Ads?

Every time you get an inbound email, it triages. Every time there's a latency spike, it has to identify which part of the codebase caused that. It has to go and do the root-cause analysis and then identify the right engineer. All these things—this is where the frontier is.

Going back to my main point, these products aren't going to be used by 100 million people. But they will generate revenue that's going to be higher than the advertising revenue of Google or Meta. It's going to happen.

I completely understand what you're saying there. I do just want to focus in on a specific element, when you were talking about the power users, because I think one of the core numbers is actually that Mark Benioff said they spent $300 million on Anthropic.

It'll be interesting to know from him if that $300 million came from—what is the distribution across employees?

That was on developers within Salesforce. It's about 3.8% of developer salaries. What percentage of developer salaries do you think will be spent on tokens in 24 months' time? Because that fundamentally changes the value of OpenAI and Anthropic. If it stays at 3.8%, they won't be $5 trillion companies. But if it's 100%, like Brandon at McKinsey said it will be in a year, they'll be $10 trillion companies.

8. The Next Massive AI Bottleneck (and Why It’s Not Models)

Well, I think they can certainly be $10 trillion companies, whether it's going to be 100% of the developer payroll today or not, because there's a lot of non-developer work that will also be done with agents. That's actually what we focus on for Perplexity Computer. We're not going after the developer market. We're going after anything that non-developers do, basically.

Your finance department, your corporate development team, your sales reps, your data science teams, and your research analysts. I think that's actually an even bigger market. It's not even like that. Think of it as Claude Code multiplied by 10. That's the size of that market.

If I push you on developer salary spend, what percentage of token spend, as a portion of a salary, do you think we'll see in 24 months?

It's hard to say. I think the costs are going to go down. That's why it's hard to say.

Do you think the costs will go down? Because this is the challenge we've had. We thought when we went from chat to agent that costs would go down and token costs would go down. They've gone up.

Yeah, for now.

Help me understand that and how that changes.

I think in software, you kind of want to pay for the frontier. It's kind of like, if you know some engineers are awesome—if you know you have the next Jeff Dean—would you rather hire that person and not hire 5 people who are medium engineers but not Jeff Dean-level, with the same amount of budget you have? Yes, right?

Let's say you had $1 million. You could hire 5 people worth $200,000 each, or you could hire 1 Jeff Dean and pay them $1 million.

What would you do?

The one Jeff Dean.

Yeah. So, I think you would pay for the frontier. But what stays frontier keeps changing. In 12 months from now, let's say—thought experiment—there is an open-source model as good as Opus 4.8.

Mhm.

And when you pair it with the right agent harness and all the connectors—GitHub, everything—and all your developer workflows work fine, why would you assume that the token spend is going to still be high? It's not going to be for the same things you're doing today. But there might be a different set of things you might do with the frontier that you're not conceiving today.

My prediction would be agents that are completely autonomous software engineers. Today, I think we're all using tools like Claude Code or Codex to write code, but not as literal software engineers.

There is a large wave of people that is now bearish on your frontier models who have o1s and your Anthropics, because they're realizing that you can actually do a lot with open models for a fraction of the price. What you're saying is actually that that is true, but—

Yeah.

—we will still pay for the frontier, and so it will still accrue great value.

That's right. And I think this distinction feels like a contradiction. It's not, though. It feels like two things cannot be true simultaneously, but that's not quite the case.

In fact, I would argue that the frontier is increasingly going to be a thing that very few individuals might even want. You could argue that after a point, it's not even interesting that AIs can write software. We've normalized it, right? Let's say that's going to be the case.

Instead of companies being built with tens of thousands of software engineers, unlike in the past, there'll be a lot more companies with smaller software teams, and each of us will be using a lot of AIs. So, that's actually good for the world. We'll be seeing a lot of different businesses. We'll be seeing an allocation of software labor in places that was never even possible.

Whatever it is, it's going to be things like AIs designing chips, AIs designing drugs, AIs figuring out how to build robots, and AIs figuring out how to cure cancer. These are applications where you don't have 10 million users. It's a few companies. But the effect of that work will touch a lot of human lives. To me, that's where the frontier is headed.

You could also see that from the moves that frontier labs are making. Anthropic bought a vet lab. It could be for the talent; it could be for the infrastructure to run vet-lab experiments. But imagine taking all those tokens and putting them into the mid-training instead of just tokens from GitHub. Right? So then that's going to produce something interesting.

Is there an asymptote to the frontier problems to be solved? I know that sounds ridiculous, but if you're continuously on the chase for the next frontier problem, you get to cancer, you get to climate change—and my word, I hope they solve both. Having that is a huge amount to solve. But if you're on the treadmill of continuously solving, is there an asymptote to that?

There's no mathematical argument to there being a cap on the amount of economic value one can create with AGI or ASI-like systems. Elon has a good argument for this: he famously says money loses all meaning in a post-AGI economy because you'll be producing an abundance of energy and labor. Fundamentally, the economy is grounded in energy and labor. If you can produce an abundance of them, what meaning does money have?

I don't think we run out of things to solve at the frontier. I think we're always going to create. Why did people even want to understand the universe? Why did we want to understand subatomic particles, quantum physics, black-hole theory, the origins of the universe? What is the purpose? But we still went ahead and did it, because that's what the purpose of humanity has always been: to understand the unknown.

David Deutsch is famous for saying this, right? We are the only species capable of being curious about what is already familiar. You can stare at a fruit, and you know that it's a mango. You know exactly how it tastes, you know how it looks, you know the shape, you know what season it grows in, and stuff. But you can still look at it and ask one more question about it that you haven't asked before.

9. The Future of 24/7 AI Agents

Other animal species cannot. Once they have it in their mental model—what it looks like, tastes, and feels like—they're going to ignore it. It's no longer interesting to them.

Can I say? You mentioned agent usage, and you said if you do repetitive tasks versus one-off, say, cron jobs. I think Sam Altman said we're going to have 24/7 AI, and they've talked about a hardware product that's going to come out. Do you think we will have continuous agents running?

Yeah. I think so. And I think that's kind of why I believe the orchestration problem—I talked about maximizing the token value.

Can you just help me—sorry, when you say the orchestration problem—

Yeah. So, okay. There are 4 objectives: accuracy, intelligence, privacy, and cost. These are all competing with each other.

You could argue that you can max out on intelligence and accuracy by building giant, giant data centers and spending a lot of power to run them. You could miss out on privacy and cost, because everything will be centralized and you're going to be paying a lot.

You could argue that everything can run locally. That'll be good for privacy and cost, but it may not be frontier intelligence. It may not be frontier accuracy. So, the solution is to figure out a sweet spot: use local models when necessary, use server-side models when necessary, and orchestrate across local models and server-side models grounded in valuable personal context.

Sometimes the intelligence might already be there, but the system might not work because the harness isn't grounded in the right set of tools, right? So, build a world-class harness that can even make an okay-ish model appear great, and be able to use the right model for the right task and the right part of the task—sub-agents. And even utilize the compute we all have in our own devices. All of that doesn't need to be always on a server.

That is an orchestration problem: a router. An awesome router, a master orchestrator router. If you do that, you can realize the vision of a 24/7 AI without people freaking out about going bankrupt.

No one's going to be able to afford a 24/7 frontier AI running on the server. Imagine you just turned it on and you could never switch it off unless something crazy happened. The thing that most people worry about with those AIs is, "Oh, what if it does something crazy?" But the real concern actually is the cost.

Nobody's going to be able to afford a cron job with a fidelity of a few seconds that runs all the time. So, the bottleneck there is actually orchestration and local compute. I believe one needs to build a continuously learning local model that can save you on compaction and context windows, so you try to preserve as much compute locally and rely on the server-side frontier only when necessary. It keeps learning, keeps adapting, keeps evolving.

That model is not just a model. It's a model plus the harness, plus the local chip in the computer, and the ecosystem of devices it controls. That system is going to be your own intelligence. Essentially, the data center moved to your local device, and you get to control it, you get to own it, and you don't get to worry about somebody spying on you or looking at all your tokens—very valuable personal tokens.

Imagine you have very sensitive deal materials. Let's say you're doing a deal, and then a frontier lab has all your tokens that you used to write a memo. Imagine somebody could hack into that server and steal your deal from you. You wouldn't want that, right?

I'm going to be honest, there's much more valuable things for people to steal from London VCs. You're not just yet another London VC. You have like a $400 million fund, last time I read it. So imagine you're already making your moves for the $4 billion fund, right?

So everyone has certain levels of sensitive stuff. And so I think that's where I believe that the 24/7, always-on agent is going to be realized by the company that wants to play the role of the orchestrator—not the model builder, not the frontier model builder, but the orchestrator. And I think that's what we want to do.

Computer has been positioned explicitly as the agent orchestrator. The musicians in the orchestra are these sub-agents that utilize these different models. Think of them as the instruments. The tools, the connectors, and the models—these are all the instruments. The musicians are the sub-agents, and the symphony is the work. The system is the orchestra, and Computer is the orchestra conductor.

That's how it's being positioned. So, what it orchestrates keeps evolving, right? It changes. It changes from models to files to tools to chips to devices.

But it doesn’t even matter. You don’t care as long as it orchestrates things correctly and maximizes the token value per user. If you can solve this problem, you will capture the most economic value in AI long term. Short term, it might look like Lambda Labs’ revenue is growing exponentially, but long term, this is the one objective that truly matters.

10. Why Perplexity Thinks It Can Become the Ultimate AI Orchestrator

Who is best positioned to do that?

I believe it’s us. You have the incentive not to token-maximize. You have the incentive to deliver the most value to the user. Every time any part of the AI stack improves, our product improves.

Since the beginning of the year, Anthropic’s models have made tremendous progress. What’s also true is that our revenue has more than tripled since the beginning of the year. A lot of that is thanks to model progress made by Anthropic. We also brought our burn down thanks to OpenAI competing with them and bringing down the cost of the same capability.

Now, with progress in open source, local models, and local chips, we’re going to move some of the inference back to local devices and bring down the cost even more. Every time any part of the AI stack—whether it’s chips, models, harnesses, or any of these—gets better, our system improves tremendously. If our system improves tremendously, our users love it, they pay more, they spend more, and so our business grows.

So, I think to your question of who’s best positioned to win in that world, for that objective of being an orchestrator, the one best positioned is the one whose product or business benefits from other people’s progress at any layer of the stack. If Jensen produces a better chip, it’s great for us. If Dario produces a better model, it’s great for us. If Apple produces a better device, it’s great for us.

I love the fact that we’re able to be a very positive-sum player at every layer of the stack and not have to rely on any one person to win.

11. The Biggest AI Bottleneck Nobody Can Ignore: Power

When we look at the different providers that we said are kind of server-side versus on-device, a lot of people talk about an AI infrastructure bubble, which I think is funny, stupid, and moronic. To what extent do we have a data center supply problem today, from what you see?

I think the biggest problem is actually power. Let’s break it down. What is a data center? Is it that you just buy a bunch of chips from Dell or Supermicro? No, that’s just one part of it.

You actually have to secure land, or you have to lease a property. You have to buy a bunch of turbines to generate power, or you have to work with power suppliers and grid suppliers. You also have to work on cooling. There’s a lot of other work you have to put in that is far, far slower. You have to get permits to do all these things.

Usually, there’s a lot of lead time to do this. The models that are already in use today have been trained on the Hopper generation. For the Blackwell generation, I think the first model in that category is Mito, and it’s already scary. People are already freaking out about it.

Imagine that everyone pre-trains a model on a million—or hundreds of thousands—of Blackwells. Those models are going to be far more powerful than what exists today. Then the Vera Rubins are coming next year in full capacity. All the data centers with Vera Rubins will be used next year. That model will be even more powerful.

I think there is a certain physical build-out time that always bottlenecks frontier capabilities. That’s why there’s value in that layer. Whoever knows how to do this, puts together a bunch of GPUs and chips and networking and power and cooling, actually orchestrates all this software on top, and is able to convert that into frontier output tokens—that vertical integration has a lot of value.

That’s why the markets are pricing infrastructure companies with a higher P/E ratio than companies like Meta, for example. Even though Meta builds a lot of infrastructure, it’s valued as a software company.

When we see that Meta’s capex spend is increasing in the last few days and that it’s thinking about raising more and more money to increase capex spend, I get it with a lot of the AI providers that you’re opening eyes around, because they aren’t making money from their AI products.

For Meta, the capex spend correlates to increasing accuracy on ads, which is like a 6% to 8% bump in revenue. I get it. But for the capex spend, it doesn’t make sense.

I believe they understand what the market’s saying. They don’t think they’re dumb and unable to see what’s being said. I think they’re introducing a lot of subscription products, from what I’m reading.

The company needs not just to be a social platform maximizing engagement and turning that into ad revenue. I think that requires them to launch a lot of agents, subscription-based products, and maybe even a Meta cloud that rents out servers, like what Elon’s doing at xAI.

Maybe once they do that, the narrative might change. But to go back to my point, it might not be inconceivable that Micron, the supplier of HBMs, might be more valuable than Meta in the next 6 to 12 months. It’s already at like a trillion, and Meta is like 1.3 to 1.4 trillion.

Can you help me understand that? Memory is already a massive bottleneck. It’s increased 5 times in price in terms of COGS.

Right.

But people are going, “Wow, Micron is fully priced at this point.” Why is it not fully priced?

Because it’s still the bottleneck. Whatever is the bottleneck will command the price.

AMD is doing really well because CPUs became a bottleneck again. Agent loops and agent harnesses are all running on CPUs. The tokens are produced by the frontier models on GPUs, but whatever work is happening—for example, Claude generates a coding script that decides to download 500 files from different websites, munches a lot of data, transforms it in certain ways, generates a plot, and then hosts it on a website that you can share with other people—all that compute is running on CPUs.

Agents are using CPUs more than humans. Suddenly, there’s a rise in enterprise CPUs, and the beneficiaries of these are Intel and AMD. They become the bottleneck. Whoever is going to be the bottleneck will win.

Infrastructure is the bottleneck right now because there’s a lot of demand and we just don’t have the supply. Whoever supplies memory, SSDs for storage, or CPU compute—suddenly, these are all interesting. They’re more important than companies that are just building data centers and don’t know how to turn that into a valuable output.

Do you believe Nebius and CoreWeave will be sustainable, multi-hundred-billion-dollar companies in the future, or are they solving a short-term supply problem?

I certainly think they can be sustainable.

Yeah.

I don’t know particularly which of those is going to win. There are also other players like Crusoe and Fluidstack and a bunch of companies.

It’s all about being resourceful. You have to take power from areas where there are a lot of natural resources. The cost to bring up the data center is pretty cheap, the time to bring up the data center is short, and your service is reliable.

If somebody commits to buying 100,000 GPUs from you, the service should be pretty good. You should be able to secure the supply ahead of time, plan well, and I think some companies are even innovating at the power layer. Generating their own power is one way to bring down the margins.

I think there’s certainly value in that layer because it’s hard to replicate the work. You could argue that OpenAI can do all the work that CoreWeave is doing. That’s kind of what they wanted to do with Stargate. But why is CoreWeave more successful at building data centers than OpenAI?

It’s hard to do. It’s operationally intensive.

Yeah, it’s operationally intensive. You have to focus. You have to spend most of your time securing permits, figuring out power, figuring out bottlenecks in the supply chain here and there, and constantly planning ahead and testing all these systems carefully.

You have to deal with random physical issues that arise in running a data center. There’s something called TCO, total cost of ownership. You have to factor that in.

Mhm.

That said, I don’t think there’s value if you’re just a server renter. If you’re just a GPU server rack renter, if you’re just leasing it to different companies at certain hourly rates, there’s not a lot of value.

You have to actually build some software on top, kind of like how AWS did. It’s called Amazon Web Services, not Amazon Servers. You have to have some software orchestration on top that allows you to get software margins on top of what you’re doing.

I think that’s why you’re seeing moves like Nebius going for AI model inference—taking open-source models or hosting your models. That’s the business model of certain other companies like Fireworks and Base 10 and all that. You could imagine a neocloud just going for that business.

12. Can Inference Companies Become the Next $100B Giants?

That was exactly going to be my question.

So, I just had the co-founder of Nebius on the show, and the really clear takeaway was the challenge that he has, which is that there's a huge amount of money that wants just capacity and compute.

Yeah.

With the awareness that he needs to build a full-stack product if he wants to have a long-term sustainable business. That was the core realization for me. When I look at the inference layer, like you said, Fireworks or Baseten, how do you think that plays out? Do we have standalone $100 billion companies in inference alone, or do we see it commoditize?

Possible. It's all about working backwards. What does it take to build a $100 billion company? Assume—

$10 billion in revenue.

Exactly

$10 billion in revenue, 30% to 40% gross margins, a good amount of net income, good cash flow. Okay, $10 billion in revenue is not that inconceivable for a company that can do both AI-hosted inference and server capacity and data center build-outs very operationally well.

There are some factors beyond their control, like open-source models continuing to be awesome. If open-source models stop actually being good, or the gap between them and the frontier is more than 12 months—15 months, 18 months—then I don't think these companies really have a business model. That's because they're not going to be able to host; they'll only be able to rent capacity to OpenAI or Anthropic. And so—

That's exactly what Emad Mostaque said. He said if consolidation happens and there are Anthropic and OpenAI, or 2 or 3 dominant providers, that is the biggest threat to them.

That's correct. You've got to make a leap-of-faith assumption that there will be enough factors in the market—models from China, or NVIDIA making good progress on its models and Nemotron—to keep consolidation from happening as an outcome. But you don't control your own destiny if you're those companies. That's basically the problem.

I totally get that. Okay, so we can have standalone companies that are $100 billion in inference alone. I'm just pillaging you for your knowledge. When we look at the model-selection companies, like OpenRouter, or Foundry AI, which just released that kind of model-selection or model-routing product that did very well on launch, are those $100 billion companies in the model-selection and routing business?

Probably not. I think you can't just be a provider of a router; you have to use the router to produce something meaningful. Actually, most of the business value of OpenRouter is less in the router. Even though the product is called OpenRouter, it's not routing across models there. It's actually just routing across different endpoints of the same model.

So, let's ask this question: If you wanted to use Claude Opus or, I don't know, GPT-5 as a developer, why would you not want to just use it with your own API key versus using it inside OpenRouter? The number-one argument, the single simplest argument as to why you would want to do that is model fallbacks.

Sometimes your API keys might not have the rate limits, or even if you have the rate limits, there might be an error on OpenAI's servers that doesn't guarantee you the response time you need to run your application. OpenRouter would pay for capacity a year ahead with the funding they have, and secure the rate limits and multiple endpoints across multiple different providers of OpenAI models, be it Bedrock, Azure, or OpenAI themselves.

And so that routing is valuable. It's essentially solving an infrastructure problem, which is reliable token supply. It's not actually, "Oh, they're lowering the cost by deciding if this prompt should go to GPT or Claude." That's not what they're actually selling to the developer. That's not actually the business model.

For a lot of these Chinese open-source models, you probably don't want your API tokens going to China. Let's say you don't have the bandwidth to work with different inference providers to verify who's good and who's not. You're just trusting OpenRouter to take care of all that, and then they're going to supply the tokens to you.

So, it's routing not at the level of deciding which model is cheaper for a task. It's more like a reliable token supply. I think there's some value in that layer, definitely. Otherwise, they wouldn't have this many users and this many trillions of tokens being routed a month.

But it's not a high-gross-margin business. The way the business model works for them is that they would secure a discount from the model providers by guaranteeing a lot of supply. But they would still charge the user list price on the API, and that difference is their margin. Do you understand?

I totally get you. We spoke about bottlenecks, and you said HBM, high-bandwidth memory, and Micron, and the value that they have today and what it can be. What bottleneck will we have in 3 years that we're not discussing today?

I think power will remain the bottleneck. It feels like that to me. Unless something dramatically changes in the way data center build-outs happen, I actually believe that there'll be a lot of resistance to building data centers.

It's because people incorrectly think that data centers consume a lot of water or eat up a lot of power, which isn't true—both are untrue. Satya Nadella even made the statement that it's like a can of water or something, in terms of how efficient these companies are.

Do you think that's why they're putting up resistance to them? I think it's because it's a symbol of job losses, increasing wealth inequality.

It's a lot of things. It's a lot of apprehension and fear about what's going to happen, channeling in so many different ways. Sometimes it's channeling through hatred for wealth inequality and wanting to tax people. Sometimes it's channeling through concerns with the environment and climate change.

Sometimes it's channeling in a way where you're like, "Oh, the price of the grid is going up because you guys are building all these data centers." Or, "I'm paying more for my phones and laptops now because the RAM prices have gone up because you guys went and bought all of it."

So, I think there's a lot of different ways in which it's getting channeled, but the common sentiment is a pretty bad sentiment about AI.

Do you think it would be meaningful to the development of those data centers? I think right now 40 out of 100 are not being developed because of public resistance.

Yeah, so that's where the power bottleneck is. You could see certain countries seize the opportunity here and allow these model builders to go build data centers there. Elon is going to space to do that, so that's going to be an interesting experiment, because there's a lot of energy from the sun that can be harnessed there.

There's a lot of natural resources in other countries. Regulations might be more friendly. So, we're still going to see data center build-out. It might not happen in the US.

The fact that you have to solve physical problems—you actually have to deal with the supply chain, the permits, securing power, making sure things work, and getting the lead times lower and lower—means you're not solving problems like cloning some SaaS apps here, right? Or building a go-to-market team, or doing better marketing against the competitors' products.

Yes, those are also hard problems, but these are much harder problems where you're not in full control of your destiny. You need a lot of capital and connections and the right people, sometimes even political help to unlock progress. That's why this will continue to remain the bottleneck, in my opinion.

There's a lot of risk as well, because if you do encounter another DeepSeek moment here, where there's a vastly more efficient model being built with a very different vertically integrated architecture, and you've built out all this capacity, you're like, "Damn, I overbuilt. There's something far more efficient that can run on people's local devices—their MacBooks, their Windows PCs." You're probably freaking out then. And so, you hope that—

How likely do you think that is, though?

It's probably a 20% to 30% chance. The reason I think there is some possibility is because of the export controls. DeepSeek is not building with NVIDIA's stack. They're building with the Huawei stack.

Because there are export controls not just on NVIDIA GPUs but also on HBM, the architectures that DeepSeek is building are far more memory-efficient. They made innovations on the KV cache to make it small enough that you can host it on SSDs. You don't need high-bandwidth memory for inference time.

They're going to have a completely different architecture for inference and a completely different architecture for storage, because they're not allowed to use 3D NAND. So, their architecture is going to look different. It's not just a model architecture.

The model architecture is already pretty different. They made innovations on the attention layer. They made innovations on the training algorithm so that it doesn't consume a lot of interconnect capacity.

So they made a lot of—basically, their whole stack is getting vertically integrated into their hardware, their chips, their fabs, and so on. That's a very different bet from what America is making.

13. Did U.S. Export Controls Accidentally Make China Stronger?

Do you think the export controls have helped or hurt us?

The jury's still out. In the short term, it's helping because, in my belief, the only reason there is even a development gap between open source and the frontier is export controls. It's definitely helped, and companies like Anthropic lobbied very hard for it.

But there is a chance that, because of that, they now get really good at the physical layer. One advantage they have is that they can actually build data centers a lot faster. Power is not a problem, permits are not a problem, people are not a problem, labor is not a problem, and expertise is not a problem. By forcing them to go out there and build all this, you're converting them into a far more potent competitor.

Do you think we still dramatically underestimate China's capabilities?

I think so. If AI is not just digital but also physical AI, you've got to build fabs, robots, and chips; harness the energy really well; and package it into local devices. I think they have a lot more advantages than America.

How important is it that we have TSMC in the U.S.?

TSMC is actually there—there is a TSMC fab in Arizona. A lot of people talk about this, but TSMC is investing around $150 billion into building American fabs. They've already invested $40 billion or something like that—$60 billion, last time I checked.

There is a TSMC in Arizona that's coming up. There's also Intel, and that's why the American government owns 10% of Intel. NVIDIA and SoftBank own 5% each, so there is a lot of investment going into an American fab, as well as TSMC investing in its American fabs. Elon is building Terafab. I think people have woken up to the importance of building fabs, but this is also why China is particularly competent.

Given the capabilities of China that we just mentioned so articulately—I know it's a ridiculous question—but if I were to say to you, “Your job is to make sure America stays competitive,” what would you do to ensure that you retained competitiveness in the face of an increasingly strong China?

I think we need to take physical infrastructure a lot more seriously and continue funding it. We shouldn't propagate fake news around data centers about how data centers are polluting and contaminating water, or how they're sucking up water. We need to actually be fact-driven.

I hope our product helps there. You can go to Perplexity, ask any question, and get fact-checked on your assumptions. It's very important that we educate the public about what's actually going on in a language they easily understand and not fear-monger.

We shouldn't be saying, “All their jobs are going to go away. There are going to be lots of amazing companies built with far fewer people, getting multibillion-dollar and multi-hundred-million-dollar valuations with 20 or 30 people and propelling trillions of dollars of new GDP.” Let's talk about how to enable that. Let's talk about how to build that and create a more positive future together, instead of saying, “90% of the jobs are going to be gone. You're all going to get screwed over by our models, and it's our moral duty to tell you all this.” That doesn't make any sense to me. You can't win by saying that while also complaining about not being able to build data centers fast enough.

14. The AI Jobs Narrative Is All Wrong

Do you think we've done a complete disservice by having the marketing message that Dario has had—that all jobs are going and it's all doom and gloom?

Yeah, I think so. They have contradictory messages in their different social engagements so far. The most recent one I heard was, “There is no evidence that AI is taking over jobs.” I think there needs to be consistent communication around this.

Very little is being said about how AIs can help you build companies in a very different way. With current AIs, even when it's generative AI, it's already true that so many things you would hire people for can be done with agents. One way of looking at it is, “What happens to all the jobs?” But the other way of looking at it is, “Hey, I never had the chance to go build a company around this idea I've been having all this time. Maybe a group of friends and I can come together and build this. Can you figure out a way to give us compute credits?”

Amazon gave a lot of compute credits to a lot of startups. When we started Perplexity, we had around $200,000 worth of AWS credits, GCP credits, and Azure credits. Together, cumulatively, this was worth almost $1 million in compute credits. In today's world, it's going to be like $1 million of compute credits, and we're doing that. We're funding this thing called the Billion Dollar Build, where we're giving $1 million of compute credits to any group of people who have a credible path to building a billion-dollar company. I want thousands of such companies to be built.

What did you think of Sam Altman giving $2 million of tokens to YC companies in exchange?

I think we should do more of that. That's the right thing to do. We should do a lot more of this because you want new companies to be built. Even if they're worth multi-hundred-million dollars, it's good. If there are thousands of them, that's a lot of new GDP.

I spoke to Amba Kak before the show, and she asked, “How has AI built the team for you?” How big is the team today?

It's around 400 people.

Four hundred people. How big will it be in 2 years' time?

I don't know. It's hard to say. Maybe 800 or 1,000.

15. Why Future Unicorns Will Need Far Fewer Employees

Will companies follow the same headcount trajectory that they have always followed, and will we just solve new problems, or will they be dramatically more efficient with a much smaller number of people?

Definitely, they'll be dramatically more efficient. That's why I am a believer in building a lot more efficient companies now and being an example for all these companies ourselves. People should look at Perplexity and be like, “With 400 people, you can build a $20 billion company.” That means with 40 people, I could probably build a billion-dollar or $2 billion company. That's totally doable.

For us, maybe that means with 4,000 people, we could be worth $200 billion. We could be worth $2 trillion with 10,000 people. That doesn't mean it's bad for all the 100,000 people we did not hire for a typical $2 trillion company. I would rather have those 100,000 people split into groups of 100, with each of those 1,000 groups worth a few billion dollars. That's awesome.

A lot more people need to be entrepreneurial. There are people who would be bad employees in any company because they're difficult to work with. They don't listen to instructions, they don't follow road maps, or they're not easy to collaborate with. But maybe the flip side of that is those are the kinds of qualities that founders typically have.

Aravind, there is a population—and a very large population—of people who are not AI-native and are not using AI to improve workflows or improve efficiency. What would you advise them?

Get started. The first step is to get started and channelize your curiosity. You don't need to use AIs to do your existing work. If your existing work is boring to you, you probably won't enjoy it even if you use AIs to do it.

You got a lot of heat for saying that people don't like their jobs.

I didn't say that. If you actually listen to my interview, I did not say that. People want clickbait articles, and they take something I said in one sentence, out of context, and make it into a headline.

What did you say?

I specifically said this: “Hey, there are a lot of people who don't enjoy their jobs.” By the way, the fact that that thing went viral is not because I was completely wrong. I think a lot of people resonated with the fact that I was actually honest in saying that a lot of people don't enjoy their jobs, and that has nothing to do with your economic position or standing in society. You might even be really wealthy but doing a job that you completely don't enjoy and destroying the peak years of your adult life working on something that is horrible or depressing.

My point is that if that's you, and if the reason you could never leave your job is because you were always worried about how you would build a company from scratch, that's changed. There are all these things to figure out—how you would hire a lot of people, set up an office, and so on. For the first time in history, you can get started on an idea with 1 or 2 other friends and maybe have a real, genuine shot at building a billion-dollar company.

I totally get that. Everything that we've discussed today has been on the back of unprecedented demand, up and to the right.

We need more memory, we need more data center supply, and we need it on demand and on the service side. Everything is up and to the right. I'm seeing some cracks in that, with Uber saying, “I'm not sure I'm getting the productivity gains that I thought.” Microsoft is lining up with them and putting a $1,500 token budget in place. Do you think we will have a continuous, up-and-to-the-right acceptance that productivity gains are unwavering—we have to do this—or will there be falterings along the way?

I'm sure there are going to be falterings along the way, and people are rightfully freaking out about token-maxing. That's why I think you need some form of hybrid agentic inference. You need some amount of inference compute to run locally that you're not paying for tokens on—unmetered intelligence, essentially.

How will the best companies of the future structure token budgets?

My hope is that they don't have to understand that. They will be able to work with an orchestrator who does it for them. It's not going to be easy for you to constantly keep track of which models are the best at what things and how you allocate everything. This is the budget for coding; this is the budget for finance. How do you even understand which models are good at each of those things, and how much do you spend on each of these divisions? You're not going to be able to keep track.

I had a friend on the show the other day say that Google will be the token king. They can produce the lowest-cost tokens of anyone. They own the full-stack team, and they use data, networking, and power procurement. Do you think that's true—that they will be the lowest-cost token producer?

They have all the advantages one needs to have to be that. But they underestimated the importance of coding models, and so they're far behind the frontier right now. Again, they could catch up. They're a totally capable, totally competent team, but today they're not quite at the frontier.

I was shocked the other day when I saw the Cloudflare announcement that agent traffic has now overtaken human traffic for them.

Why are you shocked?

It was quicker than I thought.

Okay.

Personally, I thought that would happen in 2 years, maybe, not now. How does the world change when agent traffic far exceeds human traffic?

I think people are just going to have a lot more agency. That's it.

Do websites go away? Does design not matter? Does the advertising model of the internet die completely?

No, it doesn't. My belief is that the advertising models around travel, shopping, and fashion aren't getting disrupted by agents because the judgment isn't objective. Anything where the judgment is objective—where the transaction is based on objective judgment—is going to get disrupted by agents. Anything where the transaction is more subjective, where the decisions are more subjective, will be different.

What is the best piece of furniture inside this room? Why this particular table? Those kinds of things are probably subjective. For a mic, you would make an objective decision. For the table, you probably care about the aesthetics of the room. I think that's how the world will split: subjective things will still be ad-based, and objective things will be agent-based.

I watched your commencement speech after speaking to Sam Altman at ExCeL, and he said I had to watch it, so obviously I watched it. One of the points you made was that the defining skill of the AI era is asking better questions.

Yeah.

What question is no one asking today that maybe everyone should be asking?

I think people need to ask more about this: assuming I have a lot of agency available to me, what do I do? Imagine I gave you a head count of 100,000 people or 10,000 people, and enough compute credits to run those agents. What would you do?

Let's say I ask you, Harry: you suddenly have 10,000 agents at your disposal. What would you do? I remember you telling me—or not me, but in some episode of yours—where you said you only did this podcast because you felt like you didn't have an arbitrage to go and do deals.

100%.

Yeah.

That's why I still do it. I mean, I love what I do, but yeah.

Okay, so you've gotten some amount of distribution. Now, assuming that you could spend $100 million on agentic inference and ground it with all the connectors and everything, and it's all working, what would you do with that capability to further your goals? What should your goals even be?

I think that's the question I would ask. Assuming that in the next 3 to 5 years you're going to be able to delegate whatever digital task you want, with the right harness and agents, what would you delegate?

Fundamentally, it would be to build a gigantic infrastructure to be able to find, identify, conduct outreach, set up, and win great investments, and have the media sit on top of that and power it. That is intensely difficult to do and would be the holy grail for investing.

Yeah.

That would power what my end-goal ambition is.

Yeah, so your goal is to run a 10–100x larger fund, right? That's basically what I'm hearing from you. Let's assume you have a $40 billion fund instead of a $400 million fund. All you've got to ask is: assuming I have all the head count I need to do this, how much faster can I do it? I think that's how I would frame this question.

Elon has a similar idea he spoke about once: assume somebody tells you a task is going to take 10 years. Ask the question, “What would it take to do it in 10 months?” Maybe it's impossible to do it in 10 months, but you'll probably get pretty far by asking that question compared to somebody who takes it for granted that it's going to take 10 years.

All right, interviewer, let me put it on you. What's your 10-year goal, and how does that look in a 10-month time frame?

I think our mission, beyond any level of capitalism, is to make the planet more curious. The product is always intended to help people ask the next question. My goal is to truly realize that the level of agency that needs to exist in this world is quite not there.

I think that needs to be grounded in numbers, dude, to make it possible. It's like me saying, “Oh, I want the best investments.” Well, which is why a $40 billion fund is helpful.

Sure, I could say the same thing: $2 trillion. It doesn't matter, right? One hundred x, 10x, 1,000x—these are all motivational milestones.

Do you think Perplexity will be a trillion-dollar company?

Yeah. Anyone can be a trillion-dollar company. SK Hynix and Samsung are worth a trillion dollars as of the last couple of weeks. Did you know Samsung started off as a grocery store? You didn't know that? Okay, so it's true—they started by selling dried fish. Seriously.

The SK Group started off as a textile company. Anyone can be worth a trillion dollars. You just have to work your way toward that. I mean, it's the exact same logic you laid out for how a company can be worth $100 billion. You said you need to make $10 billion in revenue. Isn't it the same for a trillion? You need to make $100 billion in revenue.

And there was actually some really interesting data that Coatue revealed—I don't know if you saw it recently—which was basically about the probability of reaching the next level of value.

Yeah.

16. Wealth Inequality, AI & The New American Dream

It's much higher. When you're at $1 billion, it's much more likely that you'll reach $10 billion; at $10 billion, it's much more likely that you'll reach the next level.

Yeah, that's true even for people. It's way more likely for a person with $100 million in liquid net worth to become a billionaire than someone with $10 million.

Are you not worried about wealth inequality? Honestly, if we were being blunt, we're both very lucky now to live in nice worlds and rarified air. Are you not worried by how much money a very small number of people have, how hard it is for everyone else, and how that gap is getting bigger?

I think the way to ensure that doesn't remain the case is to distribute the benefits more widely. You've got to let anybody benefit. By the way, the people who are using our tools—I've had an Uber driver; I'm not even making this up—an Uber driver in San Francisco once told me that he watched one of my YouTube interviews where I explained how you can build a product or a web app with AI from scratch. He went on to do it and used AI to add billing and everything else, and that makes more passive income for him than driving for Uber.

He actually reduced the amount of time he was driving for Uber because he loves building new apps with AI. That already tells you that for a person with agency and a positive outlook for the future, anything is possible.

If you keep communicating all the negative things you can about AI and wealth inequality all the time, and that's the only thing the news and press write about, I think it will perpetuate that, and people will only think about the bad things. It's essential that if you think you're already doing well, you talk about all the things that can go well and give hope to people who are down, like you. Even you—you started this podcasting circuit when you had nothing, right?

Nothing.

Exactly. So, it’s possible. You’ve got to talk more about that than be like, “Oh, I feel so guilty that I made it, and now I know—what about all these people who haven’t made it?” You can also make it.

I think I have a more pessimistic view of the actual general public, which is that I don’t think that many people have agency. I think a lot of people have their own mentality.

To help them, I think that’s the most important thing.

I think they’ve got to help themselves.

Sure, but people will help themselves once they see that: “Okay, I kind of want to be like this guy. Let me work hard.” You need an example, right? It’s not like nobody can get in shape. It takes discipline. You have to get rid of bad habits.

Now is the best time ever to change your life in 12 months. The ability to go from nothing to actually becoming a billionaire in 12 months is now possible in some respects.

And so, look, I’m not saying everyone’s going to make it and everyone’s going to be worth $1 billion. Isn’t that the caption from this show? Aravind, everyone’s going to make it—

Anyone has the potential to make it. So, it’s as likely for Perplexity to become worth $2 trillion as it is for a founder who’s yet to secure funding to be worth $1 billion. It’s equally hard. You just have to give yourself shots at the goal and be curious. That’s the message from the commencement speech: be curious.

We have SpaceX, Anthropic, and OpenAI going public. It feels like someone’s shot the gun and the race is on. Is there enough money to fund 3 such large IPOs?

There will be some reallocation, for sure. There might be some holders of SaaS stocks who would put it into Anthropic or something. Let’s say you believe that enterprise AI is going to take off. You might want to hedge between having a lot of Microsoft stock and Salesforce stock versus putting some of that into Anthropic.

Let’s say Vanguard or BlackRock cumulatively own, like, $200 billion of Microsoft and Salesforce. They might be like, “Okay, I’m going to take $30–40 billion of that and put it into Anthropic.” Fine. Not a bad bet to make.

What happens to all the enterprise SaaS companies that are public, going, “Nah, fine”?

They have to weather the storm.

Is it a storm, or is it continuous precipitation?

I think you have to bring down the costs and produce new value. Salesforce has done well because they always went and bought the next thing. If you’re just selling the same software, you’re probably not going to be around.

IBM is still around because they went and bought Red Hat and HashiCorp. Now they’re buying Confluent. There are ways for these companies to stay alive and extend their lifespans. It’s obviously going to be hard to preserve a brand that’s as relevant. I don’t think the IBM brand is that relevant anymore in terms of evoking an emotion in people to go use their products.

But as a business, it’s going to be awesome. It’s going to be fine.

You said IPO in 2028, and I had to ask this. I woke up to this in my group—we have a team WhatsApp—and it was like, “Aravind Srinivas, IPO 2028.”

I hope it can be sooner than that.

When do you know when you’re ready? Is there a $1 billion ARR threshold? You’re at $500 million ARR now?

More than that. Far, far more than that, actually.

Really? What—

We’re not yet ready to share it, but we’re growing really fast.

Revenue growth matters much more to you than profitability.

Today, I think in general, you can look at public markets. People want top-line growth more than bottom-line efficiency right now because it’s very hard. It’s rare.

But you definitely need one.

Of course. For sustainable businesses, you need to have a model in place to get to bottom-line efficiency when that becomes the objective. You also need to have a path to getting there.

Where are you cost-inefficient today, where you expect to be significantly better in 2–3 years?

17. Why Perplexity Is Training Its Own Models

We’re training our own models. We’re training on top of amazing open-source models, and that will bring down the cost that we currently spend on frontier-model tokens. We expect to continue to use frontier models for designing new experiences and new capabilities that do not exist today in our products.

But whatever exists today in our products right now, we expect it to completely rely on models we own and serve ourselves. That’s all going to be the best way to bring down the costs and increase our margins.

Will the largest enterprise in the world be fine-tuning open models to have tailored models that are much more specific to them?

Absolutely, because it’s in your incentives to bring down the costs.

Does that not provide another bear case for the large frontier-model providers?

Frontier-model providers will only remain relevant if they remain at the frontier. If, for 6 months, you’re not seeing a new capability, it’s bad for them. That’s the uncomfortable nature of this field. No one’s ever in a comfortable position.

Like I said at the start, no one can relax. This is a horse race, and it’s getting harder.

It’s going to get even harder.

That’s the nature of this. The prize is too big. Take Anthropic. I think it’s worth $1–1.5 trillion, something in that range. That’s basically the valuation of Meta, and this all was created in 6 years. Meta took 20 years to build.

The prize is so big. No one can be comfortable. Anyone who’s winning today can lose tomorrow, including the model providers.

Before this year, there was a 3-month period when people were like, “Oh, but Perplexity, what’s happening with Perplexity?” I don’t know. Do you pay attention? Do you care? There was one in particular in San Francisco. Do you remember when they were like, “What’s the company you would short?”

Of course I pay attention to all of that.

Yeah, we were voted the most likely to fail. Cursor was voted the second most likely to fail. OpenAI was voted the third, or something.

I feel like we’re all doing well.

Cursor, I think it’s getting sold. SpaceX and OpenAI—

Going public, baby.

—going public soon.

We tripled our revenue since that judgment was made. Our burn is down by more than 50%. I don’t know. My sense is that most of those people who sit on these meetups don’t actually build anything useful.

Okay, we’re going to do a quick-fire round because I could talk to you all day. What’s 1 widely held belief that you think is completely wrong?

I think a lot of people are obsessed with identifying a model in the first year or 2 of their company. But I think the only shot you have is to move fast. Velocity, in my mind—moving fast—is a way of expressing humility because you’re constantly making contact with the world and trying to question your assumptions all the time.

Where are you still moving too slowly internally today?

18. Turning Perplexity Into an AGI-Powered Company

I think we can be even more reactive. It’s insane that I’m saying this because we’re building some of the most interesting AI products, and internal adoption of our own products and our competitors’ products can be even higher. This is despite us being extremely agentic internally and trying to delegate as much to agents.

That’s a big area for us. My hope is that we can turn this company almost into an AGI. That doesn’t mean no humans work here. There will be an AGI that has all the context it needs to run different divisions of the company in a semiautonomous way, with some scaffolding provided by humans here and there.

That’s not going to feel scary at all. We’ll normalize that feeling very fast. It’s just going to feel like 10 10× engineers running certain aspects of the company.

If I gave you unlimited money, what would you do today that you’re not doing?

I would build data centers.

You would?

Yeah.

In space?

I don’t have the expertise to do that, but I would start with land on Earth. I think there’s a lot of land, and maybe you can be resourceful in securing permits and power in different countries. But I would start there.

I think physical infrastructure build-outs are the return of the Industrial Age again. The forefathers who built the Industrial Revolution—oil pipelines, steel bridges, factories producing cars, all these things that we take for granted today—were built by people who spent a lot of time thinking about how to scale these things in a cost-efficient way.

We need to do that a lot for AI. That’s what I would do. Of course, you cannot just be building infrastructure. You need to be able to utilize all that infrastructure to produce valuable output tokens for the user. But we’re already good at doing that, so infrastructure is the thing I would focus on.

19. SpaceX vs OpenAI vs Anthropic: The Best 10-Year Bet

You can buy and hold for 10 years: SpaceX, Anthropic, or OpenAI?

The 3 IPOs coming in the next few months: which would you buy and hold for 10 years, and why?

SpaceX.

Why?

It’s an easy one. Anthropic and OpenAI can claim they do whatever the other does, but SpaceX is the only company building space infrastructure for connectivity.

Have you been on a flight with Starlink?

No.

You should. You will hate being on a flight without Starlink after that. Imagine we can record this, and I can watch this podcast while flying on a plane. Starlink lets you do that. That’s just one aspect of the business.

One small aspect of the business.

Yeah. There are a lot of possibilities I’m excited about, like being able to travel from Australia to San Francisco in 30 minutes. All this feels like science fiction, but I’m excited about all these possibilities.

What job does not exist today that will be incredibly common in 5 years’ time?

I think it already exists. The forward-deployed engineer is definitely on the rise. I guess people with a really good sense of quality control.

Maybe a better answer is that most valuable jobs that exist are usually reincarnations of something that already existed. I don’t think we’re going to see completely new things. They’re going to reincarnate in different ways.

You can advise your little sibling who’s finishing university today and has just done a computer science degree. One thing: what would you advise them?

Stay curious. Don’t give in to FOMO and try to max out on something here in the short term. Don’t go to Twitter and feel like a loser because people at frontier labs are getting so rich, and everything feels hopeless to you or something.

There’s so much more to build. We’re just getting started. There’s the application-layer era and infrastructure build-outs. There are a lot of opportunities.

We’re seeing more spinouts from OpenAI, Anthropic—you name it—every single day. Do we have hundreds of these new labs and vertical models?

No. I’m not a big believer in too many of them. I think you’ve got to produce some differentiation. That’s the most important thing.

Would you call DeepSeek a new lab?

No.

Why?

I think, very stupidly, for me, I don’t call it a new lab because I attribute new labs to spinouts from larger labs.

I see.

And they’re kind of verticalized, which is probably wrong on both axes.

But it’s horizontal, and it’s not a spinout.

Yeah. I kind of like the idea of labs taking a differentiated bet. If somebody really questions the transformer architecture itself, or questions the need to build on NVIDIA GPUs, or goes out and builds foundation models for robotics, those foundational bets make sense for a lab.

I feel like there are just labs for the sake of being labs, and I don’t think they’re going to make it.

What’s the most plausible story where Perplexity becomes a trillion-dollar company? What do you do then?

Accuracy and orchestration are 2 goals that have been consistently chosen since the beginning of our company. I think we’ll continue to do that. We’ll be orchestrating across devices, chips, models, tools, files, connectors—everything, right?

So what would I do once that happens? I don’t know. We’ll chart our path to $10 trillion.

Are you happy now? Are you enjoying this?

Of course. I wouldn’t be doing this otherwise. There are so many things I could be doing if it weren’t for this.

The process is what motivates you. You asked me—I think somewhere in between, you need to give me a number for where you want to go. I don’t work like that, actually.

For example, these numbers—getting to $2 trillion or $20 trillion—are exciting, but that doesn’t motivate me. It’s hard to get motivated by wealth. You want to get motivated by impact.

Who’s the smartest person you’ve met? Final one. You’ve met Jensen Huang, you’ve met the best of the best. Who’s the smartest?

People are smart in their own ways. It’s hard to compare. I’ve met Jensen, Elon, Bezos, all these guys.

What was it like meeting Elon?

Amazing. Elon’s a very focused person. He might not appear that way on Twitter, with a lot of random tweets, but he’s extremely laser-sharp-focused on whatever he’s doing at that moment in time.

The 1 skill that, as an entrepreneur, I would really like to take from somebody like him and have for myself is that ability to just zone out of all the other things happening in your business or other businesses and focus on the limiting problem right now—the bottleneck problem—and ignore everything else.

It’s very hard to do. Even within Perplexity, I cannot just focus on 1 part of the business alone. It’s very difficult. I’m always looking at other things simultaneously.

His style is to always look at the limiting problem and ignore everything else. That’s very hard to do because you actually have to be really good at concentration. You have to be really good at ignoring even important things that are distractions to your core objective right now.

Was Jensen Huang who you thought he’d be?

Far better.

Really?

Yeah. Jensen is so truth-seeking, it’s insane. I think he, or somebody else, told me—or I read in a book—that he’s so intense that he wakes up every day and tells himself that he sucks, and he tells everybody around him that they’re 30 days away from going out of business.

Think about it, right? A $5 trillion company, guaranteed to make $500 billion in revenue in the next 2 years, with the most advanced chips in the world, and he operates with the mentality that he could be 30 days away from going out of business. That is what it takes to be Jensen Huang.

20. Elon, Jensen & Why You Should Never Retire

There’s so much to learn from these guys. There’s so much to learn. I think there’s 1 aspect of being comfortable where you are, thinking you made it. It feels good to get here so far, but these guys are not stopping.

If you look at Elon’s pay package with SpaceX, it’s structured around creating a colony on Mars with 1 million inhabitants and building enough compute in space. That’s why it’s not motivating to be worth $10 trillion in net worth or something. If he does these things, I’m sure he’s going to get there, but it’s more about making the impossible things happen and having that long-term outlook.

I think that has been the biggest thing to learn from these 2 individuals in particular. A lot of people view entrepreneurship as, “Oh, if I win and I have a great outcome and sell my company, I would have generational money. I don’t have to work ever again.” And then what?

You end up just staying at home. Your kids will obviously have trust funds, and they’re not going to get inspired watching their dad play padel.

Yeah, you know.

You’re not going to set the right example for them. They’re not going to be able to take your wealth and multiply it because they didn’t watch somebody who actually did that. You did it before they were adults.

So I think you always need to be doing something. Jensen said recently that he hopes to die on the job or something like that. That’s the attitude you need to have. You need to work forever.

I was so upset when Jensen said, “If I’d known how hard it was going to be, I wouldn’t have done it.” I don’t know if you saw that interview. I was like, “Oh, my God.”

Yeah. I think it’s pretty hard, but you don’t do it because it’s easy; you do it despite that. I think that’s how it works.

Aravind, this has been so fantastic today. I so appreciate you taking the time while you’re in London. Thank you so much for joining me.

Appreciate it.