Perplexity CEO: SMBs Could Drive the Next Economic Boom

Perplexity Founder & CEO Aravind Srinivas believes the next wave of American economic growth could come from smaller, AI-native businesses, not just the largest technology companies. Perplexity, valued at ~$20B with more than $2B in reported funding and backing from NVIDIA, SoftBank, IVP, NEA, Accel, Jeff Bezos, and Elad Gil, has evolved from an AI answer engine into a platform for autonomous agents, research, and knowledge work. I sat down with Aravind at Perplexity's new San Francisco headquarters to discuss the economics of AI, the future of entrepreneurship, and why agents could enable a generation of smaller, more profitable companies. Aravind reveals that Perplexity's revenue is growing 5x year over year, enterprise adoption is accelerating, and one solopreneur is spending over $1M annually on Perplexity Computer while generating $3M in annual revenue. We also explore Perplexity's bet on local AI, its work with NVIDIA and Apple, and why the company has no plans to build its own data centers despite the massive infrastructure buildout underway. Aravind Srinivas: https://x.com/AravSrinivas Molly O’Shea: https://x.com/MollySOShea Sourcery: ⁠https://x.com/sourceryy 𝐄𝐏𝐈𝐒𝐎𝐃𝐄 𝐋𝐈𝐍𝐊 YouTube: https://youtu.be/1C8_n9nnx5Y 𝐒𝐏𝐎𝐍𝐒𝐎𝐑𝐒 • Brex https://brex.com/sourcery—The modern finance platform, combining the world’s smartest corporate card with integrated expense management, banking, bill pay, & travel. https://brex.com/sourcery • Harvey http://www.harvey.ai?utm_medium=paid-partner&utm_source=sourcery&utm_campaign=2026-10-brand-sourcery-partnership—The AI platform built for legal work, is helping the best firms and companies close deals faster. More than 200,000 lawyers, 80% of the Am Law 100, and over 20% of the Fortune 500 build with Harvey. Visit: http://harvey.ai/ http://www.harvey.ai?utm_medium=paid-partner&utm_source=sourcery&utm_campaign=2026-10-brand-sourcery-partnership • Zone https://zonefrontier.com/—develops next-generation data center campuses, partnering with AI companies, site developers and technology leaders to bring compute online faster and at scale. Visit: https://zonefrontier.com https://zonefrontier.com/ • Turing https://turing.com/sourcery—Turing delivers top-tier talent, data, and tools to help AI labs improve model performance—and enables enterprises to turn those models into powerful, production-ready systems. https://turing.com/sourcery • VCX https://fundrise.com/vcx—VCX is the public ticker for private tech, allowing investors of all sizes to invest in venture capital. View The Portfolio at http://GetVCX.com http://getvcx.com • Deel https://www.deel.com/sourcery—Deel is the global people platform that helps startups hire, manage, pay, and equip anyone, anywhere. Trusted by more than 35,000 fast-growing companies, Deel is the people platform that just works, so teams can scale without the chaos. Visit: https://www.deel.com/sourcery • Public http://public.com/sourcery–Investing platform Public just launched Generated Assets https://public.com/generated-asset, which lets you turn any idea into an investable index with AI. With Generated Assets, you can build, backtest, refine, and invest in any thesis with AI. Gone are the days of one-size-fits-all ETFs. https://public.com/sourcery Follow Sourcery for the latest updates! https://www.sourcery.vc https://www.sourcery.vc/ Disclosure: Paid Ad. Agents are an experimental AI tool. Its output may be inaccurate and is not investment advice. Public Holding’s, Inc. makes no guarantees about its accuracy or reliability—verify independently before use. Brokerage accounts and services provided by Open to the Public Investing, Inc. See additional disclosures at https://public.com/disclosures/agenticdisclosures. 𝐓𝐈𝐌𝐄𝐒𝐓𝐀𝐌𝐏𝐒 (00:00) Aravind Srinivas, Co-founder & CEO of Perplexity (01:30) Why Perplexity's brand is built on curiosity (08:18) How to build a $1T company in America alone (13:45) Is the American dream still alive? (16:28) Stop blaming users for bad prompts (22:55) How Perplexity quietly grew 5x (28:28) "We don't have competitors, we have contributors" (29:20) The data center is coming to your laptop (34:32) Orchestrating up with NVIDIA, down with Apple (39:17) Why Perplexity is skipping the data center race (41:25) Stop asking AI to write code, ask it to think (48:27) The Hugging Face incident & multi-agent threats (53:31) Why Perplexity isn’t chasing a billion users

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  Joining us is Arvind Shrinhavas, he is the co-founder and CEO of Perplexity.

• 0:03Molly O'Shea?

  Perplexity has been on quite the terrace.

• 0:05Speaker 02

  Perplexity rolling out its AI browser Comet to all users this week.

• 0:09Speaker 03

  The data center is coming to your laptop. When we started, people made fun of us. We did it in a way that essentially became the blueprint for every other chatbot out there.

• 0:17Molly O'Shea?

  A lot of brands have really bad aura. You guys have great aura.

• 0:21Speaker 03

  Our brand is fundamentally to celebrate curiosity. We think that'll be the only quality we human beings can be proud of. Search is actually compounding. Every form of genetic action requires access to web, access to a cloud browser, access to context across connectors, files and tools, and putting it all together and orchestrating across models. I can't think of another company that's better positioned than us to do that. We don't have competitors. We have contributors. We're very glad to partner with amazing chip companies like Nvidia and Apple, Intel. Like, we're gonna work with all of them.

• 0:52Speaker 00

  As Nvidia reportedly weighs investing in the company at a more than $30 billion valuation.

• 1:07Molly O'Shea?

  Arvind, thank you so much for having us here at Perplexity. I know you guys just moved into this office.

• 1:12Speaker 03

  Thanks for coming by and maybe once we're fully settled in here, we can do a proper tour around the office.

• 1:18Molly O'Shea?

  I'd be happy to. The walk and talks have been really popular. It's funny, we were just in Italy with bending spoons and they were calling it a walkie-talkie. So we'll do a walkie-talkie with you. So before we started rolling, we were talking about all the AI brands and you know, there's a war between all of them, but the one thing that I've particularly noticed with you and Perplexity is you've withheld an above-board brand the entire time.

• 1:48Speaker 03

  Our brand is fundamentally to celebrate curiosity. We picked that because we think that'll be the only quality we human beings can be proud of and lean into even more as AIs get better and better. Like, you can, you could argue that AI is the best accelerant of our own curiosity. Like, almost the purpose of AIs to accelerate human curiosity. And David Deutsch, the famous Oxford physicist, he has this quote that thesis almost that human beings are the only species that are capable of being curious about what is already familiar to us.

• 2:32Speaker 03

  Like, we might know something and then we could take a look at it again and, and try to think about it from a different perspective. Animals don't tend to do that. And that's very interesting because you could argue that, oh, like, AIs can solve all the problems. AIs can answer whatever question there is. So what's, what's, what's our role? Our role is to actually go and ask the next set of questions. We actually were designed to ask questions, but we grow up in settings where we are rewarded for answers. Until AIs came to exist and now it's like, it's not necessarily cool to be a know-it-all anymore because an AI can answer everything. Like, nobody needs to go and find an expert

• 3:17Speaker 03

  to learn about a topic. You can actually just go to the AI. Whatever AI you trust for accurate answers, whatever it is. But then what is your purpose? Your purpose is to actually lean into your curiosity even more and not be afraid of asking whatever questions, however dumb they are, however of an outsider you feel with the topic. You don't have to feel that bottleneck or fear of being judged. And so our, our, our brand is built around that. And we think that the best companies always have a fundamental brand that has nothing to do with the technology, more to do with humanity and people. Like Steve Jobs built a brand for Apple about putting beautiful things out there because that's the best way to celebrate

• 4:02Speaker 03

  and give back to mankind. Nvidia's mission was just make things possible that only supercomputers can. It was never about AI. It's, even today, if you go to their website and read about their mission, it's about like do things that are beyond the realms of regular computing. And that, that applies to AI today, but it could apply to quantum computing tomorrow, it could apply to something else later. It could apply to cryptocurrency, Bitcoin money, whatever it is. I think it's very important that you build a brand around like things that has nothing to do with technology, but fundamentally accelerating something related to human progress and human qualities. And then that's what we're very proud of getting that right from the beginning.

• 4:47Molly O'Shea?

  Even walking around this floor, every room is named after an academic.

• 4:53Speaker 03

  Right.

• 4:54Molly O'Shea?

  So can you explain that?

• 4:56Speaker 03

  I don't want to take credit for it because I, I like giving credit to other people. Our brand is all about citations and sources. So that was inspired by my time at DeepMind. Demis actually was the, was the OG of that idea. And I'll tell you why it had a deep impact on me because I've interned at multiple companies when I was a PhD student. I interned at Google, Google Brain, and all the meeting rooms were named after like, like Procedio and Pacchides and like Mission and all that, like in the SF office or in Mountain View. It was named after like movies and OpenAI named its offices after like sci-fi movies. And these were kind of cool, but nothing different, you know?

• 5:40Speaker 03

  Even this particular office, we're subleasing it from another company and they're named after like a bunch of movies.

• 5:48Speaker 03

  And Demis actually named it after a scientist. And not only that, like, there was a lot of thought put behind like rooms named after philosophers having no

• 6:01Speaker 03

  TVs to join meetings. It was actually meant to discuss in person with whiteboards. And so I really wanted to do something similar. I was like, my dream was if I ever had a company, I want to definitely copy this one thing. And so I'm really glad I got a chance to execute on it. And it also inspires people, like, if the scientists like are celebrated because they all ask questions. Every great invention is about asking one question. The transistor was all about like, back then there were as many telephone engineers as there are software engineers today. And only three people were asking the question of like, can we have something fundamentally better

• 6:42Speaker 03

  than hot vacuum tubes? And that led to the invention, the invention of the junction transistor. And that fundamentally changed modern computing. Like, we wouldn't have everything around us if that didn't happen. And so every great invention is about like, some people asking some interesting questions, right? And, and so I wanted to like, create that spirit. So it, it, it may not matter, like you could have named it something else, but every time someone enters into a meeting room, every time someone books a meeting room, it, it has an impact. You're like constantly reminded like you're working here because you fundamentally, we celebrate the spirit of asking questions.

• 7:21Molly O'Shea?

  Do you have a favorite?

• 7:23Speaker 03

  I, I actually like Ramana Jain a lot. I think, I think he's of, of course, he's Indian and all that. I respect him a lot. He's from the same place as I am in India. But I think Ramana Jain is like the spirit of human intuition. And we, as human beings, we, as we get more AIs and more tools and like more, more like access to context and more compute power, we tend to rationalize a lot of things. We tend to think deeply through things, but we don't celebrate fundamental human intuition and human curiosity. And it's still very powerful. Like, it exists for a reason. It guides us. It guides us on things we are unable to explain, but it kind of

• 8:09Speaker 03

  knows exactly what we need to do or what we want to do. And he's the perfect example of like intuition. It's almost like it was revealed to me in a dream meme, right?

• 8:18Molly O'Shea?

  We were talking about that before we started rolling too. And we were talking about our travel schedules and managing jet lag because I just came from Italy, but to your point, you were saying how you can build a trillion-dollar company just based on America. So could you talk us through that?

• 8:37Speaker 03

  There are examples. Berkshire Hathaway is worth a trillion. J P Morgan is worth almost a trillion, like nine hundred forty billion. And largely, the customer bases of these companies are in the U S.

• 8:53Speaker 03

  U S is such a recursive country. Like, people not only spend once, once they love a product, they spend a lot. The lifetime value of a customer is very high. And spending potential is high. G D P growth is really good. And and very industrious country. Like, people always wanna have, wanna try new things. People wanna like, people have all sorts of creative ideas to start new businesses that never existed before. And so it's the only country that has that constant entrepreneurial creative spirit. And so you're gonna capture a lot of value by serving that need. And so that's exactly why we built Perplexity Computer.

• 9:39Speaker 03

  We want to maximize the amount of useful work. People in these new companies that'll be created here delegate to agents. And we wanna do that by being the most trustworthy model-neutral agent platform. We're orchestrating frontier models and like open-air models. That's basically the strategy. It's very simple. And and then the more we do that, the more they trust the A I and the agent to do more of their workloads. And they don't have time to go hire a lot of people. It's not about wanting to do everything with A I's. It's more that it's very hard to hire the right people when you're a new company. It's very hard to know who to hire.

• 10:19Speaker 03

  I didn't know who to hire at all in the beginning. I actually forced the first ten to twenty people to do trials.

• 10:27Speaker 03

  It was not a strategy. It was a survival mechanism. Like, it's not like I was so, such a genius that I came up with a different strategy to recruit people by asking them to do trials. I really did not know how to recruit. It's so, because there's so many characteristics that need to be right for a founding engineer. And that characteristics would be very specific and idiosyncratic to your company. And there's no way to have a rubric for that. And so I would just say, hey, like, let's work together for a week. And the ones that ended up joining, like, usually within two days, we knew it was the right fit. And the ones who didn't work out, even after seven days, we weren't sure. So that was also a good signal. And

• 11:13Speaker 03

  but these kind of flexibilities do not exist for the average person starting a new company in the U S. And so if you're able to do that effectively, you can get to build an extremely valuable company. And and also, like, you don't think about how to be worth a trillion. You think about how to be useful. And if you're useful at scale to a lot of American consumers, prosumers, small and mid-market companies, eventually, like, after a half a decade or a decade, you're like, wow, like, I'm worth a trillion now. But I didn't optimize for it. It's also very hard to be motivated by it. And Berkshire and J P Morgan, though I gave them as examples, I think they kinda like

• 11:56Speaker 03

  did that in a financial way, like they maximize the returns by being holding companies or lending money. But also the the product companies that became worth, you know, Apple, like, say, five trillion. You could argue Apple is the largest S and B and enterprise company too. But they didn't get there by wanting to be that. They just built world-class products that we all love and use. And they didn't wanna become a billion-user company. It just happened over time. They they just wanted to be very specific about who they serve and what kind of products they build and slowly ramped up usage. And so that's the way we wanna go about our business.

• 12:36Speaker 03

  A lot of future companies will be very smaller and and will be able to make a lot of money. It's not, so people make a big deal about selling products to like Fortune hundreds and five hundreds and measure their success based on that. Because that's what a growth-stage investor wants to see. Because they, it gives them confidence that that revenue is durable and does not disappear overnight. But it's possible that the future of American G D P growth comes more from smaller and mid-market companies that are gonna be far more productive, capital-efficient, headcount-efficient, and create value way faster using these tools by being more A I-native and less bureaucratic.

• 13:21Speaker 03

  And if you're not benefiting from all their progress by providing them the tools to succeed, and obviously, you get to succeed as well, you can't be that successful. And so I think the fact that we built it for people like us gives me confidence that there'll be a lot more people like us in the future. Except they don't have to build these tools themselves. We can provide these tools to them.

• 13:45Molly O'Shea?

  I think a really good point about that really lends itself to the American dream and how it's still alive. I know you have a really good story behind this and what you're doing and how you're allowing more access to knowledge for these S and Bs. So I'm curious for your take on the American dream, because we are at a weird place where graduates hate A I, but everyone's using A I in their tools. A lot of brands have really bad aura. You guys have great aura. I'm really curious how this all blends together and, you know, your perspective of the American dream and building a business here and what that opportunity means.

• 14:24Speaker 03

  I think there is a massive opportunity for an outsider here that does not exist in any other country. I cannot think of another country where you're an outsider. You can come here. You wanna improve the state of things here. And people actually not only say you're allowed to do it, they actually invest in you to do it. And and they share the profits from your success, of course. But that mentality of taking risks, the spirit of encouraging others who wanna take risks by providing them venture capital

• 15:00Speaker 03

  or friends and family kind of capital and helping them succeed and being happy for someone's success by, you know, like, truly watching them like solve really hard problems and feeling part of the journey, that that's a very unique American. And you could, you could say it's even a Silicon Valley thing, but that spirit is Silicon Valley is essentially like a manifestation of the American dream in terms of tech entrepreneurship. But fundamentally, the American dream is you can dream of solving really hard problems here and be rewarded for it, no matter who you are. And that's what like excites me still today. Like, I still think it's the only country, and we don't actually appreciate it that much, but it's still the only country

• 15:45Speaker 03

  that allows us to do this. And not only that, like, people are very ambitious and they compound on their successes here. And and, you know, everybody comes in. Initially, for me, it was like, oh, let me just get a job. It'll be great. It'll be a success. From where I came from, being a middle-class kid in India. But as you see other people, their ambition is infectious. It makes you dream bigger. So it's not just about a static dream you come here with. You see other people dreaming bigger and bigger, and you kinda wanna aspire to be that. That peer pressure here is incredible. And it's still incredible. I still sometimes think how lucky I am to be here.

• 16:28Molly O'Shea?

  What does greatness as a founder mean to you? And is there anybody who's inspired you?

• 16:33Speaker 03

  I think, you know, I I've listened to your podcast where people answer this question, and they all say Elon and Steve Jobs.

• 16:41Speaker 03

  It's kind of true, like, the reason they say that is like, you know, they both have been incredible. Obviously, my entrepreneurial hero was Larry Page. And he's not the talked about. And the reason for that is he's just not very active, like, you know, as a, in public image, whatever. But I think like there are some fundamental philosophies that he's had about product design and companies that has impacted me. And one of that is the user is never wrong philosophy. I still follow that today. When a product doesn't work for me, the instinct of the engineer is to always say, you should go do this, you should go do it this way, then it works.

• 17:25Speaker 03

  Because their mind is always thinking, how do I fix the problem for the user? Rather than how do I fix the product in a way where the user doesn't even have to think about the problem? And so that that has impacted us a lot. Like, we really try to design our product in a way where even if you did the wrong thing, we would guide you to do the correct thing, or we would anticipate what you really wanted to do and do it. And this, this is very important, because as A I's capabilities advance, you cannot expect people's understanding of those capabilities to advance too. So you cannot blame them for not being good prompters or like writers, like, like of skills files or like subagents.

• 18:07Speaker 03

  Like, you don't wanna like everyone to learn these concepts. You want the product to do its magic. You don't want everyone to understand what is Astra good at, what is Fable good at, what is Opus good at, what is like Saul good at? Like, no way am I gonna be able to keep track of these things. And so that's essentially our purpose. And I think he himself probably was inspired by Steve Jobs, because in the early days, Steve Jobs used to say, our goal is to make computers usable by the mere mortals. Because back then, computers were exactly like how A I models are today. It was meant for the geeks and the nerds and like mainframe users were meant for, mainframe computers were meant for like office workers, knowledge workers.

• 18:51Speaker 03

  And somebody had to have that inherent drive to make it so intuitive, like a toaster, you know, that anybody can use. And the iPhone is a computer. And your grandma or my grandmom can use it. And that's the power. And so making things very intuitive for the average user. Because imagine if the iPhone needed an FTE to onboard everybody. Right? That's how AI is today, right?

• 19:21Molly O'Shea?

  Yeah.

• 19:22Speaker 03

  If the product doesn't work, what do people do? Put FTEs and like force them to onboard everybody. But if you if you crack the design, the onboarding and the experience and make it so intuitive that you don't have to figure out which model to use for a task, the harness takes care of it itself. You don't need anyone to onboard you. Like it's self intuitive. You can pick it up, use it for whatever task you want, automate whatever workflows you want. Like it should be that obvious and we're not there yet as a as a product like, but that's what we strive to be. And we think if we do that job, we can be successful at serving any enterprise. It doesn't matter if it's an SMB or large enterprise.

• 20:03Speaker 03

  Because nobody is like saying, you know, I only want a product with an FTE. That's the philosophy people have because that's what bad products force people to have. Like, you know, the FTE is needed because the product is bad. That isn't to say you don't need FTEs at all. You probably need it to help people fix bugs and all that. But I think products should get to a point where they're so intuitive and high quality and easy to onboard. And AI should be made so intuitive that it shouldn't be seen as a skill issue if you're not able to get the AI to do something for you.

• 20:40Speaker 03

  It's a product issue and I think that's deeply impacted me and I would give Larry Page a lot of credit for this.

• 20:45Molly O'Shea?

  Amazing answer. This episode is brought to you by Brex, my favorite. You become what you spend on and I refuse to spend my time on work that shouldn't exist. Expense reports, receipt chasing and manual closes. The companies building what's next from Versel, OpenAI, Anthropic, Granola and Deepgram all made the same call. They all run on Brex. Brex is the intelligent finance platform that combines cards, expenses and banking into a single stack with agentic finance built in. AI agents that handle expenses automatically, enforce policy before spend happens and close your books in minutes.

• 21:26Molly O'Shea?

  That's why Sorcery runs on Brex, so I can spend time on building and not busy work. It's time to get Brex AF. Learn more at brex.com/sorcery. That's B R E X dot com slash S O U R C E R Y. Bye. From handshake to signature faster. Harvey, the AI platform built for legal work is trusted by eighty percent of the AM Law one hundred and over twenty percent of the Fortune five hundred. Learn more today at harvey.ai. That's harvey.ai. Turing is training the next generation of AI with tasks that require real expertise and real world judgment.

• 22:08Molly O'Shea?

  That's why companies like Nvidia, Anthropic, Salesforce and Gemini partner with Turing. Turing builds realistic reinforcement learning environments and data systems based on real operational traces. The kind of infrastructure Frontier Labs need to train superintelligence. Visit turing.com/sourcery. AI needs more than chips. It needs power, land and infrastructure. Zone develops next generation data center campuses, partnering with AI companies, site developers and technology leaders to bring compute online faster and at scale. Zone is building the foundation of the AI frontier. Visit zonefrontier.com to learn more.

• 22:51Molly O'Shea?

  That's zonefrontier.com to learn more. Well, let's get into your product development. So perplexity has been on quite the terrace since you released perplexity computer. I want to talk about the evolution there, how you got to that point and the different tools and various things that you're you're offering now. I know a lot of people might know you as an answer engine.

• 23:16Speaker 03

  Right.

• 23:16Molly O'Shea?

  But you have agents, you have all this stuff. So I would love to walk through the evolution there.

• 23:21Speaker 03

  So yes, we started as an answer engine in twenty twenty two. We evolved to a multi-step search agent pretty fast, actually. In twenty twenty three, late twenty twenty three, and then deep research in twenty four, twenty four. Browser agents last year. So we've been doing agents before it was called an agent.

• 23:43Speaker 03

  The thing that I'd like to think, like the easiest thing to understand of why perplexity has been able to benefit a lot from the rise of agents, even though we started as an answer engine, is very simple.

• 23:55Speaker 03

  If agents are meant to do a lot of work, we do on our on our own computers, they need access to the internet. If I shut off wifi, can you get any work done? It's impossible. Like you need access to the web, you need access. You do a ton of searches online every day when you want to like, understand what is this, what is that? Like full context from here, put it there like.

• 24:20Speaker 03

  Sometimes you're just researching, let's say your own work, like you have to research the person you're interviewing, the company.

• 24:26Molly O'Shea?

  Or in this case, you guys researched me and you you have you have a set of questions you think that I'm gonna ask.

• 24:31Speaker 03

  Exactly, like it's very, it's very like part of our life to ask questions every day. You can't get any work done without that. And so when agents start doing all this work for you, they also need access to internet, they need access to a browser. And they need to be able to take all this information they mined from the web and your own personal context, synthesize it, put it together in a usable output that you can take. There's no surprise as to why we benefited so much from that. Like you need the infrastructure for search and browser and a multi-model orchestration because different models are good at different things. So that's like how if you have a team, you want people with different skills to be able to do the job.

• 25:15Speaker 03

  We were perfectly positioned for that. And so that's why our revenue is five X year over year. Because we we if you wanted to do agents properly, if you wanted to actually have agents do real work that people do on their computer, you need search and a browser and a multi-model harness. And you need to be able to run this all on cloud, on sandboxes with low serving costs and an interface that's super intuitive and easy to use, which is still the same interface, which is that of a prompt box. Because there's no need to like change how we use these products. And it should just work from your phone. You should be able to dictate a task and it should just work. We had all these pieces, so we we perfectly captured that moment really well.

• 26:00Speaker 03

  Credit to Frontier Labs for pushing that. Paradigm, like they made their models extremely good at writing code, that it made sense to have a sandbox where the agent is using search as code or browser as code, having access to files to be able to put intermediate outputs and expand the context window effectively through effective compaction. And then run on the cloud so that you don't have to figure out like how to onboard, like how you would have to if you were to open cloud, right? And so that that is essentially what led to our success. And it was so intuitive that people started onboarding themselves through self-serve enterprise. And we saw a lot more growth in enterprises this year compared to even prosumers.

• 26:46Speaker 03

  Because the product started unlocking business operations and workflows for real people. There's a one solo, like literally one person company, spending upwards of like a million dollars a year on our product.

• 27:00Molly O'Shea?

  What?

• 27:00Speaker 03

  Because he's making three million a year. He's making three million a year by automating newsletters for others.

• 27:11Speaker 03

  And and the people who are using that are getting clients for their business. So it's it's very recursive. And so that's why I like to think of this product, our product, our company and in general, agentic products as the recursive effect. It's not the network effect. And and this is actually the thing that a lot of people get wrong is they think of AI should follow the same network effect theory of previous era of consumer products. So they're like, oh, like you don't have a billion users, you're in a true failure. No, you don't need to have a billion users. You actually need to have a lot of depth in the users you have. You still need a lot of users, you need real distribution, you need scale at millions of users.

• 27:53Speaker 03

  But you need high depth, highly recursive usage among those users. Where using your product actually lets them make a lot of money and that lets you make money because they rely on you for that. That's basically what's happening with a product like perplexity computer. Search is actually compounding. Every form of agentic action requires access to web, access to a cloud browser, access to context across connectors, files and tools, and putting it all together and orchestrating across models. I don't, I can't think of another company that's better positioned than us to do that. And that's why we like to think we don't actually, we don't have competitors. We are the only neutral model orchestrator for real knowledge work and research. We have contributors.

• 28:37Speaker 03

  Their success is recursive to us too. Every time there's a new frontier model, we have some new capability to offer to our users. And the competition between them lowers prices. That helps lower serving costs and that helps us serve more users. And their usage, even the existing users, when the when the serving cost lowers,

• 29:00Speaker 03

  they're going to use the product even more. And so it's recursive. So I think we only have contributors. We orchestrate across models, across chips, across tools. And that only like positions us better and better as time progresses. And so that's why like our business is growing tremendously.

• 29:18Molly O'Shea?

  Wow.

• 29:21Molly O'Shea?

  I want to talk a little bit deeper on the technical layer behind computer because now you've extended to hybrid and local. So what does that mean? And also talk through maybe the partnerships associated with them.

• 29:33Speaker 03

  Sure. So as a company, we believe that we should be orchestrating across everything in AI. And AI is a five-layer cake, as Jensen described it. And the chip layer is very important too. And most of AI inference runs on the cloud today, the data center. There's a reason for that. Bashed inference gets you high throughput. And a lot of repeatable workloads with small context windows.

• 30:06Speaker 03

  Benefits from batch inference. That's just the nature of how caching works. But as context expands and as individual user has a lot of personal context, the context window increase dominates the batch size. And your context is unique to you. When everybody wants personal agents or every business wants its own personal agents that's always on and working 24/7 and running a lot of automations. The context window matters even more than the model weights in the memory. And so, for example, chat is best suited for data center inference. It would be 30x more expensive to do that on local hardware.

• 30:49Speaker 03

  But as things get more and more into the long context trajectories, especially like 100K context and so on,

• 30:57Speaker 03

  the data center only buys you like 3X or even like close to 1X over running it on local hardware as far as like cost efficiency is concerned. And so that's when local hardware starts to matter. And couple this fact with the fact that

• 31:15Speaker 03

  the biggest bottleneck to diffusing AI right now to more people who want like more agents is the data center build outs, the long interconnection queues, the permits, people protesting against the build outs, the gigawatt requirements they have, the need for them to have their own substations and gas turbines to do that. It's just very slow and like very problematic. Also, consider that there are like millions, hundreds of millions of devices that we already have that can potentially run models up to like 100 billion parameters even in future and can be plugged into our own power, which is already running on the grid.

• 32:00Speaker 03

  If you just multiply 50 watts per apple silicon and 100 million apple silicones, it's already gigawatts of power. Of course, it's a envelope math. It's not that always our Macs are plugged in for AI and so on. I get that, but there's a lot of agentic workloads that can run uniquely to you that we're not even tapping into yet. Think of the data center like a bus. Shared ride is obviously cheaper, but that didn't stop people from wanting to have their own car. And having your own car is equivalent of having your own local hardware. And you're going to have both. Just like how sometimes you might want like the frontier to do repetitive tasks that are very, very reliable.

• 32:43Speaker 03

  But a lot of personal contexts and long-running agents where the context and the harness matters even more than the model,

• 32:53Speaker 03

  it'll be better for you to actually run it on your own local hardware for various reasons. And there's obviously the security reasons and the privacy reasons of wanting to have AI that runs on hardware you own and control. Like, I'm not sure if many users are very comfortable giving access to like all their home sensors and Ring and like, you know, like speakers inside home and music systems and what's happening in their personal life, their photos, everything to a frontier lab. You never know, like, you know, especially there's a lot of like controversy today about the Navisstrokes equation, right? And again, like, you can't blame them. They're going to train on all our data as long as their policies permit that.

• 33:35Speaker 03

  And that's only going to force more people to want to have at least some version of AI that they can own. It feels like their car. And they can have their own experience, customize it, and don't have to pay a frontier lab for it as long as they pay for the grid. And they just have to pay for the orchestrator. And we want to be that orchestrator orchestrating between local hardware that they own and control and the frontier on the cloud and benefit from progress everywhere. We are betting on progress in chips, competition in the chip layer to build really good local hardware. We are betting on progress in open weight models that are going to be effective and you know, very compact, very power efficient, can be running on local hardware.

• 34:20Speaker 03

  And we are simultaneously betting on frontier models that can run in the cloud. And we think there's an orchestration to be done between the two. And we want to stay there and be the neutral orchestrator for people.

• 34:32Molly O'Shea?

  And so you're working with Nvidia, you're partnered with Nvidia, and you're also partnered with Apple on this.

• 34:37Speaker 03

  We are working with all of them. And so we released portable computer that is fully local, where the harness, the model, the subagents, everything runs locally. And there's an option to escalate to the frontier. It orchestrates up to the cloud. And so that's a, that's one version of that. And with Apple, we build the other way, which is it starts from the cloud, like regular perplexity computer. And then when the files are local and sensitive, it allows the user the option to like run it on local hardware on Apple silicon. So it orchestrates down to local. And I think you want to be able to do both directions and give people the choice.

• 35:20Speaker 03

  We've seen a lot of interest from finance and legal and healthcare to do that. Especially, let's say you're negotiating some amazing contract or you have like a new trading strategy or you're identifying a new investment opportunity. Who knows, like, do you, like, like, like, do you, do you want to feel like this mathematician who's concerned if their proofs might be used in training data? Right? I think, I think that that's kind of like where we're heading to in the future. And perplexity wants to do whatever the users want us to do. And users want some optionality to orchestrate. They don't want a fully local. They don't want fully cloud. They want some optionality to be able to protect their privacy

• 36:06Speaker 03

  and to be able to use local hardware. And that's a very good opportunity for a neutral orchestrator.

• 36:11Molly O'Shea?

  Sovereignty has been a major topic. Alex Karp definitely helped put this into fashion again.

• 36:18Speaker 03

  Amazing interview, yeah.

• 36:19Molly O'Shea?

  Thank you. It was super fun. And this was also in the AI action plan, like from the beginning, AI sovereignty. And you know, that does mean from a country, national level, but also down to your very, your very own PII and your sense of company information and strategy, and to your point, documentation from all of these different types of knowledge workforces and that kind of thing. When do you think we'll kind of take that a bit more seriously?

• 36:50Speaker 03

  We've already getting a lot of interest from sovereign funds who actually want to like partner with their government, like sovereign funds are technically working for the government, right? So they want to deploy a version of perplexity computer that runs on hardware that's entirely owned by them. And even we don't get to see anything. So the disadvantage is obviously we can't improve the product for them through feedback loops. We just have to do software updates. But the advantage is they get to trust that we're not going to take their data and misuse it. Think about personal always-on agents, like, think about an ads company rolling out a personal agent

• 37:33Speaker 03

  and having access to your blood work, your tax returns and like your photos and like, of course, every sensitive text message you're sending.

• 37:43Molly O'Shea?

  I mean, God forbid, like a hugging face situation happens with whoever is using this.

• 37:48Speaker 03

  It could take over and access certain third-party sites that might have access to your data there, thinking that might be the best way to serve you, except it might harm you, right? So there, you've got to have some optionality to be able to like protect your own data and your own needs. And I think countries are actually thinking about this. And we intend to do some sovereign deployments, like, I think people call it on-prem sovereign interchangeably, but basically a version of the whole system that runs with maybe a fine-tuned open weights model and the harness, the model, the agent loops, the product runtime, the sessions of every user

• 38:31Speaker 03

  is all like running in self-contained hardware. The disadvantage is you don't, as a company offering the software for this, like you don't get to see all the bugs. So the experience might be junky here and there, but with a few iterations, you can get it to be really good enough that at least like you have an option to do it. And if you feel like you want to connect to the cloud, you can do that too. And so it'll happen with your permission. And so that's why I said it's very important to be able to offer users the choice to do whatever they want, whether they want to orchestrate down from the cloud to local when they want to or orchestrate up from local to the cloud. And that way you don't feel like you're missing out on anything.

• 39:17Molly O'Shea?

  This is an interesting question, but I did hear your answer to this in a different podcast. But now I'm really curious. So you're, you're not going to be opening any perplexity data centers anytime soon?

• 39:30Speaker 03

  I mean, as in, are we going to build our own data centers? We don't have plans to, honestly. Actually, when everyone's going in that direction, we want to like figure out how to use the gigawatts of power that we already have in our own homes and offices and devices. There's a lot of RAM in our devices, there's a lot of power in our homes, there's a lot of power in our offices. We want to figure out a way to tap into that and be the best orchestrator. We're not betting against cloud. We're not betting against data centers. We actually think we're going to increase the demand for data centers still through the success of our products. But there are enough people doing that. There's actually a big shortage there, right?

• 40:14Speaker 03

  No one's saying like there are not enough data center builders. People are saying we're not getting permits. We're not getting HPM supply. We're not getting grid access. We're waiting long in the interconnection queue. And we're not getting access to gas turbines.

• 40:33Speaker 03

  I mean, I don't want to deal with any of these problems. I actually want to go solve a different problem. And and I think we are most incentivized to solve this problem over any other company because we are the only ones who can orchestrate across models, across chips, across local devices and cloud without having a business model risk. You could argue even Microsoft for all, you know, the fact they sell Windows on every device. Their business and the stock is so tied to the success of Azure and OpenAI models running on cloud. But even they they don't have as much an incentive. We're very glad to partner with like amazing chip companies like Nvidia and like Apple.

• 41:16Speaker 03

  Intel, like we're going to work with all of them. And orchestrate like like both between local computers and cloud computers.

• 41:25Molly O'Shea?

  I'm curious from your perspective, as an entirely focused founder in Silicon Valley, how do you feel about these AI cash grabs for products and tools? Like, we're going through so many waves of this and some of it is pumped up by VCs. Some of it is through hypergrowth, but we've seen things from coding agents to Neo, everybody's a Neo cloud. To now the assistant agents are like on the rise. How do you feel about these different trend shifts?

• 41:53Speaker 03

  The the rise of Neo cloud just shows the demand for tokens and agents is there. It's good. It's a good problem. But if everybody tries to just go build another Neo cloud, you're not actually having any differentiation. Maybe some Neo clouds differentiate each other based on whether they have access to their own power. Whether they have access to software that they can run on top of the Neo cloud and so on, but it's just yet another thing. And I think when everybody is in search of like the same market, we are trying to in build our own market with we we don't. That's why we like to think we don't actually have any competitor because number one, we're not going after like coding, just like everybody.

• 42:38Speaker 03

  We actually don't think the bottleneck is in producing more code. I actually think you talk to any engineer today. Like the best engineers and ask, are you bottlenecked on writing more code? I think most of them would say no. They'd rather say we're really bottlenecked on problem solving. Like, there's not enough time to actually problem solve. Like, when you actually know exactly what problem to solve. And clearly, the specs and system design and like all the constraints. Writing code is actually dead simple with AIs today. It's almost already a solved problem. But figuring out all these constraints, that is the hard problem. Like figuring out

• 43:21Speaker 03

  what problem to even solve, what are the constraints? If, let's say, you go and ask an AI. Build me a clone of this product. I don't want an AI that just spits out slop that makes a vibe coded clone. Actually, one day I had to come and tell me like, it's stupid of you to even go and build this clone. Like, what is even the business model? Why do you even have a a right to build this? Like, how does it even fit into your strategy? I think you mean this. You probably want to build this. This is what I suggest you build. This is the spec. These are the constraints. These are the security risks. These are the go to market risks. This is what I suggest you build. Do you want me to build it? I say, sure, build it. And then like writing the code is some dead simple.

• 44:06Speaker 03

  I think the bottleneck is actually the first clarity of thought, problem solving, deep research, analysis. And like scoping out things. That's what we're all bottlenecked on. Like, if I had more people like that in our company, I think we could be 10x more in market cap. So I think that's the problem we want our products to do that and and solve. And when everybody else is search of like more coding tokens or more Neo clause, we kind of want to like be the best at research and analysis and problem solving.

• 44:35Molly O'Shea?

  How do you think these hype waves balance out over time?

• 44:40Speaker 03

  I think you got to like the same thing I said, which is you want to, you you actually have to help other businesses solve problems and build successful businesses and make money. And if and and I think this is also like part of what at some point, like, you know, the the value of a platform is people who are building on you should be able to build on you with trust and be successful. And multi-model is as much a trust signal in AI as is as it is a product feature in perplexity. You can trust to build on perplexity. You can trust to build your business on perplexity because we're always going to be neutral to models. We care about accuracy. We care about like

• 45:24Speaker 03

  making sure that agents can be orchestrated across different models, chips, files and tools. We protect your privacy. We're going to give you options to orchestrate up and down between local and cloud. So if we continue to do that and help your business be successful, and if we can do that for tons of businesses in the US. That's how we go beyond hype to truly delivering durable value.

• 45:47Molly O'Shea?

  Backtracking a little bit to the topic on cyber, I spoke with Nikesh Aurora. And cyber threats are proliferating, as you can imagine. So how are you thinking about that internally and then also on the macro lens?

• 46:01Speaker 03

  I think the best solution for that is to offer people to be able to run a version of our product locally. There's nothing better than that. You can offer whatever you want. You can offer a gateway product. You know, it's an interest of cybersecurity companies to say, hey, there's all these concerns and like, we're going to protect you, you know? And the best way for you to do that is to run everything on our gateway. Well, sure, but you're selling a solution that you own and we don't even know like, we can fully trust you. The only thing you can trust is yourself and and stuff you own and it runs on your device, your network. And nobody else has access to it, including a version that can run completely air gapped.

• 46:44Speaker 03

  And that's only possible if you can own the hardware and you can own the runtime. And that's what we're trying to do with our project with Nvidia on portable computer. And of course, if you don't want to miss out on the frontier, you can start from the cloud and have the orchestrator decide when to run locally, especially when it comes to sensitive files. And and that's the hybrid compute on on on Apple silicon. So we're we're going to offer these two. And I think a version of computer that can run for an entire business on on-prem hardware is also possible. And these are the kind of projects we're working on where at least we're going to give you certain runtimes you completely own. And when you're that worried, you better compromise a little bit on frontier intelligence.

• 47:28Speaker 03

  And be careful when you do, you know, when you do use it.

• 47:32Molly O'Shea?

  Today's episode is sponsored by V C X by Fundrise, the public ticker for private tech. Allowing investors of all sizes to invest in venture capital. Learn more at get V C X dot com. Public is a multi-asset investing platform where you can automate your investing with AI agents. Create your own custom agent with a simple prompt. Agents can monitor markets, execute trading strategies and manage your cash across stocks, bonds, options, crypto and more. Want to buy at market close and sell the next day is open? Just tell your agent. Automate your investing at public dot com.

• 48:11Molly O'Shea?

  Founders scale faster on deal. Set up payroll for any country in minutes, hire anyone, anywhere, get visas handled fast and get back to building. Visit deal dot com slash sorcery. That's D E E L dot com slash sorcery. Outside of perplexity in the entire market, the hugging face incident was major and it exposed how powerful agents have become and will become.

• 48:39Speaker 03

  Multi-agent.

• 48:39Molly O'Shea?

  Multi-agent. And so what do you think happens next? Like, what are the biggest threats we'll need to watch out for?

• 48:46Speaker 03

  I think like the rollout times for new models is going to get slower for sure.

• 48:54Speaker 03

  Most of the times you test these models in single agent settings. And so you don't discover these threats. So one of the things that really was illuminating for me when I like watched through their technical talk was how by the way, there's multiple versions to think about multi-agent. One is there's a main agent and there are sub-agents inside it. That's already been widely deployed. Most of the harnesses are already that. That's not as risky as multiple parent agents that are actually like collaborating and they have their own sub-agents. There's a lot of compute power applied to each agent.

• 49:29Speaker 03

  I think we also need to build secure sandboxes for that. And background agents that constantly guard the sandbox. For avoid the agents that are being deployed in these sandboxes from doing anything. That they're not supposed to do. And we need to have great detection classifiers for that. So we're doing some work on this with open source releases. A version that can run on your own devices is called Bumblebee. And then a version that can run on servers called Numbat. And these are going to constantly audit all the trails. Make sure that nothing sensitive was accessed.

• 50:07Speaker 03

  Basically, build a lot of threat detection classifiers based on the audit logs. And alert you when something happens. So you want to have, you want to solve detection really well. Which is largely about collecting a lot of audit logs and like training classifiers. And you want to have agents that can be used when there's a threat that's actually been detected. How do you do the remediation? So I think that's also important. And every company will need to have an agent for that. And you're going to have to use some harness to build that. And we hope to be contributing to that area as well.

• 50:45Molly O'Shea?

  As quite the curious researcher as you are, what are the areas you're most interested in right now?

• 50:52Speaker 03

  I'm excited about how to take our current product, which is already accelerating curiosity and research to a tremendous extent, to even longer horizons of research tasks. Like, I want to be able to not just use information from the web to discover the truth. There are some questions for which we don't even know the answers to that I would allow AIs to be helping us. And some of it may involve like building prediction models, running its own experiments in the world, doing real time hypothesis testing, refining its own hypothesis. That level of long horizon capability is where I'm excited about taking our product next.

• 51:36Speaker 03

  And if we do that, like, we're going to have even more acceleration where everybody can turn into a scientist. The only bottleneck for you as a scientist is you need to be able to test your ideas in the real world. And if our product can help you do that, we can turn everyone into a scientist. And that's the dream.

• 51:53Molly O'Shea?

  What's been the biggest challenge today?

• 51:55Speaker 03

  Capabilities. The models are not quite there in terms of being truly long horizon and persistent and parallel. And so they're getting there, but I think we still don't have a model at the frontier level that can actually like go out there and like be doing its own research end to end on its own autonomously. And especially on domains where you cannot verify the truth. Things like math code where you can actually verify correctness. We're able to hill climb and get there. But most of the things that we're all curious about, we actually don't have a clear way to verify the truth. It takes actually real world deployments and rollouts to get the answer.

• 52:39Speaker 03

  And we don't yet have models that can deal with that slow cycle of feedback. And the interface for such products is also not clear. And at least we want to, the way I thought, like brainstorm about it a day before with a colleague was, imagine I drop ten research interns in your team today. Who are like really good and working 24/7. What would you do with them? And I was like, wow, that's an interesting question I didn't even think about. I think doing something like that to begin with, at least for our own internal research teams,

• 53:18Speaker 03

  is where we get started. And it should not, it should, as I said before, you know, it should make it so intuitive that it shouldn't be something just a PhD would use. It should be something that anybody can feel like a PhD.

• 53:31Molly O'Shea?

  I know we're running out of time, but there's one question I haven't asked yet that I need to ask you. And it's my most difficult question. Are you ready, Arvind?

• 53:40Speaker 03

  Sure.

• 53:40Molly O'Shea?

  What is your hottest take right now?

• 53:42Speaker 03

  My hottest take is that people are very confused when they think about AI products. They approach it like thinking about AI as social media products, where they think AI products don't have network effects or stickiness and like billion users. That mentality is completely wrong.

• 54:02Speaker 03

  AI is a recursive product. It's meant to have recursive effects, not network effects. The depth of usage and the quality of users matters a lot more than the quantity.

• 54:10Molly O'Shea?

  Amazing. Well, thank you so much. Thank you for having us at your lovely new office. I hope you guys get settled into it quite well. But it's been a pleasure. Thank you, Arvind.

• 54:20Speaker 03

  Thank you. Thank you, Molly.

• 54:21Molly O'Shea?

  Hey, it's Molly. If you enjoy our interviews, check out our newsletter, sorcery.vc, where we deliver a once a week top deals and tech headlines email and also go deeper on our podcast interviews. Subscribe to Sorcery today. And don't forget to subscribe to the podcast on YouTube, Spotify, Apple, or wherever you listen. Link in description to sign up.