'The data center is coming to your laptop,' says Perplexity CEO

[00:00:00] Speaker 1: We believe that the future is going to be a hybrid of server and local for agentic inference. The data center is coming to your laptop. People are tired of token maxing and paying half a billion dollars a month on, you know, whatever frontier model tokens that are running on the server. I think what you really want long term is really high token value per watt per user. And I think the best way to do that is to orchestrate intelligence, accuracy, privacy, and costs all simultaneously. And that's an orchestration problem. And so that's why we built Perplexity Computer, which we think is an AI operating system. It takes objectives instead of instructions and puts together everything AI can do in one single unified system. And earlier, Perplexity Computer orchestrated across models, files, and tools. Now it can orchestrate across chips. That is the chip that's on your local computer, as well as the chips that are on the servers, like GPU racks. And that's the hybrid agentic inference that we announced yesterday with Intel. But we are chip agnostic and we are working together with everybody, including NVIDIA on RTX and Apple on Mac. So I think that's the power of Perplexity. It's a neutral model agnostic, chip agnostic orchestrator. [00:01:21] Speaker 2: And you also have a partnership for your personal computer, your agent product, for Microsoft now, if you can tell us more. So is that targeting enterprise customers? [00:01:31] Speaker 1: That's correct. So both enterprise and consumer. So we announced a bundle of Perplexity Computer integrations with all of Microsoft 365 software. So, first of all, there is another thing I need to introduce, which is a personal computer, which is a version of Perplexity Computer that connects your local files. It lives inside your desktop, so it has access to all your local files, your local apps, and so on. And we made a big announcement on the Mac ecosystem about a month ago. And obviously, a lot of our Windows users were like, we want this, we want the same thing. And so we first started off with Office 365 integrations, and then tomorrow we're going to announce a version of a personal computer for Windows. And so it will do everything that we did on Mac on the Windows ecosystem. And so that's what we are excited about. We're going to put out a wait list, and we're going to be launching it later this month. [00:02:27] Speaker 2: Okay, so it will be this month that you're looking to roll that out. [00:02:30] Speaker 1: The wait list is going to go out tomorrow, and the actual app will be shipped later this month. [00:02:35] Speaker 2: And, you know, like for companies like Microsoft and Apple, they have their own ecosystem or hardware. So, you know, for Perplexity, how does not having, you know, your own ecosystem or hardware impact the long-term viability of your product? Is there a risk that these big ecosystem players could favor their own AI products down the line? [00:02:58] Speaker 1: I think they absolutely will try to build their own AI systems, but we believe we are building the most versatile operating system by making it work across different models, across different chips, across different, you know, traditional operating systems, different hardware providers, different laptops. And that hybrid, neutral orchestration layer is what we are building, and that allows us to balance all these different objectives simultaneously. Now, for example, like, you would want to, like, specifically focus on Microsoft software inside Windows. But there's a lot of, like, desktop apps and hardware advantages the Mac ecosystem has that, you know, a company like Microsoft wouldn't be able to focus on. Whereas Perplexity can do all of that. Like, we can build models that can be very efficient locally on Mac Winnie's, as well as RTX, as well as Intel Arc. And all these, like, all these different work streams allows us to build something far more versatile than any of these individual companies that are focusing purely on their own ecosystem. [00:04:05] Speaker 2: Yeah, and also as more companies go deeper into AI agents now, I want to ask you about the competition and how you're differentiating yourself a bit more. So, say, if other companies were to copy your feature, what will still be hard for others to replicate? [00:04:19] Speaker 1: I think the primary focus of Perplexity since the beginning has always been accuracy and model agnostic orchestration. So, OpenAI and Anthropic are actually model providers inside Perplexity. So, anytime their models actually get better, Perplexity as a product also gets better. Just to give you, like, another way to think about it, Anthropic has made a tremendous amount of progress in their models since the beginning of the year. And what has that led to for Perplexity is that three times revenue growth since just the beginning of the year. In five months, we have tripled our annualized revenue, thanks to model advances that have been made by Anthropic. So, it might look like we're competing with Claude or something, but what's actually going on is every time there's a competition of the model layer between companies like OpenAI and Anthropic, or every time there's a competition on the local chip layer between Intel and NVIDIA, or any time there's a competition on the traditional OS layer between Apple and Microsoft, the biggest beneficiary of this is Perplexity because we are orchestrating across all of these different providers, all of these different ecosystems, all of these different hardware providers. And so, that's why, like, our business is going really, really well.