NVIDIA is reportedly in discussions to invest in Perplexity as part of a new equity round that could value the AI search startup at more than $30 billion. If completed, the deal would mark another major vote of confidence in the idea that the next big AI companies may be built not just around chatbots, but around products that can search, reason, cite sources, and increasingly perform work on a user’s behalf.

The talks are still just that: talks. Perplexity declined to comment on the report, while NVIDIA did not immediately respond to a request for comment. But the reported numbers are striking. A valuation north of $30 billion would represent an increase of more than 50% from Perplexity’s previous financing, which reportedly placed the company at $20 billion last year.

That kind of step-up is not simply a reflection of AI market exuberance. Perplexity’s business has apparently accelerated sharply in 2026. The Information reported that the company’s annualized revenue had climbed above $750 million, from less than $250 million at the beginning of the year. Even allowing for the usual caution around privately reported revenue figures, the direction is clear: Perplexity is growing fast enough to be treated as more than a promising alternative search app.

The company has spent much of its public life being described as an “AI-powered answer engine,” a label that is both accurate and increasingly incomplete. Its original appeal was straightforward: ask a question in plain English and receive a concise response with links to sources rather than a traditional page of blue search results. That experience turned out to be especially compelling for users who wanted a researched starting point, not a scavenger hunt across browser tabs.

Now Perplexity is pushing beyond search. A key part of its reported momentum has come from Perplexity Computer, a cloud-based AI agent designed to handle computer-based tasks for professional users. The larger ambition is familiar across the industry: rather than merely generate text, AI should be able to take a goal, use software tools, navigate the web, gather information, and complete multi-step work. That is a considerably more valuable proposition than answering trivia questions, but it is also much harder to execute reliably.

For NVIDIA, a Perplexity investment would fit a broader strategic pattern. The chipmaker is no longer only selling the computing infrastructure behind the AI boom; it is also becoming deeply enmeshed with the companies creating demand for that infrastructure. An investment in an AI company can strengthen a commercial relationship, secure a place in an important application layer, and offer NVIDIA insight into where developers and enterprise customers are taking AI next.

There is an obvious commercial logic here. Companies building agentic products tend to consume significant amounts of compute, especially as they move from single prompts to multi-step tasks that may call several models, browse websites, process documents, or run parallel workflows. The more capable the product becomes, the more infrastructure it may require. NVIDIA’s GPUs sit at the center of that economic equation, even when a company runs its workloads through a cloud provider rather than directly owning the hardware.

Perplexity, meanwhile, has been broadening its infrastructure relationships. Earlier this year, it signed a three-year agreement worth about $750 million to use Microsoft Azure, according to Bloomberg. The arrangement gives Perplexity access to Microsoft’s Foundry platform and models from providers including OpenAI, Anthropic, and SpaceXAI, while the company has said it is not abandoning its existing AWS relationship. For a startup operating at this scale, cloud capacity is not a back-office procurement detail. It is central to product performance, cost control, model choice, and speed of expansion.

The Microsoft agreement also highlights a more interesting feature of Perplexity’s strategy: it does not appear to be betting its future on one foundation-model provider. That could give the company flexibility to choose models based on quality, price, speed, or the nature of the task at hand. In a market where model leadership can shift quickly, building a product layer that can work across different models may prove more durable than becoming dependent on a single supplier.

Of course, a $30 billion-plus valuation raises the bar considerably. Investors at that price are not buying into a useful AI search tool alone. They are placing a bet that Perplexity can build an enduring consumer brand, persuade businesses to pay for agentic workflows, keep infrastructure expenses under control, and defend itself against companies with much deeper pockets.

Those rivals are formidable. Google remains the incumbent in search and has woven generative AI into its own products. OpenAI has the consumer reach of ChatGPT and a rapidly widening product portfolio. Microsoft can distribute AI across Windows, Office, Azure, and Copilot. Anthropic is gaining enterprise traction. Then there is the uncomfortable possibility that the features making AI search attractive – direct answers, research summaries, task automation – become standard capabilities inside every browser, operating system, and productivity suite.

Perplexity has another challenge: trust. Search has always involved a trade-off between speed and verification, but AI raises the stakes because a polished answer can make weak or incorrect information feel persuasive. The company’s emphasis on cited answers is an important differentiator, yet citations alone are not a permanent solution. Users need sources that are relevant, accurately represented, and easy to inspect. Publishers also want a clearer answer to a difficult question: if AI systems increasingly summarize their reporting, how will the economics of original journalism hold up?

Still, the reported revenue growth suggests that the company has found demand beyond curiosity. Perplexity Computer, in particular, signals a shift from a product that helps people find information to one that may help them act on it. That transition is where the real commercial upside lies. A search subscription can be useful; an AI colleague that meaningfully reduces routine research, monitoring, analysis, and administrative work could become embedded in a professional’s daily workflow.

Perplexity CEO Aravind Srinivas has said the company plans to go public in 2028 regardless of how the public listings of OpenAI and Anthropic are received. Whether that schedule holds is impossible to know, particularly in a market moving this quickly. But a high-profile NVIDIA investment, if it materializes, would give Perplexity more than fresh capital. It would add one of the AI era’s most influential companies to a group of backers that already includes Amazon founder Jeff Bezos and SoftBank.

The bigger takeaway is that the AI market is starting to place a premium on applications, not just models and chips. NVIDIA’s own success has been powered by the infrastructure race. Perplexity represents the next question: which products will become the everyday front door to that infrastructure?

A $30 billion valuation does not answer it. But it makes clear that investors think Perplexity has a credible chance to be one of them.