SAN FRANCISCO — Nvidia, the chip maker whose processors power much of the artificial-intelligence boom, is in discussions to invest in the AI search company Perplexity at a valuation above $30 billion, The Information reported Monday. The talks, which people familiar with the matter said are ongoing and not yet final, would extend Nvidia’s reach from the hardware that trains AI models to the applications people use to search the web.
The potential investment is the second in the application layer for Nvidia in recent months. Earlier this year the company backed Poolside, the model factory building open-source coding models, and it has been quietly accumulating stakes in AI startups across the stack. The Perplexity deal would be the most visible: Perplexity has become the best-known challenger to Google in search, its app and website used by millions who prefer answers composed by AI over lists of links.
For Nvidia, the logic is defensive as much as offensive. The company controls the overwhelming majority of the market for AI training chips, but that dominance invites countermeasures: cloud providers are building their own accelerators, and model makers are exploring alternatives. Owning pieces of the companies that buy the chips — and of the applications that depend on them — hedges the hardware monopoly with an ecosystem of allies. Nvidia chief executive Jensen Huang has described the company’s strategy in terms of building the “AI factory,” selling not just chips but the entire platform a customer needs to train and run models. An investment in Perplexity extends that platform story to the front door of consumer AI.
The talks come at a moment of renewed attention to Nvidia’s pricing power. The company recently raised prices on its AI server products by more than 15 percent, a move that rippled through the industry and prompted customers from cloud providers to startups to recalculate their budgets. Investors read the increase as evidence that demand still exceeds supply, and Nvidia’s shares have climbed in response. Adding a search investment on top of a price increase sends a clear message: Nvidia intends to profit from AI at every layer, from the silicon to the search box.
Perplexity’s valuation has risen quickly. The company, founded by Aravind Srinivas, was valued at around $9 billion in a round earlier this year, and reports of the current talks put the new figure above $30 billion, a jump that reflects both its user growth and the market’s hunger for AI application companies with real revenue. Perplexity has said its annualized revenue is in the hundreds of millions of dollars, and it has been signing enterprise contracts at a brisk pace.
The investment would also deepen Nvidia’s ties to the search wars. Google, Perplexity’s main rival, has responded to the AI search threat by rolling out its own AI Overviews and by striking deals with publishers; OpenAI, which launched its own search product, is another competitor. Nvidia has no consumer search business of its own, which makes Perplexity a natural ally rather than a conflict.
Analysts said the deal, if completed, would sharpen the question of how far Nvidia’s ambitions extend. The company has historically described itself as an infrastructure vendor, content to sell the picks and shovels of the AI gold rush. Investments in Poolside and Perplexity suggest a different appetite: a stake in the miners as well, and a voice in which applications succeed. For startups in the AI application layer, Nvidia’s capital comes with an implicit promise of access — to chips, to engineering help, to the ecosystem — and an implicit pressure to build on Nvidia’s stack rather than a rival’s.
People familiar with the talks cautioned that terms could change or the deal could fall apart, as such negotiations often do. But the direction is consistent. Nvidia has moved from selling chips, to renting them through its own cloud, to funding the companies that use them. A Perplexity investment would complete a circuit: the company whose hardware powers the AI boom would also hold a stake in the search engine that shows millions of people what that boom can do.
The Perplexity discussions also illuminate how the AI investment cycle has shifted. In 2023 and 2024, most big checks went to model labs — OpenAI, Anthropic, xAI — as investors bet that foundation models would capture the value of the boom. That thesis has matured, and money is now flowing to the applications and infrastructure around models. Search, with its billions of users and proven advertising economics, has become the most contested application of all, and Perplexity has emerged as the standard-bearer for the challenger camp. A Nvidia investment would give it a patron with deep pockets and a strategic interest in its success.
For Nvidia, the deal also guards against a future in which its chip dominance fades. Every major cloud provider is designing its own accelerators, and OpenAI and others have explored custom silicon. If the hardware market becomes more competitive, Nvidia’s influence will depend increasingly on its network of allies across the stack — the startups that buy its chips, run its software and champion its platform. An investment in Perplexity, like the one in Poolside, converts a customer relationship into an ownership stake, binding the company’s fortunes to Nvidia’s ecosystem. Analysts described the pattern as classic platform strategy: make the partners strong, and the platform grows with them.
Investors will look for the deal to close with terms near the $30 billion-plus figure reported, and Nvidia taking a board seat if talks follow the usual shape. Neither company has commented. Whatever the final terms, the talks themselves are evidence of how the AI economy is knitting itself together — chip makers buying into search, search companies selling stakes to their suppliers, and everyone racing to own a piece of the interaction that most people will use to reach AI every day.
