Nvidia Discusses Perplexity Investment at $30 Billion-Plus Valuation
The Takeaway
- Nvidia previously considered paying Perplexity billions of dollars to license some of the startup’s technology and hire some staff. Over the past year, Nvidia struck similar licensing and hiring deals with AI inference chip designer Groq and networking hardware designer Enfabrica.
Nvidia is discussing investing in Perplexity as part of an equity round that would value the AI startup at more than $30 billion, according to people with knowledge of the discussion.
The round would be worth billions of dollars and boost the startup’s valuation more than 50% from its last financing a year ago, the people said. The proposed deal comes as the two companies have forged stronger business ties and Perplexity’s annualized revenue has climbed to more than $750 million from less than $250 million at the start of the year, thanks in part to Perplexity Computer, an AI agent that professionals use to automate computer tasks, according to one of the people and another person familiar with the matter.
Nvidia previously considered paying Perplexity billions of dollars to license some of the startup’s technology and hire some of its staff as part of the chip leader’s push to develop AI models and related software, according to two of the people familiar with the deal talks. It isn’t clear what prompted those discussions to morph to Nvidia making a more traditional investment deal.
Nvidia has been acting as a kind of central bank to the AI startup field for several years, pumping billions of dollars into dozens of companies that rent or buy its vaunted AI chips. And Nvidia has already made multiple investments in Perplexity. Nvidia CEO Jensen Huang has said in interviews that he uses Perplexity as a go-to chatbot.
While Nvidia has spent tens of billions of dollars on investments and partnerships involving chip, networking, and data center-related firms to entrench its AI hardware dominance, it recently started pursuing licensing or hiring deals with startups to strengthen its Nemotron AI models and other offerings.
Last week, Nvidia raised eyebrows by agreeing to pay $6 billion to license AI model development software from Poolside, a startup that has pivoted its business multiple times since it was founded three years ago. Nvidia planned to make offers to hire more than 100 Poolside employees, including those that developed the startup’s recently launched Laguna open-weight AI model. Separately, Nvidia is also investing $1 billion in Poolside at a $12 billion pre-money valuation.
Nvidia told Nemotron employees that the Poolside deal aimed to help Nvidia develop Nemotron models more quickly and effectively, though it’s not yet clear how the new employees will be incorporated into the organization, according to a person with knowledge of the communications.
Perplexity was an early darling of venture capitalists in the current AI boom. The company launched an AI search engine long before ChatGPT enabled web search. Perplexity has since expanded its products, which now include Perplexity Computer, which resembles newer offerings from OpenAI and Anthropic and open source tools like OpenClaw.
Nvidia separately developed an OpenClaw-like software for AI app developers to use for free.
In March, Perplexity joined Nvidia’s Nemotron Coalition, a group helping Nvidia develop its open models. Representatives from the two companies have been meeting multiple times a week to collaborate on both hardware and software development, according to someone with knowledge of the meetings. The coalition is part of Nvidia’s aggressive push, particularly over the last year, to develop Nemotron open-source models more quickly and effectively.
While the Nemotron push could be competitive with many of Nvidia’s customers and users, Nvidia has felt compelled to do so as a counterweight to open AI models being developed by Chinese developers. Nvidia believes that giving app developers access to open, Western-made AI models will drive greater demand for Nvidia hardware.
Nvidia’s open model pursuits also mean that Nvidia is now one of the biggest users of its own chips; it has agreed to spend tens of billions of dollars over the next few years to rent Nvidia servers through cloud providers such as Lambda, Amazon and other buyers of Nvidia chips.
A deal could give Perplexity more breathing room to develop its business ahead of a potential initial public offering in the coming years. CEO Aravind Srinivas told CNBC that he was considering a 2028 time frame for such an event.
In addition to the search engine and Perplexity Computer, the company also sells access to open source and closed source AI models developed by other companies and is now among the biggest “AI native” startups by revenue. The company’s gross profit margin has stayed around its prior mark of 60%, said one of the people familiar with its business, though it historically hasn’t included expenses from nonpaying or trial users in that calculation.
The four-year-old Perplexity doesn’t train its own models from scratch but says it routes AI requests to the best model, depending on the task. Perplexity has said it plans to use Nvidia’s Vera central processing units to run agentic AI queries, and it previously struck a deal to use Nvidia chips through CoreWeave, a publicly traded cloud-server provider that Nvidia heavily backs.
Perplexity has raised more than $1.7 billion from Nvidia and other investors including New Enterprise Associates, Accel and Softbank.