Nvidia is reportedly considering a new investment in Perplexity, the artificial intelligence company best known for its answer engine. The discussions come as Perplexity attempts to evolve beyond AI-powered search and build a broader platform capable of completing work on behalf of users.
According to information first reported by The Information and relayed by Reuters, Nvidia could participate in a new equity funding round valuing Perplexity at more than $30 billion. That would represent an increase of over 50% from the approximately $20 billion valuation attached to the startup’s previous financing.
The possible deal remains unconfirmed. Perplexity declined to comment on the report, while Nvidia had not publicly confirmed its participation at the time of publication.
The most striking part of the story may not be the valuation, however. Perplexity’s annualized revenue reportedly climbed from less than $250 million at the beginning of 2026 to more than $750 million by August. Part of that acceleration has been attributed to Perplexity Computer, an AI agent designed to execute complex tasks rather than simply answer questions.
Nvidia would be strengthening an existing relationship
A new investment would not mark Nvidia’s first involvement with Perplexity. The chipmaker is already one of the startup’s financial backers, alongside investors including SoftBank and Amazon founder Jeff Bezos.
The two companies are also becoming more closely connected technologically.
In July 2026, Perplexity confirmed that it planned to use Nvidia’s new Vera central processors. These CPUs have been designed in part for the continuous workloads created by AI agents, which can browse, code, analyze information and interact with software for extended periods.
Nate Kupp, Perplexity’s vice president for Computer Enterprise and Infrastructure, told Reuters that Vera completed certain AI-agent coding workloads approximately 1.5 times faster than conventional processors in the company’s tests. This remains a performance claim based on Perplexity’s own evaluation, and the company did not disclose how many processors it intended to purchase.
An additional investment would therefore reinforce a relationship in which Nvidia could be both a shareholder and an important infrastructure supplier.
What does $750 million in annualized revenue mean?
The revenue figure needs some clarification.
Annualized revenue is not necessarily the amount Perplexity has already generated during the year. It represents an extrapolation of its recent revenue rate over a twelve-month period.
For example, if a company generates $62.5 million during one month and assumes that pace will continue, its annualized revenue would be approximately $750 million. The calculation can be useful for measuring the trajectory of a fast-growing company, but it does not guarantee that the same performance will be maintained for a full year.
Perplexity’s reported progression is nevertheless significant. Its annualized revenue has increased from less than $250 million at the beginning of 2026 to more than $750 million in August, suggesting that its revenue rate has roughly tripled in a matter of months.
The figures have not been published in audited financial statements. Perplexity is privately held, meaning it is not subject to the same quarterly disclosure requirements as listed companies such as Nvidia, Microsoft or Alphabet.
Perplexity is moving beyond AI search
Perplexity originally became known as an alternative to traditional search engines. Instead of presenting users with a page of links, its service generates a direct response supported by sources retrieved from the web.
That model helped establish the company as one of the most visible challengers to Google’s dominance in online search. But search alone is no longer the full extent of Perplexity’s ambitions.
The company is increasingly positioning itself as a provider of AI agents that can perform work for consumers and professional teams.
This reflects a broader industry shift from chatbots that answer questions to systems capable of completing multi-step assignments, a distinction Demystia also explored in its guide to ChatGPT Chat versus Work.
Its central product in this strategy is Perplexity Computer, which the company describes as a general-purpose digital worker. After receiving an instruction in natural language, the system can deploy subagents to research information, browse websites, write code, create reports and connect with external tools.
Perplexity says Computer can execute workflows in the background for hours or even months. It can also perform recurring tasks, monitor information continuously and interact with services such as Gmail, Slack, Notion and online calendars.
The product is currently available to Pro and Max subscribers across desktop and mobile devices, with integrations for Slack and Microsoft 365. Although professional use is an important part of its positioning, Computer is not exclusively an enterprise product.
Why agents can generate more revenue than search
The transition from answers to actions changes the business model.
A conventional AI search request may require a handful of searches and model calls before generating a response. An agentic task can be much more demanding. It may involve dozens of steps, several specialized models, browser automation and continuous execution over a longer period.
This creates more value when the agent successfully completes meaningful work, but it also generates higher computing costs.
Perplexity has responded by combining subscriptions with usage-based pricing. Users receive a certain allocation of computing credits and may need to purchase additional capacity when they run more demanding workflows.
This model can increase the amount of revenue generated by highly active customers. It also allows Perplexity to charge according to the resources consumed rather than relying exclusively on a fixed monthly subscription.
The effect became visible earlier in 2026. The Financial Times reported that Perplexity’s estimated annual recurring revenue increased by around 50% in March, exceeding $450 million following the launch of Computer and the introduction of usage-based pricing.
At the time, the company reportedly had more than 100 million monthly users across its search and agent products, including tens of thousands of enterprise customers.
However, higher revenue does not automatically translate into profitability. Agentic systems are expensive to operate because they require repeated access to AI models and computing infrastructure. Perplexity must therefore ensure that the price paid by customers remains higher than the cost of running their tasks.
Why Perplexity could matter to Nvidia
Nvidia has repeatedly invested in companies across the AI ecosystem. Supporting model developers, cloud providers and AI applications can encourage the adoption of services that ultimately require more computing infrastructure.
That strategy is becoming more important as major AI developers invest in alternatives to Nvidia’s hardware. OpenAI, for example, has introduced Jalapeño, a custom inference chip designed to compete with Nvidia systems.
Perplexity fits that strategy particularly well.
Its agents are designed to remain active for long periods and complete multi-step workflows. If this type of product gains widespread adoption, it could generate substantially more computing demand than occasional chatbot or search queries.
An investment would therefore give Nvidia greater exposure to the application layer of artificial intelligence while potentially supporting demand for its GPUs, CPUs and networking technologies.
Perplexity could, in return, secure a closer relationship with the company that controls much of the hardware needed to operate increasingly complex AI services.
The arrangement would illustrate how tightly connected the AI market is becoming: chipmakers invest in software companies, while those software companies become major customers for the investors’ infrastructure.
A $30 billion valuation still carries substantial expectations
Perplexity’s revenue trajectory helps explain why investors might consider a valuation above $30 billion. It does not remove the risks associated with such a price.
The company competes with some of the world’s largest technology groups. Google is integrating generative AI into search and productivity tools, while OpenAI, Anthropic and Microsoft are developing agents capable of researching information, writing code and interacting with software.
Perplexity must also pay for access to models and infrastructure supplied by other companies. Rapid usage growth can therefore increase both revenue and operating costs.
Several important details about the reported funding round remain unknown:
- the amount Nvidia could invest;
- the other investors participating in the round;
- the final valuation and financial terms;
- whether the discussions will produce a completed transaction;
- Perplexity’s current profitability and operating costs.
Until the companies make an official announcement, the investment should be treated as a reported discussion rather than a finalized agreement.
From answer engine to agent platform
The possible Nvidia investment is notable, but Perplexity’s transformation is the more consequential development.
The company is trying to expand from a service that finds and summarizes information into a platform that can research, decide and act. Its reported revenue growth suggests that customers may be willing to pay more for AI systems capable of completing work than for chatbots that only generate answers.
Whether that growth can become profitable remains an open question. Running autonomous agents requires significant computing capacity, and Perplexity faces competition from companies with greater financial and technological resources.
For Nvidia, however, that computing intensity may be precisely what makes Perplexity attractive. If agents become a major way people interact with AI, the companies operating them will require more chips and infrastructure — creating another source of demand for Nvidia’s expanding hardware platform.