Is Nvidia Buying Perplexity? What a $30 Billion Deal Could Really Mean

Nvidia is not, at least according to credible reporting, trying to buy Perplexity. The actual development is smaller, but strategically more interesting. Nvidia is reportedly discussing an investment in a funding round that could value the AI search and agent startup at more than $30 billion. That distinction matters. A minority investment would give Nvidia exposure to Perplexity’s growth without forcing it to absorb the startup’s losses, legal disputes and competitive conflicts. A full acquisition would be a completely different and far more questionable bet.

Let’s break down what has actually been reported, why Nvidia could want deeper exposure to Perplexity, what the two companies already do together, how the synergies could work, and whether an investment or acquisition would create enough value for Nvidia shareholders.

Nvidia Perplexity Deal: What Has Actually Been Reported?

The original report came from The Information and was subsequently reported by Reuters. It said Nvidia was discussing an investment in Perplexity as part of an equity funding round that could value the startup at more than $30 billion.

As of August 25, 2026, neither Nvidia nor Perplexity has announced a completed transaction. Perplexity declined to comment on the report, while Nvidia did not immediately respond to Reuters.

Here is what is known and what remains speculation.

The reported valuation would be more than 50% above Perplexity’s previous $20 billion financing valuation. But revenue appears to have grown even faster. Perplexity’s annualised revenue reportedly increased from less than $250 million at the beginning of 2026 to more than $750 million by August.

That means this is not simply another AI startup raising money on a new story. Perplexity has reportedly tripled its revenue run rate in less than a year, although investors must understand what that number actually represents.

Nvidia’s Existing Investment in Perplexity: Timeline and History

Nvidia first invested in Perplexity in January 2024, when the startup raised $73.6 million at a valuation of around $520 million.

At the time, Nvidia described Perplexity as its first consumer investment from its corporate investment arm. That is important because Nvidia was already looking beyond enterprise software and data centres. It saw AI search as a possible consumer gateway into the AI economy.

Perplexity’s reported valuation has since climbed from approximately $520 million to more than $30 billion. That is almost 58 times higher in under three years.

The value of Nvidia’s existing stake is unknown because the company has never publicly disclosed how much it invested, how much dilution it has experienced or whether it participated in every subsequent round.

Therefore, the current discussion is not about Nvidia suddenly discovering Perplexity. It is about an existing shareholder potentially increasing its exposure after Perplexity has become a much larger business.

What Is Perplexity AI and How Does It Make Money?

Perplexity began as an AI answer engine. Instead of displaying a page of links like a traditional search engine, it searches the web, reads multiple sources and produces a direct answer with citations.

But describing Perplexity as only an AI search engine is now outdated. The company is trying to become a broader AI work platform.

The most important development is Perplexity Computer, launched in February 2026.

Computer can divide a complex task among multiple models, use external tools, search the internet, write code, analyse data and continue working after the user has stepped away. Perplexity says it can orchestrate around 20 models, assigning different parts of a task to whichever model is best suited.

Think of it as a general contractor. The user explains the final goal, while Computer hires different digital specialists, coordinates their work and returns the completed project.

The product also introduced usage-based credits on top of Perplexity’s subscription plans. That allowed Perplexity to earn more from heavy users whose agents consume substantially more computing resources.

According to the Financial Times, Perplexity had more than 100 million monthly active users and tens of thousands of enterprise customers by April 2026. Its estimated annual recurring revenue had reached more than $450 million in March. The latest report places annualised revenue above $750 million.

Perplexity AI Revenue Explained: Growth, Business Model and Risks

Perplexity is private, so it does not release audited quarterly financial statements. The reported $750 million figure is an annualised revenue run rate.

In simple terms, annualised revenue usually takes a recent month or quarter and assumes that level continues for a full year. It is useful for measuring current momentum, but it is not the same as revenue already earned during the previous 12 months.

This distinction has become more important because Perplexity Computer uses consumption credits. Usage revenue can rise quickly when customers experiment with expensive agent tasks, but it may not be as predictable as a fixed monthly subscription.

Investors should not automatically treat all $750 million as traditional recurring software revenue. The following information remains undisclosed:

• Recognised revenue during the last 12 months
• Gross margin
• Operating loss
• Cash burn
• Customer retention
• Revenue coming from subscriptions
• Revenue coming from usage credits
• Revenue concentration among enterprise customers
• Cost per completed agent task

Perplexity is growing extremely quickly, but the quality of that growth matters as much as its speed.

Why Does Perplexity Need More Funding Despite Rapid Growth?

AI search and AI agents are expensive to operate. Every time a user asks a complex question, Perplexity may have to perform multiple searches, retrieve documents, run more than one model, check the results and generate a final response.

An ordinary software product might retrieve information from a database. An AI agent can perform dozens of computing steps for a single user request.

Perplexity signed a three-year, $750 million agreement to use Microsoft Azure, equal to an average commitment of approximately $250 million a year. The company said Amazon Web Services would remain a preferred infrastructure provider, meaning the Azure agreement did not replace its existing AWS spending.

Perplexity has also signed a multiyear deal to use Nvidia Grace Blackwell systems through CoreWeave.

The $750 million Azure commitment alone averages around one-third of Perplexity’s latest annualised revenue. It does not mean Perplexity spends one-third of revenue on Azure today because the commitment runs for three years and actual usage can vary. Nevertheless, it demonstrates how capital-intensive AI agents can become.

The company also needs money for product development, enterprise sales, browser distribution, model access, content licensing and legal expenses.

How Deep Is Nvidia’s Existing Relationship With Perplexity?

The two companies already work together across almost every important part of the AI computing stack.

Nvidia says TensorRT-LLM helped Perplexity reduce certain inference costs by four times and save approximately $600,000 a year at an earlier stage of its development. Moving from A100 to H100 GPUs reportedly doubled speed and increased throughput by 200% under the tested configuration.

More recently, Perplexity said Nvidia’s Vera CPU completed AI-agent coding workloads around 1.5 times faster than traditional CPUs. Perplexity called Vera a strong fit for its agent workloads, although the company did not disclose how many units it intends to use.

Nemotron 3 Super, Nvidia’s open AI model, is also one of the models available to Perplexity users and one of the models orchestrated by Computer.

This leads to the central question: if Nvidia already receives nearly all these benefits through commercial agreements and a minority investment, what additional value would it gain by buying Perplexity?

Why Nvidia Wants Exposure to Perplexity and AI Search

1. Nvidia Would Gain Exposure to the Application Layer

Nvidia currently earns most of its money from AI infrastructure. It sells the chips, networking systems and software used to train and run AI applications.

Perplexity operates much closer to the final customer. It sells AI search, research and agent services to consumers and businesses.

If AI hardware eventually becomes more competitive, some economic value could move from chip suppliers toward applications that own customer relationships. A stake in Perplexity gives Nvidia exposure to that possibility.

It is similar to a company selling kitchen equipment while also owning a small stake in a fast-growing restaurant chain. Nvidia continues selling equipment to the entire industry, but it also participates if one customer becomes exceptionally valuable.

2. Perplexity Is a Real-World Testing Ground for Inference

Training a model is expensive, but it happens periodically. Inference is what happens every time someone uses that model.

As AI adoption grows, inference could become the larger, more durable computing market. Perplexity’s search and agent workloads are particularly valuable because they involve:

• Low-latency responses
• Long context windows
• Retrieval from external data
• Multiple models working together
• Tool usage
• Continuous agent execution
• High volumes of small computing tasks

Perplexity gives Nvidia a live environment in which it can optimise Vera CPUs, GPUs, TensorRT, Dynamo, NeMo and Nemotron.

This is more valuable than a simple hardware order. Perplexity can help Nvidia learn which parts of an agent workflow create bottlenecks and where new hardware or software improvements are needed.

3. Perplexity Can Help Nvidia Prove the Vera CPU Business

Nvidia is expanding beyond GPUs into central processing units through Vera.

The CPU market is dominated by Intel and AMD. Nvidia needs well-known customers to demonstrate that Vera is not merely an accessory to its GPU systems.

Perplexity saying that Vera completed agent coding tasks 1.5 times faster gives Nvidia an important reference customer. A deeper investment could align Perplexity more closely with Vera’s product roadmap.

However, Nvidia does not need to acquire Perplexity to obtain this benefit. A supply agreement and technical partnership can achieve most of it.

4. Perplexity Can Distribute Nvidia’s AI Models

Nvidia is building open models under the Nemotron brand. Perplexity can place those models in front of consumers, enterprises and developers.

This can generate:

• More usage data
• Better model feedback
• Broader developer adoption
• More demand for Nvidia-optimised inference
• Evidence that Nvidia offers more than hardware

Again, this synergy already exists. An investment may strengthen it, but ownership is not essential.

5. Nvidia Could Capture Financial Upside From a Future Perplexity IPO

Perplexity CEO Aravind Srinivas has said the company continues to target a public listing in 2028.

If Perplexity becomes a major independent AI platform, Nvidia could earn a return on its equity stake in addition to earning money from the infrastructure used by Perplexity.

Nvidia is effectively trying to capture value at two points:

1. When Perplexity buys or rents Nvidia-powered computing.
2. When Perplexity’s equity value increases.

That can be attractive, but only if Nvidia does not overpay or fund demand that would not otherwise exist.

The Demand Recycling Test: Is Nvidia’s AI Investment Strategy Creating Real Demand?

Nvidia’s expanding investments have created concern about circular AI financing.

A simplified loop looks like this:

1. Nvidia invests in an AI company.
2. The AI company spends money with a cloud provider.
3. The cloud provider buys Nvidia systems.
4. Nvidia records hardware revenue.

Not every arrangement like this is artificial or problematic. Perplexity has independent customers, subscription revenue and enterprise usage. Its products generate external demand.

Still, investors should apply what we call the Demand Recycling Test.

Perplexity passes the first test better than many pre-revenue AI startups because it reportedly has more than $750 million in annualised revenue. But the remaining questions cannot be answered until the funding and commercial terms are disclosed.

Perplexity AI Valuation: Is a $30 Billion Valuation Justified?

At $750 million in annualised revenue, a $30 billion valuation implies a multiple of approximately 40 times revenue.

Forty times revenue is expensive, especially for a company that is reportedly unprofitable. But there is a useful detail hidden beneath the headline.

At the beginning of 2026, a $20 billion valuation against less than $250 million in annualised revenue would have represented a multiple of more than 80 times. Revenue has therefore grown faster than valuation, bringing the implied multiple down significantly.

That does not make Perplexity cheap. It means its valuation has partially grown into its underlying business.

The greater concern is that annualised usage revenue may deserve a lower multiple than predictable subscription revenue. Until Perplexity discloses its revenue mix, gross margin and retention, investors cannot know whether 40 times represents premium software revenue or expensive and volatile computing consumption.

What Would a Minority Nvidia Investment Buy?

The report does not disclose how much Nvidia may invest. The following table simply shows the approximate ownership a new investment could purchase if the round uses a $30 billion post-money valuation.

These percentages would change depending on whether the reported valuation is pre-money or post-money, the rights attached to Nvidia’s existing shares and the final round size.

For Nvidia, even a $1 billion investment would be small relative to its financial scale. As of April 2026, the company held $13.2 billion in cash, $37.1 billion in marketable debt securities, $30.2 billion in marketable equity securities and $43.4 billion in non-marketable investments.

Nvidia also generated $48.6 billion in free cash flow during its fiscal first quarter alone.

What Return Would Nvidia Need at a $30 Billion Entry Valuation?

Suppose Nvidia invests at a $30 billion valuation and wants the investment to generate an illustrative 15% annual return over five years.

Perplexity would need to be worth approximately $60.3 billion after five years, ignoring future dilution.

What revenue would support that valuation?

This is not a Perplexity forecast. It is a valuation hurdle.

Perplexity has recently grown much faster than these rates, but maintaining more than 50% annual growth becomes harder as the revenue base expands. Competition from Google, OpenAI, Anthropic, Microsoft and other agent platforms will also intensify.

A successful investment therefore requires Perplexity to become a multi-billion-dollar revenue company, not merely a popular search tool.

Could Nvidia Acquire Perplexity? Acquisition Scenario Explained

No credible report currently says Nvidia is negotiating a takeover. Nevertheless, the acquisition scenario is worth analysing because it has become part of the online discussion.

A buyer normally needs to offer more than the latest private-market valuation to persuade shareholders to sell. The following scenarios use illustrative premiums and are not reported offers.

Nvidia could finance an acquisition using shares, cash, debt or a combination. At Nvidia’s approximately $5.05 trillion market value on August 25, a $37.5 billion all-stock acquisition would cause less than 1% dilution before considering employee compensation and other adjustments.

Therefore, affordability is not the problem. Value creation is.

Perplexity’s $750 million annualised revenue is only around 0.3% of Nvidia’s trailing 12-month revenue. Perplexity is also unprofitable, while Nvidia’s trailing operating margin is above 60%.

An acquisition would add very little near-term revenue or profit while introducing a very different and less proven business model.

The Acquisition Hurdle: How Large Must Perplexity Become?

Assume Nvidia paid an illustrative $37.5 billion and wanted the acquired business to eventually produce a 10% pre-tax operating return on that purchase price.

Perplexity would need to generate $3.75 billion in annual operating profit.

The model ignores taxes, integration expenses, share-based compensation, the time required to reach maturity and the opportunity cost of capital. In reality, Nvidia would probably need even more value.

This is why a Perplexity acquisition cannot be justified by its current revenue. Nvidia would have to believe Perplexity could become one of the world’s leading consumer and enterprise AI platforms.

Why Owning Perplexity Could Actually Reduce Its Value

The most overlooked risk is neutrality.

Perplexity’s appeal comes partly from its ability to use different models and infrastructure providers. Computer can route tasks across models from Nvidia, OpenAI, Anthropic and other developers. Perplexity also works with AWS, Microsoft Azure and CoreWeave.

If Nvidia owned Perplexity, several questions would immediately arise:

• Would Perplexity still choose the best model for each task?
• Would Nvidia favour Nemotron over competing models?
• Would Microsoft, Amazon or Google continue supporting the same integrations?
• Would Perplexity be expected to prioritise Nvidia hardware?
• Would enterprises trust its claims of model neutrality?
• Would rival model developers share technical access with a platform owned by Nvidia?

Nvidia’s position as a relatively neutral infrastructure supplier is strategically valuable. It sells to companies that compete fiercely with one another.

Owning a major downstream AI application could introduce what we call a neutrality tax. Nvidia may gain control over one customer but weaken the confidence of several larger customers.

Google Search alone generated $63.3 billion in revenue during the June 2026 quarter. That is around 84 times Perplexity’s entire current annualised revenue. Google’s Gemini application had 950 million monthly users in the same quarter, compared with Perplexity’s reported base of more than 100 million users earlier in the year.

Nvidia should be careful about creating friction with customers and partners that are substantially larger than the company it might acquire.

Investment vs Acquisition: Which Option Creates More Value for Nvidia?

This is the central conclusion of our analysis.

Nvidia already sells the computing, provides the software, supplies the models and owns an equity interest.

Buying Perplexity would mainly add control. But control is only valuable if Nvidia cannot obtain the required benefits through contracts.

This creates a simple Control Premium Test:

If Nvidia can obtain nearly all the strategic benefits through a supply agreement, model partnership and minority investment, paying tens of billions for control is unnecessary.

Perplexity appears to pass that test in favour of remaining independent.

Perplexity AI Risks: Legal Challenges and Business Concerns

Perplexity’s products rely on retrieving and summarising information from the internet. That has led to disputes with publishers and online platforms.

CNN filed a lawsuit in May 2026 alleging that Perplexity unlawfully copied and distributed thousands of its stories, videos and images. Perplexity responded that facts cannot be copyrighted.

The company has also faced lawsuits from The New York Times, Reddit, Dow Jones and others.

Separately, Amazon sued Perplexity over the way Comet’s agent accessed Amazon accounts and placed orders. A district court initially issued a preliminary injunction, but the Ninth Circuit vacated that injunction in August 2026. The appeals court concluded that Amazon was unlikely to succeed under the computer-access laws it had relied upon because the user, rather than Perplexity itself, was accessing Amazon through the agent.

The case has returned to the lower court, and Amazon can continue pursuing other contractual or legal claims. A minority investor has limited exposure to these liabilities. An acquirer inherits them.

Nvidia would need to evaluate:

• Potential content licensing costs
• Whether injunctions could restrict core product features
• Data privacy practices
• Enterprise security controls
• Agent permissions and liability
• Reputational risk with publishers
• Historical data collection practices

These issues do not make Perplexity uninvestable, but they make a full acquisition considerably less attractive.

What Does Nvidia’s Investment Mean for Perplexity AI?

Under a Minority Investment

Perplexity would receive additional capital without giving up control. It could use that money to:

• Expand its computing capacity
• Fund Perplexity Computer
• Improve Comet
• Build enterprise sales
• Enter more distribution partnerships
• Pay for licensed content
• Prepare for a future IPO

Nvidia’s participation would also act as technical validation for Perplexity’s agent infrastructure.

However, Perplexity would need to prevent the relationship from weakening its multi-model and multi-cloud positioning. Its ability to choose among Nvidia, OpenAI, Anthropic and other models is part of its product value.

Under an Acquisition

Perplexity would gain access to one of the strongest balance sheets in technology. It could receive large amounts of computing capacity without raising money every few months.

But it would lose much of its independence. A sale could also conflict with management’s stated 2028 IPO plan.

The larger risk is that Perplexity could stop being perceived as an independent AI answer engine and start being seen as Nvidia’s distribution channel. That may reduce its appeal to users, model developers, cloud providers and enterprises seeking vendor neutrality.

Our Take: Should Nvidia Invest in or Acquire Perplexity?

Our view is clear.

A measured minority investment in Perplexity makes strategic sense for Nvidia. A full acquisition does not, at least at the current valuation and with the information available today.

The minority investment case works because Nvidia can:

• Participate in Perplexity’s financial upside
• Strengthen a major inference customer
• Promote Vera CPUs
• Expand Nemotron distribution
• Learn from real agent workloads
• Preserve its position as a neutral infrastructure provider
• Limit its exposure to Perplexity’s operating and legal risks

The valuation is still demanding. At approximately 40 times annualised revenue, Nvidia would be paying for years of exceptional growth. The investment would make the most sense if it is modest, does not guarantee Perplexity’s obligations and does not require Perplexity to use Nvidia exclusively.

The acquisition case is much weaker.

Nvidia would be paying at least $30 billion, and probably more, for a company producing around 0.3% of Nvidia’s trailing revenue and no reported profit. It would inherit legal disputes, expensive infrastructure commitments and a highly competitive consumer business.

Most importantly, Nvidia already gets almost every meaningful synergy without owning Perplexity.

The ideal outcome for Nvidia is therefore simple:

Own enough of Perplexity to participate in its success, but not enough to become responsible for running it.

For Perplexity, the best outcome is also likely to be funding plus a deeper technical partnership while preserving its independence, multi-model approach and planned IPO route.