The deal in brief
Nvidia is reportedly in discussions to invest in Perplexity AI at a valuation of more than $30 billion, as part of a new funding round worth several billion dollars. The report, first published by The Information on August 23, 2026 and confirmed by multiple financial outlets, had not been formally acknowledged by either company at time of writing. Discussions are described as ongoing. If completed at the reported valuation, the round would represent a 50-plus percent increase from Perplexity’s previous fundraising valuation roughly one year ago.
Perplexity’s revenue trajectory makes the number plausible: annualized revenue climbed above $750 million by mid-2026, up from under $250 million at the start of the year — a threefold increase in under eight months. The company has expanded from consumer AI search into enterprise AI agent development, launching Computer for Enterprise at its Ask 2026 developer conference, which competes directly with Microsoft Copilot and Salesforce Einstein for the enterprise AI productivity layer.
What Nvidia is reportedly buying into
Perplexity launched in 2022 as an AI-powered search interface built to surface cited answers rather than links. By 2026 it had grown to more than 100 million monthly active users, powered by a multi-model backend that queries foundation models from OpenAI, Anthropic, Google and proprietary Perplexity models simultaneously before synthesizing a response with source citations.
The revenue growth that underpins the $30 billion reported valuation comes from two directions: a consumer subscription at $20 per month (Pro plan) and an enterprise product suite that gained serious traction through 2026. The Sonar API — which lets developers embed Perplexity’s cited-search capability into their own applications — is used by teams across FinTech, HealthTech and enterprise SaaS to give their products access to live, cited external knowledge without building or maintaining their own retrieval pipelines.
Nvidia’s reported interest follows a clear pattern. In December 2025 it invested in Groq, the inference-chip startup. In August 2026 it reportedly explored paying billions to license Perplexity’s technology before pivoting to an equity discussion. The move mirrors what Nvidia did with Poolside — another AI-code generation company — earlier in August 2026. In each case Nvidia is acquiring influence over the companies that consume the most GPU inference compute at the application layer.
Perplexity Computer: the enterprise pivot
The product that most directly explains Perplexity’s enterprise revenue surge is Computer for Enterprise, announced at the Ask 2026 developer conference. Computer is a multi-model orchestration agent that routes tasks across 20 foundation models from OpenAI, Anthropic and Google depending on the nature of the request — reasoning tasks go to models that score highest on reasoning benchmarks; code tasks go to specialist coding models; retrieval tasks use Perplexity’s own search stack.
For enterprise customers, Computer connects into Slack, Snowflake, Salesforce and other business systems. A team member asking Computer to “summarise everything that changed in the Snowflake pipeline this week and flag anomalies” receives a cited, synthesised answer drawn from live operational data, not a link to a dashboard. The pricing at $200 per month per seat positions it between Microsoft Copilot ($30/month) and bespoke AI agent implementations that typically cost ten times more to build and maintain. More than 100 enterprise customers reportedly contacted Perplexity over a single weekend requesting early access after the product was announced.
The enterprise pivot is also what explains Nvidia’s interest. Computer and the Sonar API together drive substantial GPU inference workloads — exactly the kind of sustained, high-volume demand that Nvidia’s H100 and Blackwell clusters were built for. Backing Perplexity financially gives Nvidia preferred positioning as those workloads grow.
Why Nvidia is backing AI application companies
Nvidia has historically been a chip company, not a software investor. Its moves into AI application equity in 2025–2026 reflect a structural shift: the company that controls compute infrastructure wants visibility into where inference demand is growing fastest, and influence over the companies most likely to drive that demand at scale.
The pattern — Groq (inference chips), Poolside (AI coding), Perplexity (AI search and agents) — covers three of the highest-volume categories of AI compute consumption. Enterprise AI search and agent orchestration, in particular, generate continuous, latency-sensitive inference workloads that differ from periodic model training runs. They require always-on GPU capacity, not batch compute, which makes them a more predictable and recurring source of chip demand.
For enterprise teams evaluating AI vendors, this consolidation raises a practical question that goes beyond Perplexity specifically: as GPU infrastructure and AI application companies merge into fewer, larger stacks, the integration of generative AI into existing enterprise workflows increasingly means navigating vertical supply chains that start at the chip level and end at the user interface.
What it means for US & EU software teams
AI search is becoming retrieval infrastructure, not a tool. Enterprise teams that have adopted Perplexity as a productivity tool — individual seat licences for researchers and analysts — are now looking at it as a platform component. The Sonar API enables embedding live, cited search into production applications. If Nvidia’s investment closes and Perplexity continues its growth trajectory, its retrieval layer becomes a credible alternative to building custom RAG pipelines for certain use cases. The trade-off is standard: managed convenience versus control.
Multi-model orchestration is the new normal for enterprise AI products. Computer’s architecture — routing tasks across 20 models based on performance characteristics — is a public demonstration of the multi-model pattern that most serious enterprise AI products are moving toward. Teams building internal AI tools, customer-facing AI features or autonomous agents in 2026 should expect to manage model routing, fallback logic and provider dependencies as first-class engineering concerns, not afterthoughts.
Vendor concentration risk is growing. The convergence of Nvidia (compute), Perplexity (AI search and agents), and the foundation model providers (OpenAI, Anthropic, Google) into increasingly interlinked financial relationships narrows the number of genuinely independent options for enterprise AI procurement. Teams with existing Microsoft or Salesforce contracts should model whether Computer for Enterprise creates a meaningful competitive alternative or whether platform lock-in economics will ultimately dictate the choice. EU teams should also note that Perplexity’s data residency and GDPR compliance posture has not been audited at the same depth as established enterprise vendors.
The revenue benchmark matters for talent pricing. Perplexity at $750 million ARR with fewer than 400 employees is generating roughly $1.9 million in revenue per employee — one of the highest ratios in enterprise software. That ratio is achieved through AI-augmented workflows rather than headcount. It benchmarks what is becoming possible for AI-native software companies and is already influencing how US and EU engineering leaders think about team sizing and productivity expectations for 2027 and beyond.
Frequently asked questions
What is Nvidia reportedly investing in Perplexity AI?
According to The Information, reported by Yahoo Finance, Benzinga and PYMNTS on August 23–24, 2026, Nvidia is in discussions to invest in Perplexity AI as part of a new funding round that would value the startup at more than $30 billion. Neither Nvidia nor Perplexity had confirmed the deal at the time of publication; discussions are described as ongoing.
What is Perplexity’s annual revenue as of mid-2026?
Multiple sources report that Perplexity’s annualized revenue climbed above $750 million by August 2026, up from less than $250 million at the start of the year. The threefold increase in under eight months reflects accelerating enterprise adoption of its AI search products and the Sonar API.
What is Perplexity Computer for Enterprise?
Perplexity Computer for Enterprise is a multi-model AI agent that orchestrates 20 foundation models from OpenAI, Anthropic, Google and Perplexity’s own stack to automate complex tasks. It integrates with Slack, Snowflake, Salesforce and other enterprise tools and is priced at $200 per user per month. It was announced at the Ask 2026 developer conference and targets the same enterprise AI productivity market as Microsoft Copilot and Salesforce Einstein.
Why does Nvidia invest in AI search and agent companies?
Nvidia generates revenue from GPU chips used to run AI models. Enterprise AI search and agent orchestration create continuous, high-volume inference workloads — exactly the kind of sustained GPU demand that differs from periodic model training. By investing in companies like Perplexity (AI search), Poolside (AI coding) and Groq (inference chips), Nvidia secures influence across the AI compute chain from infrastructure to application layer.
Sources
Yahoo Finance — Nvidia Reportedly Weighs Perplexity Investment at More Than $30 Billion Valuation (August 24, 2026)
Benzinga — Nvidia in Talks to Back Perplexity at Over $30 Billion Valuation as Revenue Tops $750 Million (August 2026)
PYMNTS — Nvidia Considers Backing Perplexity at $30 Billion Valuation (August 2026)
VentureBeat — Perplexity Takes Its Computer AI Agent Into the Enterprise (2026)