Perplexity signs a deal with Microsoft; sources say the $750M, three-year commitment will let Perplexity deploy AI models through Microsoft's Foundry service
AI startup Perplexity signed a $750 million deal with Microsoft Corp. to use its Azure cloud service, spreading its business beyond longtime cloud partner Amazon.com Inc.
Context & Ripple Effects
Perplexity’s cloud commitment follows a rapid expansion from its 2023 Series A into an answer-engine business that reported sharply higher query volume and was later said to be targeting a valuation increase alongside roughly $50M in annualized revenue. Its reported revenue growth and fundraising plans made dependable model infrastructure a more consequential operating decision.
The deal also arrives after Perplexity’s shopping and advertising efforts struggled to scale, leaving its ability to turn product usage into durable revenue central to the economics of a large multiyear infrastructure obligation.
First-order effects
- Perplexity gains a second major cloud route for deploying models through Microsoft Foundry, reducing its dependence on longtime partner Amazon while committing $750M over three years.
- Microsoft secures a sizable, long-duration Azure/Foundry customer commitment; Perplexity takes on a clearer fixed infrastructure cost base.
Second-order effects
- Amazon now faces a more explicitly multi-cloud Perplexity account, while Microsoft can use the deployment as evidence that its Foundry service can win AI-application workloads from rival cloud ecosystems.
- The commitment raises pressure on Perplexity to align model deployment, product demand, and monetization, particularly while its adjacent commerce and ad businesses remain unsettled.
Third-order effects
- If comparable AI application companies split workloads among cloud providers, cloud differentiation will increasingly depend on managed model platforms and commercial terms rather than exclusive hosting relationships.
- Large multiyear compute contracts could make infrastructure commitments a more visible constraint on AI startups’ operating models, favoring providers that can convert usage into predictable platform revenue.
The trend: AI application companies are shifting from single-cloud dependence toward longer-term, multi-provider commitments centered on managed model platforms.