Exa Labs raises $250M at $2.2B valuation led by a16z for AI search

Exa Labs closed a $250M round led by Andreessen Horowitz on May 20, valuing the AI search company at $2.2B — a substantial step up from its $85M Series B less than a year ago, which had included NVIDIA and Y Combinator. SiliconANGLE's report frames the use of funds as primarily infrastructure: Exa is buying more compute to scale its embedding pipelines and serve growing agent-side query volume.

Exa's product is 'search built for agents' — an API that returns semantically ranked, content-rich results designed to be consumed by LLM tool calls rather than rendered for humans. The technical bet is that agentic workflows have different retrieval requirements than human web search: cleaner extracted content, ranking by usefulness-to-an-LLM rather than click-through, and pricing that works at high call volumes from automated systems.

The competitive landscape is heating fast. Perplexity (in the same source set today shipping 'query-aware compression') is going broader as a consumer answer engine; Tavily and Brave are courting the same agent-API niche; OpenAI shipped its own browsing capability; and Google's I/O keynote leaned hard on Search-as-agent-infrastructure with the AI Mode redesign and 2.5B AI Overviews users. Exa's pitch is that a neutral, agent-first search layer beats vertical lock-in to any single model provider.

For context on the round: a16z leading and pricing $2.2B less than a year after the $85M round implies serious revenue traction; otherwise pure-software AI infra rounds at this stage normally see 2x markups, not closer to 5x. Watch for Exa to disclose ARR or developer numbers in the coming weeks to justify the multiple — and watch whether Google or OpenAI respond by tightening or differentiating their own search-for-agents pricing.