· investment-strategies · 1 min read
Exa's $250M Series C: AI Search Isn't a Feature — It's Infrastructure at $2.2B
Exa Labs raised $250M led by a16z at $2.2B valuation to scale AI-native web search — hiring Meta and Yandex search veterans.
Google won search for humans. Exa is betting agents need their own index.
The round
$250M Series C at $2.2B valuation, led by a16z (Sarah Wang joins board). Benchmark, Lightspeed, YC, and AI leaders including Scott Wu participated.
What changes with this capital
Exa’s blog says the plan is to train next-gen models and scale infra to hundreds of thousands of searches per second — they’ve hired retrieval leads from Meta and search backend talent from Yandex.
Mental model (Julia-style)
| Layer | Job |
|---|---|
| Model | Reasoning |
| Retrieval (Exa) | Fresh facts from the web |
| App | User-facing product |
Without retrieval, agents hallucinate or go stale. That’s why this isn’t “another search startup” — it’s pick-and-shovel for agent stacks.
Practical takeaway
Founders building agents: Budget for retrieval quality like you budget for model tokens.
Investors: $2.2B says the market believes AI-native search is a standalone layer, not a feature absorbed overnight.