Exa secures $250M Series C

- Exa said on May 20 it raised a $250 million Series C led by Andreessen Horowitz, valuing the AI search company at $2.2 billion. - The company said it already serves more than 5,000 companies and 400,000 developers, with customers including Cursor, Cognition, HubSpot, OpenRouter and Monday.com. - Exa said the new capital will fund next-generation models, higher-capacity infrastructure and a global go-to-market expansion led by new CRO Marcus Holm.

Exa said on May 20 that it raised a $250 million Series C led by Andreessen Horowitz at a $2.2 billion valuation, giving the San Francisco AI search company one of the larger funding rounds this year for infrastructure tied to autonomous agents. The company disclosed the round in a post by co-founder and chief executive Will Bryk, while a16z published a separate investment note the same day. Exa said the money will go toward training new search and retrieval models, expanding infrastructure capacity and building out its commercial team. The round positions Exa as one of a growing group of startups trying to supply the data layer for AI systems that need fresh information from the web. ### Who led the round, and what valuation did Exa disclose? Andreessen Horowitz led the Series C, according to both Exa and a16z posts published May 20. Exa said the financing valued the company at $2.2 billion. a16z partners Sarah Wang, Jennifer Li, Stephenie Zhang and Jason Cui said in their post that the firm backed Exa because it believes search must be rebuilt for AI agents rather than human users. The $250 million figure appeared in Exa’s announcement, which described the raise as capital to “build the search engine for AIs.” The company did not, in the materials reviewed, disclose the full list of participating investors in its own post. (exa.ai) ### What does Exa actually sell? Exa said it operates a web search API built for AI products and agents. Bryk wrote that the company shifted in early 2023 to focus on a “perfect search API” for AI, and said the number of companies using Exa has since grown to more than 5,000. (exa.ai) More than 400,000 developers use the platform, Exa said, and named Cursor, Cognition, HubSpot, OpenRouter and Monday.com as customers. The company said its service is designed for cases including code search, enterprise research and agent workflows that need current web information rather than model training data alone. (exa.ai) ### Why are investors framing this as an AI infrastructure bet? a16z said in its investment note that “every meaningful AI workflow” increasingly depends on external information, and argued that large language models are “frozen in time” without search. (exa.ai) The firm said Exa’s advantage is in handling harder, long-tail queries and balancing cost, latency and comprehensiveness for agent use cases. Bryk made a similar case in Exa’s post, writing that AI agents will search the web more than humans this year and that search demand from large language models could grow far beyond current consumer-search volumes in the next few years. (exa.ai) That forecast is Exa’s characterization, not an independently verified market total. ### What will Exa do with the money? Exa said the funding will be used to train its next generation of models and expand infrastructure to handle “100s of thousands of searches per second.” The company also said it is hiring globally across sales, solutions and marketing. (a16z.com) Marcus Holm, previously president of LaunchDarkly, is joining as chief revenue officer to lead Exa’s global go-to-market organization, the company said. (exa.ai) Exa also said recent hires include a former head of retrieval infrastructure from Meta, a former head of search backend at Yandex and a research team from Google. ### How big is the technical build-out Exa is describing? Exa said its crawlers track more than 500 billion URLs and that it has built its own web-scale search engine, embedding models and vector databases to serve high query volumes from AI agents. (exa.ai) The company also said its search API can return results in under 200 milliseconds and that its text extraction models reduce token counts by more than 20 times. Those performance claims came from Exa’s own announcement. In the near term, Exa said the next step is scaling infrastructure and model training “in the coming months,” while Holm builds out the company’s global sales organization. The company is also continuing to recruit worldwide as it expands capacity for agent-focused search products. (exa.ai)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.