Exa was worth $700 million in September.
Now it's worth $2.2 billion.
The San Francisco startup builds a search engine for AI agents, not human users.
It raised $250 million in a round led by Andreessen Horowitz.
That triples its value in about eight months.
Google has owned online search for two decades, with every typed question flowing through ten blue links and a wall of ads.
AI is rewriting that model.
When an AI agent runs a task - like booking a flight or pulling a research report - it doesn't need a list of blue links.
It needs raw data it can act on.
That's the gap Exa is going after, one CEO Will Bryk and co-founders Dan McArdle and Jeffrey Wang set out to fill when they started the firm in 2021 to build a search layer made for machines, not people.
The firm is small but growing fast, with about $12 million in yearly revenue at last count.
Market Briefs covers the moves that actually matter in AI investing - and joining comes with a free investing masterclass.
Exa isn't alone.
Tavily, TinyFish, and Parallel Web Systems are all going after the same prize.
That prize is the search layer for AI.
Big tech is moving in too.
OpenAI, Anthropic, and Google are building their own search and pull tools, while Perplexity has raised billions to take on Google with a direct AI search product.
What sets Exa apart from that crowd is the audience.
The firm isn't trying to win readers - it's trying to win the AI tools those readers use.
The value jump isn't because Exa built a Google killer overnight.
It's because investors see AI search as one of the few spots in software where a new leader can still emerge.
If AI agents are going to do real work on the web, they need a search engine built for them.
Whoever wins that layer gets paid by every other AI firm on top, which is why Andreessen Horowitz tripled its bet on Exa in eight months.
That tells you what at least one major venture firm thinks of the timing, on top of the prior round in 2025 backed by Nvidia and Lightspeed Venture Partners on the back of Exa's pitch that AI agents need a search tool built for machines, not one built for human users that the agent has to work around.
For investors watching the AI stack, the bet that follows is simple.
Search may stop being a Google story and start being a layer that any AI agent calls when it needs to look something up.
If that shift plays out, Exa is in the running to own a slice of it, though the next test is the same one every fast-growing startup faces.
It has to turn investor cash into real, paying customers, while knowing big AI firms could try to build the same search tool in house instead of paying Exa to do it for them.
Either way, the value of the firm now hinges on how fast AI agents pull real work off the web.
If you want this kind of read on AI startups and where the money is moving, join 350,000+ investors reading Market Briefs - and pick up a 45-minute investing course as a bonus.
