Databento Raises $100 Million at a $1.1 Billion Valuation on 39 Employees and Contracts With Five of the Six Biggest AI Labs

Databento, a market data company that sells exchange feeds through a single API, has raised $100 million in a Series C at a $1.1 billion valuation. Fifth Down Capital led the round, with existing investors NEA, DRW Venture Capital, Tribe Capital and Hedosophia returning, and Fitch Ventures joining as a strategic investor. The company announced the financing on November 12, roughly four months after a $97 million Series B.

The numbers attached to the round are unusual for an infrastructure company of this size. Databento says revenue has more than doubled since the previous raise and that it has stayed profitable throughout, with 39 employees at the time the round closed. Its chief executive Christina Qi, who co-founded the company with Luca Lin after running the high-frequency trading firm Domeyard, has said roughly 90 percent of revenue comes from large financial institutions and AI labs.

That second customer category is the reason the valuation reads the way it does. Databento says five of the six largest frontier AI labs hold enterprise agreements with it, and that OpenAI and NVIDIA are customers. Training and evaluating models that reason about markets requires historical price data at tick or nanosecond granularity, with clean timestamps and survivorship-bias-free history, which is precisely the product traditional vendors sell through negotiated annual contracts and weeks of procurement.

Databento's model is deliberately unglamorous. It hosts its own servers and capture hardware inside exchanges and adjacent data centers rather than reselling a wholesaler's feed, then sells the output on demand: full order book data normalized to a single schema, raw packet captures with nanosecond timestamps, and historical data priced by the byte actually consumed. A developer can sign up without talking to a salesperson, which is how an AI research team ends up as a customer.

The fresh capital is going into geography and asset classes. Over the past four months the company has stood up nine data center deployments, including Hong Kong, Japan and London, to serve more markets and deeper history. It also plans to widen coverage beyond futures, options, equities and indices.

The contrast with incumbents is the point of the pitch. Established market data vendors grew up around fixed terminals, bundled entitlements and long contracts, and their cost base reflects physical distribution at scale. Databento is arguing that a small team controlling the capture-to-delivery chain can undercut that structure the way cloud infrastructure undercut owned servers, and the 39-person headcount is being marketed as proof rather than a limitation.

There is a strategic reading here that goes beyond one vendor. As model capability converges, proprietary and hard-to-obtain data becomes the differentiator, and financial tick data is one of the few large, machine-readable corpora that cannot be scraped from the open web. If frontier labs keep buying it at enterprise scale, the market data layer becomes a quiet chokepoint in the AI supply chain, priced accordingly. Databento's $1.1 billion mark is a bet that this holds.