Databento raises $97M to chip away at Bloomberg's terminal moat

The line that stuck from Fortune's write-up of Databento's Series B wasn't the $97 million or the NEA name on the term sheet. It was the shape of the customer list. Nvidia is a paying customer and a technology partner. OpenAI is in there too. So is the Abu Dhabi Investment Authority. Founder Christina Qi told Fortune that '90% of revenue does come from the large financial institutions and AI labs.' That mix, quant desks next to frontier labs, is a real signal about who is buying clean market data now and why.

The pitch itself is straightforward. Bloomberg Terminals reportedly run $20,000 to $27,000 per seat per year, LSEG Refinitiv is priced in a comparable band, and Databento wants you to add stock data to a cart like an e-commerce order and pay only for what you use. The company captures raw price data from more than 80 trading venues on specialized chips, and Qi's claim is that it is 'the only vendor delivering that full picture over a regular internet connection.'

Take the round numbers as reported, not as settled. Fortune says the raise attracted over $300 million in demand, that total disclosed funding is now around $127 million, and that the 24-person company is profitable every month while sitting on most of the new cash. Qi's own words: 'We haven't really touched much of that $97 million. Our investors are telling us, spend, spend money, and we're trying.' The plan she describes is expanding to 20-plus data centers and more than 100 petabytes of storage.

The honest caveat is that a profitable 24-person outfit built on à-la-carte pricing has not yet been stress-tested against the incumbents deciding to compete on price, and 'profitable every month' at this stage rests on a narrow revenue base with heavy concentration among a handful of large institutional and AI-lab accounts. What the reporting doesn't give you is gross-margin detail on the chip-based capture stack, the exchange licensing terms Databento pays to redistribute, or how it handles the compliance and audit-trail workflows terminal buyers actually rely on.

Still, the interesting read is who is buying. When Nvidia and OpenAI show up on the same invoice list as sovereign wealth and quant desks, the market for clean, real-time price data is quietly widening beyond trading floors into AI training and inference. That is the lane Databento is trying to own before Bloomberg and LSEG decide they need a pay-per-use tier of their own.