Databento.

A data shop that bills like a software company

It is 9:28 a.m. somewhere, and a quant who has never spoken to a salesperson is about to do something that used to take a quarter of a year. She types an API key. She picks an exchange. She streams a live order book onto her screen and pulls back four years of tick history in the same breath. No signed contract. No minimum spend. No three-week procurement dance. She pays for exactly what she used, and goes back to work.

That quiet, unremarkable moment is the whole point of Databento. The company is a licensed market data provider that delivers live exchange feeds and decades of historical data through a single API and a no-code browser - across 45+ trading venues, covering equities, options, and futures. It is headquartered in Salt Lake City, runs across eight time zones, and serves a customer list that runs from pre-product startups to funds managing billions.

The unglamorous truth: market data is one of the largest recurring costs on a trading desk, and for decades it came wrapped in long contracts, opaque pricing, and integration projects measured in months. Databento's bet is that none of that is actually necessary.

The data was good. Getting it was miserable.

Anyone who has worked on a trading floor knows the ritual. You need a feed. You email a vendor. A salesperson appears. There are PDFs. There are minimums. There is an integration team translating one exchange's idiosyncratic format into another's. Months pass. Budgets balloon. And the data itself - the actual ones and zeros of what traded, when, and at what price - was never the hard part.

The friction was the product. Legacy providers had built businesses on the assumption that market data was scarce, complicated, and best sold by humans over long horizons. For a small fund or a solo researcher, the cost of admission was simply too high. For a large one, the overhead was a permanent tax.

The founders had lived inside this problem. They had built and run trading strategies that depended on clean, fast, normalized data, and they had paid the toll personally. So the question that started the company was less "what's missing in the market" and more "why do we keep putting up with this?"

From the hedge fund floor to a software counter

Databento was founded in 2019 - originally incorporated under the decidedly less catchy name Elisify - by Christina Qi and Luca Lin. Qi had co-founded Domeyard LP, a high-frequency trading hedge fund, and had done time in sales and trading at UBS and Goldman Sachs. In other words, she had been the customer she would later try to rescue.

The bet was specific: take institutional-grade data and sell it the way modern software is sold - self-service, pay-as-you-go, instantly available. Charge people for what they use, not for a year they might not need. It sounds obvious now, which is usually the sign that it wasn't.

They staffed it with people who had felt the same pain - traders and engineers from Two Sigma, Flow Traders, and Tower Research Capital, alongside builders from Google, Stripe, and AWS. The result is a team that is 86% science and tech backgrounds, 46% institutional buy-side veterans, and 60% women and minorities. For an industry that historically looked like a 1990s trading pit, that last number is its own kind of statement.

One login for the whole data stack

What you can actually do with Databento is the part that converts skeptics. You can stream a live feed, replay raw packet captures, pull historical tick data going back years, or enrich it all with reference data - without leaving one account. Developers get official client libraries in Python, C++, and Rust. The data arrives in DBN, the company's open-source binary encoding built to be fast enough that nobody has to apologize for it.

Six products, one credit card. The market data equivalent of finally finding everything in one aisle.

The numbers do the talking

Plenty of companies claim they're disrupting an incumbent. Fewer can point to a revenue line bending almost vertically. Around its 2024 raise, Databento reported 985% year-over-year revenue growth and more than 7,000 new customers acquired in a single funding cycle - a sign that the self-service model isn't just elegant on a slide, it's converting.

The money came with notable names. The 2021 round drew Indicator Ventures, Blindspot Ventures, Redpoint, and Tribe Capital. The 2024 top-up added trading and capital-markets insiders - Belvedere Trading, Clear Street, and Lightscape Partners - the kind of investors who use this category of product and are therefore harder to fool. Databento also sits inside the ecosystems that matter: it's an approved data vendor in CME Group's marketplace and its datasets are available on QuantConnect.

Make market data boring

Databento's stated mission is almost suspiciously plain: create a better, simpler, faster data experience, and let customers pay only for what they use. There is no manifesto about reinventing finance. The ambition is to make a historically painful purchase so unremarkable that nobody thinks about it - which, for infrastructure, is the highest compliment available.

The longer game is to become the default source of institutional-grade market data for every firm, not just the ones with a Bloomberg budget. The 2024 funding is pointed squarely at that: extended data history, deeper European coverage, global indices, and new pricing plans. The map keeps growing; the friction keeps shrinking.

The next desk that never had to ask permission

Trading is being rebuilt by people who write code, not people who sign contracts - and the same is true of the AI and machine learning teams now treating market data as training fuel. Both want the same thing: clean data, instantly, priced like a utility. That is precisely the world Databento is building toward, and it's why a 36-person company in Utah keeps showing up on the buy lists of firms a thousand times its size.

Return, then, to that quant at 9:28 a.m. A decade ago she'd have been on hold with a vendor, watching a quarter evaporate before her first backtest. Now she's already three feeds deep, paying by the byte, building. The data was never the bottleneck. Databento just removed everything else around it - and handed the morning back.