Institutional market data has long been associated with complex licensing, specialist infrastructure and lengthy procurement processes. Much of that complexity is unavoidable: exchange data can be technically demanding, commercially restricted and costly to distribute at scale. However, the expectation around how that data should be accessed is changing.
As financial firms become more accustomed to cloud services, APIs and self-service technology platforms, there is growing pressure on market data providers to make their products easier to discover, price, integrate and consume. The underlying data may remain complex, but the process around it is increasingly expected to look more like modern software infrastructure.
Databento is one of the providers pursuing that model. Founded in 2019, the company says it now serves over100,000 users across 189 countries, with financial institutions accounting for the majority of its revenue. More recently, it has also seen demand from large AI labs and other newer categories of data user.
The company provides live and historical market data, including Level 1, Level 2 and Level 3 order book data, reference data and packet captures. Users can access the data online or programmatically, including through APIs for real-time streaming.
“Our goal is to make it easy for any use case. To be able to come in, even if it’s your first time ever looking at data, and for it to be understandable, intuitive and easy to use,” says Christina Qi, CEO of Databento, in conversation with Market & Alt Data Insight.
Changing Expectations Around Access
That focus reflects a broader change in how data buyers think about infrastructure. For many users, especially developers and quantitative researchers, the ability to see a price, select a dataset, choose an output format and start working with it quickly is becoming less of a differentiator and more of an expectation. Databento allows users to configure elements such as date range, file format, timestamping and file splitting before downloading data directly.
At the same time, simplicity at the point of access does not remove the need for significant infrastructure underneath. Qi says Databento holds petabytes of historical data, including OPRA options data back to 2013 and CME data back to 2010, with history varying by dataset. The company has also built and deployed its own hardware across multiple locations and operates its own NTP service.
For more latency-sensitive users, the technical characteristics of that infrastructure remain important. The questions are familiar: where is the data captured, how is it timestamped, how are redundant feeds arbitrated, and what does end-to-end latency actually mean in practice?
“Customers ask us, ‘What’s your latency? How do you timestamp the data? How do you arbitrate feeds? How is the network built?’ And we’ll literally diagram it for them,” says Qi. “We’re transparent down to the FPGA and packet level.”
That kind of disclosure reflects another shift in the market. Sophisticated users increasingly want visibility not only into the data itself, but also into how it is collected, processed and delivered.
“Buying market data shouldn’t require getting on the phone with a salesperson just to find out what it costs,” says Qi. “We believe customers should be able to see the price, understand exactly what they’re getting, and know they’re being charged fairly.”
New Types of Data Consumer
The audience for institutional-grade market data is also broadening. Qi says around 90% of Databento’s revenue still comes from financial institutions, but the company is seeing increasing demand from large AI labs, which is potentially significant for the wider data market. Historically, high-resolution exchange data has been consumed primarily by banks, brokers, hedge funds and proprietary trading firms. AI developers, technology companies and independent researchers now represent additional categories of demand, although the specific use cases are often less visible.
Databento also serves students and individual developers. They account for a much smaller share of revenue, according to Qi, but they form part of a broader community around the product.
The company operates an open product roadmap where users can request new datasets and features, giving some indication of where demand is developing. Qi points to interest in areas including fixed income, crypto and FX, although Databento’s current business remains centred largely on exchange-traded data.
Market data is likely to remain a technically and commercially complex business. But the experience of buying and consuming it seems to be gradually moving towards a more open, software-like model. For data providers, that means competition is increasingly shaped not only by coverage, quality and latency, but also by how easily users can understand what they are buying, how much it costs, and how quickly they can put it to work.
Christina Qi is a speaker at the A-Team Group/Eagle Alpha Alternative Data Conference in New York on 10th September.
Subscribe to our newsletter
