A $97 million round led by New Enterprise Associates, with participation from DRW, Redpoint Ventures and Tribe Capital, has thrust Databento into the fintech spotlight. The startup aims to build what it calls the market data platform for modern finance, openly competing with the Bloomberg terminal. Investor demand made the round oversubscribed – a signal that the market sees room for a credible alternative.

Behind this raise lies a disruptive thesis: the Bloomberg terminal, with its monolithic interface and high costs, is a 20th-century model, while financial institutions need granular data, accessible via APIs and integrable into proprietary software pipelines. It’s not just about saving on subscription fees; the real stake is data control and the ability to train or query AI models on real-time and historical market data.

For teams working with Large Language Models and quantitative analysis, the difference is structural. A terminal enforces passive consumption; an API-first platform like Databento lets you directly feed backtesting frameworks, algorithmic execution and LLM inference without intermediate manipulation. Those deploying on-premise, in particular, gain a missing piece: a normalized, programmable data flow that can live on local infrastructure and comply with data residency policies without relying on proprietary appliances.

Sovereignty is crucial. Banks and asset managers must navigate strict regulations on where and how financial data is processed. Bloomberg’s approach – data visible only inside a closed ecosystem, often on cloud terminals or dedicated hardware – is incompatible with the need to keep models and data within the same on-premise perimeter, especially when fine-tuning LLMs on sensitive market data. Databento promises usage licenses and deployment flexibility that, if well executed, shift the balance toward self-hosted architectures.

This is not a simple vendor war. The deal signals a shift in incentives: value no longer lies in exclusive access, but in the ability to compose data sources as building blocks of proprietary AI infrastructure. Quantitative traders and research divisions at large banks already build their own stacks, but until now they had to bypass the market data ingestion problem with costly and fragile hybrid workarounds. The emergence of a well-funded pure player like Databento could accelerate the unbundling of the classic financial desktop.

Who stands to gain? Those who have already invested in internal MLOps skills and inference hardware (GPUs, high-VRAM servers, low-latency storage), because they can finally leverage institutional-grade data without suffocating licensing constraints. Who loses? Businesses built on closed data redistribution and, naturally, Bloomberg itself, if the API-first migration becomes structural. Whether Databento can cover the asset classes and historical depth that make the Bloomberg terminal a de facto standard remains to be seen; but the capital raised suggests that for the first time, the threat is being taken seriously.