EquiLibre Technologies, based in Prague and started by three former DeepMind researchers, has completed a Series A funding round led by Creandum, bringing its valuation to over $500 million.<\/p>\n
Creandum says this is the largest investment it has ever made in one company.<\/p>\n
EquiLibre says its AI agents trade billions of dollars each day on the S&P 500 and Nasdaq, but other companies also claim to have similar technology, challenging EquiLibre\u2019s claim to be first.<\/p>\n
The founders of EquiLibre Technologies<\/a>, who previously built an AI that beat professional poker players, are now using their skills in financial markets. Martin Schmid<\/a>, Rudolf Kadlec<\/a>, and Matej Moravcik<\/a> started the company in Prague in December 2021. <\/p>\n EquiLibre just closed a Series A round led by Creandum at a valuation of over $500 million. Creandum has not shared the size of the round, only the valuation and that it is their largest investment in a single company.<\/p>\n \u201cThe market is the judge, and the verdict updates every millisecond. That\u2019s what drew us to the problem, and it\u2019s why reinforcement learning is such a natural fit. The question is no longer whether this approach works. It\u2019s how big it can get. We\u2019ve proven the technology in the world\u2019s biggest and most liquid markets,\u201d says Schmid, EquiLibre\u2019s CEO. <\/p>\n The case for reinforcement learning in trading<\/p>\n Schmid, Kadlec, and Moravcik left DeepMind to create DeepStack, the first AI to beat professional players at no-limit poker. They used reinforcement learning, training the model through trial and reward instead of labelled examples, in a competitive, high-stakes setting with clear scoring. Schmid says markets are similar.<\/p>\n The founders returned from North America to start the company in Prague, thinking it would be easier to retain talented people than in Silicon Valley. <\/p>\n \u201cWe want to build a global business from Prague. We are by far the most exciting company working on the frontier of applied AI research here, and this allows us to attract and retain amazing talent. Our ambitions are global, and we aim to build the world\u2019s leading AI trading company,\u201d Schmid says. <\/p>\n This choice helped EquiLibre build a team of about 25, mostly Czech engineers who have worked at Google and other US tech companies.<\/p>\n EquiLibre trains reinforcement learning agents using both historical and live market data, then lets them trade on their own. After trying its approach on crypto markets in 2025, EquiLibre moved into US stocks. Its agents now trade billions each day on the S&P 500 and Nasdaq with Tower Research Capital, a detail not included in EquiLibre\u2019s funding announcement.<\/p>\n The \u201cfirst\u201d problem<\/p>\n EquiLibre says it is the first AI lab to use reinforcement learning live on US markets. However, this is hard to confirm and is different from what Jane Street, a top trading firm, has said.<\/p>\n Jane Street uses reinforcement learning with LLMs, or whatever else we need to train good models, and has tens of thousands of high-end GPUs, much more than EquiLibre. Schmid sees this as a strength, highlighting EquiLibre\u2019s focus on efficiency. <\/p>\n The company says it has not had any negative months since it started, but it has not shared the time period, methods, or drawdown data. <\/p>\n Cameron Sellers<\/a>, vice president at Creandum, says the firm is focused more on scaling than on being first:\u00a0\u201cThis is the largest investment we have ever made, showing the belief that we have in the future scaling of the technology. EquiLibre is doing what the best frontier labs do: picking a domain where the feedback loop is brutal and honest, and letting the technology speak for itself. Martin and the team have proven the approach in the hardest possible environment. We\u2019re thrilled to back them as they scale.\u201d<\/p>\n