DeepMind Trio Behind Record-Breaking Poker AI Pivot to Quantitative

What happened

EquiLibre Technologies, an AI research lab based in Prague, has reached a valuation of $500 million after securing a Series A funding round. The investment was led by the venture capital firm Creandum, which confirmed that the deal represents its largest single investment in any company to date. The startup was founded by a trio of former DeepMind researchers—CEO Martin Schmid, CTO Rudolf Kadlec, and CSO Matej Moravcik—who are best known for their pioneering work in game theory. At EquiLibre, they have successfully adapted the reinforcement learning techniques originally used to master poker for the purpose of trading stocks and cryptocurrencies. The firm has already begun operating at scale, partnering with the quant firm Tower Research Capital to trade billions in daily volume across major indices like the S&P 500 and Nasdaq.

What's new in this update

The most significant development in this update is the confirmed transition of EquiLibre’s technology from experimental crypto markets into the heart of global finance. While the startup began deploying its agents in cryptocurrency exchanges in 2025, it has now officially integrated with major stock exchanges. According to company disclosures, the algorithms have maintained a perfect record of zero negative months since inception, meaning every monthly period has concluded with an overall investment gain. This level of consistency in the volatile Nasdaq and S&P 500 environments has validated the firm's self-learning models. Additionally, the Series A funding marks a shift in the VC landscape, where traditional European firms are now placing record-breaking bets on specialized AI labs that operate outside of the typical Silicon Valley or London hubs.

Key details

EquiLibre’s approach is rooted in reinforcement learning, a training method where AI models learn through a system of rewards. In the context of trading, CEO Martin Schmid notes that the reward mechanism is uniquely straightforward: the amount of money the agent generates. This simplicity allows the models to iterate rapidly based on market feedback. The startup currently employs approximately 25 people in its Prague headquarters, drawing heavily from the 'Czech diaspora' of researchers who previously worked at Google and other global tech giants. The technical advisory board is equally formidable, featuring Rich Sutton, a 2024 Turing Award recipient often cited as one of the founding fathers of reinforcement learning. This combination of top-tier academic pedigree and proven financial performance has made EquiLibre a standout in the increasingly crowded field of AI-driven quantitative finance.

Background and context

The founding trio first gained international acclaim as visiting PhD students at DeepMind’s research office in Edmonton, Canada. During their tenure, they developed DeepStack, the first artificial intelligence program to beat human professionals at no-limit Texas hold ’em. Poker is considered a 'game of imperfect information,' making it a significantly more complex challenge for AI than games like chess or Go. The researchers realized that the same uncertainty and strategic depth found in poker are inherent in financial markets. After Alphabet closed the DeepMind Edmonton office in 2023, the team chose to return to their home country, Czechia, to build a startup that could commercialize these insights. This move mirrors a broader trend of DeepMind alumni founding frontier AI companies, such as the U.K.-based Ineffable Intelligence, though EquiLibre is unique in its specific focus on quant-driven capital markets.

What to watch next

As EquiLibre scales its operations, the industry will be watching to see if its 'zero negative months' record can survive longer-term market cycles and increased volatility. The startup explicitly identifies as a research lab first rather than a finance firm, suggesting that their future roadmap may include applications beyond trading. However, the immediate focus remains on capturing a larger share of the total addressable market in global financial trading. Observers should also monitor the recruitment efforts in Prague; the city is becoming a central node for AI research as EquiLibre attracts talent back from major tech hubs. Further partnerships with other high-frequency trading firms or hedge funds are likely as the company seeks to deploy its Series A capital to further refine its self-learning agents for even more complex asset classes.