EquiLibre Lands $500M as Creandum Places Its Largest-Ever Bet

Prague-based EquiLibre Technologies has closed an undisclosed-sum Series A at a $500 million valuation. The round was led by Creandum, which confirmed it was the largest single investment the European early-stage investor has ever made in one company. EquiLibre builds reinforcement-learning AI agents that trade billions of dollars a day on the S&P 500 and Nasdaq through a partnership with quant firm Tower Research Capital.

The Prague lab was founded in late 2021 by three former Google DeepMind researchers who built DeepStack, the first AI to beat professional players at no-limit Texas hold ’em poker. The new capital lifts EquiLibre’s valuation sharply from its $10 million seed round led by Blossom Capital, the round’s investors say. Most of the fresh money is earmarked for compute, with EquiLibre planning one of the largest AI clusters in Central and Eastern Europe.

From Poker Bots to Wall Street

The three founders met as visiting PhD students at DeepMind’s first international research office in Edmonton, Alberta. Working under Michael Bowling’s computer poker research group, they built DeepStack, an algorithm for imperfect-information games described in the DeepStack paper first published in 2017. Schmid, Kadlec, and Moravcik all hold PhDs in algorithmic game theory or episodic memory modeling, and all later worked at DeepMind in research or engineering roles. They co-authored Player of Games while still at the lab, an algorithm that extended DeepStack-style reasoning to perfect-information games.

The founders returned to their home country and started EquiLibre in Prague in late 2021, before Alphabet shut the Edmonton office in 2023. Schmid told TechCrunch the Czech location was a deliberate choice to tap a large Czech diaspora at Google and other U.S. tech firms. Prague now hosts EquiLibre’s 25-person team and a sister company, BottleCap AI, in the same building. The lab has recruited high-profile advisors, including 2024 Turing award winner Rich Sutton, who is recognized for foundational work on reinforcement learning. Other advisors include Bowling, DeepMind RL lead Csaba Szepesvari, ex-Avast CTO Michal Pechoucek, and Deep Blue co-author Murray Campbell.

Creandum Places Its Biggest Bet on EquiLibre

EquiLibre’s Series A closed in early July 2026 at a $500 million / €438 million valuation. Creandum led the round, and Creandum vice president Cameron Sellers confirmed it was the largest single investment the firm has ever made in one go into a company, the team’s pivot from poker to Wall Street. The size of the round itself was not disclosed, by either EquiLibre or Creandum.

The new round follows a $10 million seed round led by Blossom Capital, which valued EquiLibre at €122.8 million. That seed came after an earlier pre-seed investment from Credo Ventures, a CEE-focused firm whose other bets include ElevenLabs and UiPath. The Series A is a sharp step up in valuation from the seed by any measure. EquiLibre declined to disclose its total funding to date.

Most of the capital will go toward purchasing compute power to scale the operation, the startup said, with plans to bring online one of the largest AI compute clusters in Central and Eastern Europe. Creandum’s earlier AI bets include Black Forest Labs, the FLUX image-model maker. Schmid framed the bet as an investment in reinforcement learning as a category, broader than any single company. ‘This 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,’ Sellers said.

How the Agents Learn to Trade

Reinforcement learning is a class of AI training in which an agent learns by trial and error, maximising a reward signal rather than relying on labelled data. In EquiLibre’s case, the reward signal is profit and loss on actual trades, as the lab describes its approach to building self-learning trading agents. ‘The nice thing about trading and markets is that the scoring is super simple: how much money did the agent make?’ Schmid said. The same approach underlay DeepStack, where the agent learnt to win chips instead of dollars, and the lab now ports those algorithms to the world’s largest equity markets.

The agents train on historical and live market data, the company says. They execute trades autonomously in continuous interaction with other market participants. The lab’s pitch is that the same class of algorithm that cracked poker can find strategies in markets that human quants would not have discovered.

• Train on historical and live market data across crypto and U.S. equities
• Learn through a profit-and-loss reward signal rather than labelled examples
• Execute trades autonomously through partner broker routes
• Adapt continually to changing market conditions and other participants

EquiLibre first validated the technology on cryptocurrency markets before expanding into U.S. equities. The lab started running its reinforcement-learning agents on the S&P 500 and Nasdaq in 2025, the company says. The startup claims its agents have logged a perfect record of zero negative months since inception, a phrase EquiLibre uses to mean that each monthly return has been positive. EquiLibre has not disclosed the methodology or audited performance data behind that claim. The Recursive flagged the claim as unsupported by third-party verification.

For Creandum, that claim is part of why the bet makes sense. ‘The potential total addressable market of trading in the financial markets is one of the biggest on earth, and there are countless funds over the years that have generated quantums of profit that make most venture-backed successes look small,’ Sellers said. Schmid has framed the lab’s work as research more than finance, with the goal of building new things that have never been built before. Creandum described EquiLibre as a research operation first, with finance as the application domain.

Tower Research and the Billions in Daily Volume

EquiLibre’s algorithms do not run a proprietary book. The lab routes its trades through Tower Research Capital, one of the larger independent quant trading firms, with daily volume running into the billions across the S&P 500 and Nasdaq.

The volume figure is one the company uses as a marketing credential. Tower Research brings the execution and market access; EquiLibre contributes the strategy layer that decides when and how to trade. From the lab’s side, the relationship is a statement that reinforcement learning has graduated from research curiosity to production code that other firms want to run.

Schmid describes the markets themselves as the cleanest test of any AI technique. He argues the scoring is brutal and the feedback loop is honest, and that reinforcement learning is now a better fit for trading than for almost any other commercial problem. He frames the field as one where multiple approaches will coexist, since ‘this is not a winner-takes-all market.’ That framing has limits: the company reports no audited performance data and no track record against a public benchmark.

Trading is one of the few fields where technology is the entire game. There’s no sales cycle, and no marketing spend can rescue a weak product.

Martin Schmid, CEO and co-founder of EquiLibre Technologies, said the quote captures his pitch: at EquiLibre, the only thing standing between the research and the market is the agents themselves. The lab describes its culture as ‘researchers and engineers who love hard problems,’ drawn largely from Google and other U.S. tech firms now based in Prague.

EquiLibre Against the Jane Street Compute Wall

Reinforcement learning in markets is compute-bound, and the competitors on this front are not other Prague labs. Trading giant Jane Street states publicly that it already uses reinforcement learning with large language models ‘or whatever else we need to train good models,’ and claims to operate tens of thousands of high-end GPUs. EquiLibre is candid about the gap, with Schmid saying the lab is trying ‘to get more from less’ on compute than rivals with much larger GPU bases.

The Series A is meant to close some of that gap. EquiLibre plans to build one of the largest AI compute clusters in Central and Eastern Europe, the company says, funded largely by the new round. The same round also lets the lab hire more engineers in a market where frontier-AI talent costs have ballooned. Schmid argues the lab’s four-year head start on reinforcement learning in markets counts for something even against a larger compute budget. ‘Because we started four years back, we believe we are ahead,’ he said. The test of that claim will come in 2026 and 2027 as both firms deploy more capacity.

Why Prague, Why Now

Prague is not the obvious place to base a quant-trading AI lab chasing S&P 500 volume, and Schmid is candid about why the team chose it. The founders wanted to recruit from a Czech diaspora at Google and other U.S. tech firms, many of whom had worked with them at DeepMind’s Edmonton office. ‘These were our friends, so we told them, “Hey, guys, we are moving back to Prague, do you want to join us?”‘ Schmid said. Compared to San Francisco, he added, ‘it’s much easier to keep the good people here, because there’s not a new sexy AI thing happening every two months.’

EquiLibre’s 25-person team now includes engineers and researchers from Google and other U.S. tech companies, as well as local Czech hires from the Czech Technical University. The lab shares its building with BottleCap AI, another Prague-based AI startup. The location also keeps the team close to two of its key advisors, Rich Sutton and Michael Bowling, both at the University of Alberta.

The Prague location shows up in the cap table too. Credo Ventures, the CEE-focused firm that led EquiLibre’s pre-seed, also backed ElevenLabs and UiPath. The Czech startup scene has produced some of Europe’s larger recent funding rounds, including Hypefy AI’s €6.3M Series A and Almetra’s €16.3M round in 2026, per The Recursive. EquiLibre’s Series A adds to that list of major Czech funding rounds. Schmid says Prague is the foundation of the lab’s recruitment pitch: ‘We 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.’

What the Numbers Don’t Say

The headline numbers around EquiLibre are striking, and each carries a footnote. The Series A valuation is confirmed by both Creandum and EquiLibre; the dollar size of the round is not. The ‘zero negative months since inception’ claim comes from EquiLibre itself, not from an audited third party. The lab’s stated compute strategy of squeezing more from fewer chips is a research pitch, not a published benchmark against Jane Street or any other major quant firm.

The single most concrete figure EquiLibre shares is the daily trading volume it claims to handle through Tower Research Capital: billions of dollars a day, with no monthly down period since going live. The Recursive has flagged the methodology behind that claim as undisclosed, while EquiLibre’s research lineage through DeepStack, Player of Games, and advisor Rich Sutton is verifiable. For all that, Schmid frames the field as one where multiple reinforcement-learning approaches will coexist, each lab claiming different bets. Whether EquiLibre’s compute strategy keeps it ahead of Jane Street is the open question for the next twelve months. The Series A buys time, talent, and one of the largest AI clusters in Central and Eastern Europe, with a public track record still to be earned.

• $500 million – Series A valuation, led by Creandum (early July 2026)
• €438 million – Series A valuation, in euros (per The Recursive)
• 25 – Employees at EquiLibre’s Prague office
• $10 million – Seed round size, led by Blossom Capital, at a €122.8 million valuation
• Zero – Negative months since inception claimed by EquiLibre (methodology not disclosed)

Frequently Asked Questions

What is EquiLibre Technologies?

EquiLibre Technologies is a Prague-based AI lab founded in late 2021 that builds reinforcement-learning trading agents. The agents route trades through Tower Research Capital and handle billions of dollars in daily volume across the S&P 500 and Nasdaq.

Who founded EquiLibre?

EquiLibre was founded by three former Google DeepMind researchers: CEO Martin Schmid, CTO Rudolf Kadlec, and CSO Matej Moravcik. The trio built DeepStack at DeepMind’s Edmonton office, the first AI to defeat professional players at no-limit Texas hold ’em poker.

What was Creandum’s Series A bet?

Creandum led EquiLibre’s Series A at a $500 million / €438 million valuation in early July 2026. The dollar size of the round was not disclosed. Cameron Sellers, Creandum’s vice president, confirmed it was the largest single investment the firm has ever made in one go into a company.

How does EquiLibre’s AI trade the markets?

EquiLibre trains reinforcement-learning agents that learn by trial and error using a profit-and-loss reward signal. The agents train on historical and live market data, execute trades autonomously, and adapt continuously to changing market conditions.

What is DeepStack?

DeepStack is the first AI program to defeat professional players at no-limit Texas hold ’em poker, built by Schmid, Kadlec, and Moravcik at DeepMind’s Edmonton office and first published in 2017. It is the research foundation EquiLibre’s founders cite for applying reinforcement learning to financial markets.