EquiLibre Technologies, a Prague-based AI lab founded by three former DeepMind researchers who built a poker-beating AI, has raised a Series A at a $500 million valuation. The round was led by Creandum and is the largest single check the Swedish VC has ever written, vice president Cameron Sellers confirmed. The valuation marks a sharp jump from the startup's $10 million seed, led by Blossom Capital at a $140 million mark, per Dealroom data.
EquiLibre is applying reinforcement learning — the same technique that powered its founders' poker work at DeepMind — to quantitative trading. In partnership with Tower Research Capital, the startup's agents are now executing billions in daily volume across the S&P 500 and Nasdaq. EquiLibre claims a perfect record of zero negative months since inception, with the system going live on crypto markets in 2025 before extending to equities.
The premise is that markets reward the same kind of agent that wins at poker: one optimized through self-play against a clear scoring function. CEO Martin Schmid argues the loop is cleaner than most RL domains because the reward signal is unambiguous — dollars made or lost.
“The nice thing about trading and markets is that the scoring is super simple: how much money did the agent make?”— Martin Schmid, EquiLibre CEO
Key facts
- 01EquiLibre Technologies closed a Series A at a $500M valuation, up from a $140M seed round led by Blossom Capital.
- 02The round was led by Creandum and is the largest single check the firm has ever written into a company, per VP Cameron Sellers.
- 03EquiLibre's agents trade billions in daily volume across the S&P 500 and Nasdaq in partnership with Tower Research Capital.
- 04The startup claims zero negative months since inception, with crypto rollout in 2025 and stock markets following.
- 05Founders Martin Schmid, Rudolf Kadlec, and Matej Moravcik built DeepStack at DeepMind's Edmonton office, the first AI to beat pros at no-limit Texas hold'em.
The three founders — Schmid, CTO Rudolf Kadlec, and CSO Matej Moravcik — are not finance people. They met as visiting PhD students at DeepMind's Edmonton, Alberta office, the Google-owned lab's first international AI research outpost, which Alphabet shut down in 2023. There they built DeepStack, the first AI program to defeat professional players at no-limit Texas hold'em. Their advisory board now includes Rich Sutton, who won the 2024 Turing Award for foundational work on reinforcement learning.
Creandum's thesis is that automated trading is a market large enough to dwarf most venture outcomes if the technology actually works.
“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”— Cameron Sellers, Creandum VP
Sellers stressed that EquiLibre positions itself as "a lab first, not a finance firm" — a framing that matters because the founders' motivation is research-driven, not financial. Schmid said he is not chasing market efficiency for its own sake; he wants to build systems that have not existed before. That posture echoes the pitch from frontier labs like Ineffable Intelligence, another DeepMind-alumni venture that recently raised $1.1 billion, though most peers in that bracket are U.K.-based.
EquiLibre built its initial team in 2022 by recruiting from the Czech diaspora at Google and elsewhere. Headcount sits at 25. Schmid argues Prague is a retention advantage versus San Francisco, where employees get pulled into a new high-profile AI launch every few months. BottleCap AI operates from the same building, but the regional AI scene is thin enough that EquiLibre faces less internal poaching pressure.
Next on the roadmap is compute. EquiLibre plans to bring online what it expects will be one of the largest compute clusters in Central and Eastern Europe. Pre-seed backers included Credo, the regional VC that previously backed ElevenLabs and UiPath.
“Because we started four years back, we believe we are ahead.”— Martin Schmid, EquiLibre CEO
The competitive risk is real. Trading giant Jane Street has publicly stated it uses RL with LLMs and runs tens of thousands of high-end GPUs — orders of magnitude more silicon than a 25-person Prague startup will field anytime soon. EquiLibre's counter is efficiency: Schmid frames the strategy as "get more from less," arguing the four-year head start on RL-for-trading gives the team a research lead that capital alone will not close. Skeptics will note that quant finance has buried plenty of clever algorithms whose edge decayed once the market caught on.
EquiLibre's pitch to become "the AI lab in trading" lands at a moment when frontier labs are increasingly being judged on revenue, not just benchmarks, and a system that can show audited P&L has a credibility advantage over one that only posts SWE-bench scores. If the zero-down-month record holds through a real drawdown, EquiLibre becomes the first DeepMind-alumni venture to point at cash flow rather than research papers — and a template for RL labs that want commercial validation without selling tokens. The harder question is whether Tower's volume stays exclusive, or whether the next round forces EquiLibre to spin up its own fund and compete with the Jane Streets of the world directly.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
