Three former DeepMind researchers, who previously developed an AI capable of defeating human poker players, have now turned their attention to stock trading—and the strategy seems to be yielding results. The connection between poker and Wall Street lies in their suitability for reinforcement learning, an AI training method where self-learning models are driven by rewards. As EquiLibre CEO Martin Schmid explains, “The nice thing about trading and markets is that the scoring is super simple: how much money did the agent make.”

This isn’t just theoretical. In collaboration with quantitative firm Tower Research Capital, EquiLibre’s algorithms have been handling billions in daily trading volume across the S&P 500 and NASDAQ. The startup reports that its agents have performed well since entering crypto markets in 2025 and now stock exchanges, boasting “a perfect record of zero negative months since inception,” meaning each month has ended with overall investment gains. By applying its AI to quantitative hedge funds, the company operates in a field where automation is standard, and success can quickly translate into profit. This potential attracted Creandum, according to a partner there: “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.” However, he noted that EquiLibre explicitly defines itself as “a lab first, not a finance firm.” Schmid added, “I’m not doing this because I’m excited about making markets efficient. I’m doing this because we are all excited about building new things that have never been built before, and this is a lot of fun to build.”

EMBED_PLACEHOLDER_0

The prospect of frontier AI from DeepMind alumni is a hot area for venture capitalists. Another recent example is Ineffable Intelligence, which raised $1.1 billion. Most such ventures are based in the U.K., but EquiLibre is a notable exception. The founding trio were visiting PhD students at Google’s first international AI research office in Edmonton, Alberta, Canada (shut down by Alphabet in 2023). There, they created DeepStack, the first AI to beat professional players at no-limit poker, or Texas hold’em. They also worked with professors who now serve on the startup’s high-profile advisory board, including Rich Sutton, who received the Turing Award in 2024 for his work on reinforcement learning.

To launch their startup, EquiLibre’s founders returned to their home country, Czechia. “This is where we had a lot of people we had worked with, and there was a large Czech diaspora at Google and other places,” Schmid said. “These were our friends, so we told them, ‘Hey, guys, we are moving back to Prague, do you want to join us?’” That helped EquiLibre build its initial team in 2022 and reach its current headcount of 25. Schmid says the location continues to pay off. Compared to San Francisco, “It’s much easier to keep the good people here, because there’s not a new sexy AI thing happening every two months.”

EquiLibre isn’t the only hot AI startup in town—BottleCap AI is in the same building—but it stands out as a notable talent hub in the region. The company next plans to scale its compute infrastructure, aiming to bring online one of the largest compute clusters in Central and Eastern Europe (CEE). While it declined to disclose total funding, Schmid said it previously raised two rounds, with pre-seed backers including CEE-focused VC firm Credo, which also backed ElevenLabs and UiPath. According to Dealroom data, EquiLibre’s $10 million seed round was led by Blossom Capital at a $140 million valuation. The Series A, at a $500 million valuation, marked a significant jump, as confirmed by a investor. This comes as reinforcement learning (RL) has gained traction, including in trading. “When we started, people were skeptical,” Schmid said. But now RL is standard. “Because we started four years back, we believe we are ahead.”

EMBED_PLACEHOLDER_1

Still, the startup risks being leapfrogged by competitors. Trading giant Jane Street, for instance, says it already uses RL with LLMs, “or whatever else we need to train good models.” It also claims to have “tens of thousands of high-end GPUs,” while EquiLibre aims to extract more compute from fewer chips and “get more from less,” according to Schmid. Given Jane Street’s profitability, EquiLibre will need to play its cards well to achieve its goal of being known as “the AI lab in trading.” But unlike poker, this may not be a zero-sum game. As Schmid puts it, “This is not a winner-takes-all market.”