Three former researchers from DeepMind, known for creating an AI that triumphed over human opponents in poker, have now adapted their technology for stock trading. Their Prague-based AI venture, EquiLibre Technologies, has reached a valuation of $500 million after securing an undisclosed amount in a Series A funding round, as reported by TechCrunch.
The funding was led by Creandum, which has confirmed that it is the largest single investment made by the firm in a company at once, although exact figures were not disclosed, according to vice president Cameron Sellers.
The link between poker and financial trading lies in their compatibility with reinforcement learning, an AI training method where models learn through a system of rewards. Martin Schmid, CEO of EquiLibre, noted that the evaluation of trading performance is straightforward: “the measure is simply how much profit the agent generates.”
EquiLibre’s algorithms have been executing trades worth billions daily across major indices like the S&P 500 and Nasdaq in collaboration with quantitative trading firm Tower Research Capital. The startup asserts that its algorithms have maintained a flawless record without a single negative month since launching in 2025, indicating a consistent positive performance in their investments.
By integrating its AI into quantitative hedge funds, EquiLibre is operating in an environment where the automation of trading is already prevalent, and successful innovations can convert rapidly into profits, making the venture enticing for investors like Creandum, according to Sellers.
“The overall potential market for trading in finance is among the largest globally, with numerous funds historically generating profits that dwarf those of many venture-backed companies,” Sellers explained. However, he emphasized that EquiLibre identifies itself primarily as “a lab first, not a finance firm.”
Schmid, along with co-founders Rudolf Kadlec and Matej Moravcik, does not come from a financial background, which is not their main motivation, as Schmid expressed. “I’m not driven by the goal of making markets efficient. Our enthusiasm comes from constructing new systems that have yet to be created, and it’s incredibly enjoyable to develop this,” he stated.
The pursuit of cutting-edge AI by former DeepMind researchers is garnering attention from venture capitalists. Another recent instance is Ineffable Intelligence, which raised $1.1 billion. While many such firms are located in the UK, EquiLibre stands out as a notable exception.
The founding trio met while pursuing their PhDs at DeepMind’s initial international AI research center in Edmonton, Canada, before its closure in 2023. During their time there, they developed DeepStack, the first AI to defeat professional players in no-limit poker. They collaborated with leading professors who are now part of EquiLibre’s advisory board, including Rich Sutton, who received the Turing Award in 2024 for his advancements in reinforcement learning.
After establishing their startup, the founders decided to return to Czechia, where they had numerous professional contacts and where a considerable Czech diaspora existed, particularly at Google. “We reached out to friends and said, ‘We’re going back to Prague, want to join us?’” Schmid shared.
This initial outreach helped them assemble their founding team in 2022, growing to 25 employees currently. Schmid noted that choosing this location continues to yield benefits. “In contrast to San Francisco, it’s easier to retain talented individuals here since there isn’t a flashy AI project emerging every other month.”
EquiLibre is not the only emerging AI company in the area, with another venture, BottleCap AI, operating from the same facility.
Nonetheless, EquiLibre is recognized as a standout AI startup in the region and plans to enhance its computing infrastructure, aiming to establish one of the largest computing clusters in Central and Eastern Europe.
While the startup has not disclosed the total funding it has garnered to date, Schmid revealed it previously closed two earlier funding rounds, including pre-seed investments from Credo, a VC firm focused on CEE, which has also supported ElevenLabs and UiPath. Data from Dealroom shows that EquiLibre’s $10 million seed round, valued at $140 million, was led by Blossom Capital.
Sellers confirmed a significant increase in EquiLibre’s valuation to $500 million during the Series A funding, particularly as the landscape for reinforcement learning in trading has shifted favorably. “Initially, there was skepticism,” Schmid remarked. “However, RL has now become the norm. We consider ourselves ahead since we began this journey four years ago.”
However, competition poses a risk for the startup. Trading behemoth Jane Street claims it already implements reinforcement learning with large language models, and boasts thousands of advanced GPUs compared to EquiLibre, which aims to maximize efficiency from a smaller number of chips.
To be recognized as “the AI lab in trading,” EquiLibre will need to navigate these competitive challenges carefully. Schmid commented, “This market doesn’t operate on a winner-takes-all model.”
