DeepMind Poker Masters Turn AI to Stock Trading Success
Former DeepMind poker AI creators launch a $500M valued startup applying reinforcement learning to stock trading, boasting zero‑negative months and reshaping global quant finance.
Key Takeaways
The former DeepMind team that built a poker‑beating AI has launched EquiLibre Technologies, a Prague‑based startup now valued at $500 million after a Creandum‑led Series A.
Using reinforcement learning, the company trades billions of dollars daily across the S&P 500 and Nasdaq, reporting a perfect record of zero negative months since launch.
EquiLibre’s agents, developed with Tower Research Capital, were first tested on crypto markets in 2025 and have since expanded to stock exchanges.
Founders Rudolf Kadlec, Matej Moravcik and CEO Martin Schmid come from a poker research background, not finance, and stress that the venture is a laboratory experiment rather than a traditional fund.
Investors cite a massive addressable market, and the startup plans to scale a large compute cluster in Central and Eastern Europe, aiming to out‑perform rivals such as Jane Street.
The company previously raised a $10 million seed round led by Blossom Capital at a $140 million valuation and an earlier pre‑seed backed by Credo, with a Series A valuation jump to $500 million.
EquiLibre aims to squeeze more compute from fewer chips, positioning itself against large firms that rely on tens of thousands of GPUs, while emphasizing a “lab‑first” mindset over pure profit.
Potential Impact Areas
- Enables more efficient, data‑driven investment strategies for asset managers.
- Reduces reliance on human traders, potentially lowering costs.
- Shows feasibility of reinforcement learning in high‑stakes financial markets.
- Creates pressure on incumbents like Jane Street to expand compute resources.
- Inspires new AI startups to explore niche markets with limited data.
- Highlights need for robust validation to avoid systemic risks.
Our Insight
EquiLibre illustrates how AI research originally aimed at games can be repurposed for finance, offering a template for applying reinforcement learning to real‑world markets.
Opportunities include faster trade execution, lower operational costs, and the chance to test novel algorithms in a low‑latency environment.
However, the startup faces significant challenges: intense competition from firms with far larger GPU fleets, the need for rigorous safety checks, and regulatory scrutiny as automated trading expands.
Its “lab‑first” stance may limit short‑term profit focus, but could attract talent seeking innovative research over pure monetary returns.
For the broader industry, success could accelerate adoption of RL techniques, yet it also raises concerns about market stability if many agents act similarly during stress events.
Overall, the venture underscores both the promise and the caution required when scaling AI beyond simulation.
External Credit
Original source: techcrunch.com
Full credit goes to the original publisher. We link to this content for informational and commentary purposes only.