EquiLibre hits €438M valuation to scale AI trading agents

Prague's EquiLibre Technologies just repriced from a €122.8 million seed valuation to a €438 million Series A in under two years — a roughly 3.5x mark-up on a company that runs reinforcement-learning agents against live financial markets. The round is led by Creandum. The stated use of proceeds is compute: a large cluster in Central and Eastern Europe to run agents that, per the company, trade billions of dollars daily.

What actually shipped

Strip the valuation and here is the substance:

• EquiLibre Technologies, founded in late 2021, builds AI trading agents using reinforcement learning.
• Series A closed at a €438 million valuation, led by Creandum.
• Prior funding: a $10 million seed led by Blossom Capital at a €122.8 million valuation, after a pre-seed from Credo Ventures.
• Capital targets a CEE compute cluster to scale the agents.

Note what is absent from the disclosure: the size of the Series A itself, revenue, assets under management, and the trading performance of the agents. "Trades billions daily" is turnover, not profit. Those are different numbers, and only one of them justifies a €438M mark.

Why the valuation ran ahead of the disclosure

The interesting mechanism here is who is paying. Creandum, Blossom Capital and Credo Ventures are frontier-software investors — Credo backed UiPath and ElevenLabs. This is an AI-lab thesis applied to trading: back the research team and the compute, not a hedge fund's track record. The pitch is that reinforcement learning generalises to markets the way it did to games, and that the moat is the training loop plus the cluster.

That is a bet on a mechanism, not on audited returns. RL agents that win in simulation routinely fail on live markets because markets are adversarial, non-stationary, and reflexive — the agent's own trades move the price it is learning from. A €438M valuation prices in the assumption that EquiLibre has crossed that gap. The round does not prove it has.

What it signals for CEE

The talent-to-capital arbitrage is now visible

The deeper story is geographic. A Prague team is raising a top-tier Western Series A and building the compute in the region rather than relocating to London or New York. That is the UiPath pattern repeating — CEE engineering, Western capital, ambition priced globally. For Romanian and CEE founders, the read is concrete: frontier-AI valuations are now available without a US zip code, provided the technical claim is strong enough to attract funds like Creandum.

Compute is becoming the CEE bottleneck

EquiLibre's plan to build a regional cluster lands on the same constraint we flagged in the data-center flexibility piece: CEE capital is chasing GPU capacity faster than the grid and facilities can supply it. Expect compute — power, hookups, hosting — to be the gating factor for the region's AI ambitions, not funding.

For allocators, the caution is straightforward. Trading is one of the few AI verticals where the product's quality is measurable to the decimal and where losses are immediate. The market will grade this bet in P&L, not in press releases.

Key takeaways

• EquiLibre raised a Series A at a €438M valuation led by Creandum — a ~3.5x jump from its €122.8M seed, priced on RL research and compute, not disclosed returns.
• The round size, AUM and performance are undisclosed — "trades billions daily" is turnover, and the hard question of whether RL agents survive live markets remains open.
• A Prague team is building its cluster in CEE — reinforcing the UiPath-style pattern of regional talent, Western capital, and compute as the next binding constraint.