Eustella says European AI model matches US systems

The company’s core claim is not just technical performance but control. Eustella says it uses freely available open-weight models, operates them on its own infrastructure in Europe and can audit, adapt or fully take them offline — a contrast with closed US systems that remain under the vendor’s control.

That positioning matters in a market where governments and enterprises are increasingly wary of where data is processed and who can access the underlying model. Eustella is leaning on a broader European policy push for AI sovereignty, arguing that an open model hosted in Europe remains under European oversight even if its base technology originates elsewhere.

The start-up also said it is trying to strip political bias from the model. In its latest version, the share of answers reflecting the Chinese government’s narrative fell to 1.9% from 30.4%, it said, underscoring both the opportunity and the geopolitical complication of building on top of China-origin open models.

For investors, the message is that competition in AI is broadening beyond the US frontier labs and chipmakers to a second layer of companies focused on localized deployment, compliance and data control. That could create opening for European cloud, software and public-sector AI contracts, while also challenging the pricing power of US incumbents in regulated markets.

The timing also fits a wider scramble across the sector, with major US groups accelerating model releases and governments weighing stricter rules around frontier AI. Eustella said it wants to help secure “a place in the global race” for Europe, but its near-term test will be whether sovereignty and performance are enough to win customers against better-capitalized rivals.