The French AI provider Mistral has completely separated data processing for customers by region since August 11, 2026, and has set up separate access points for Europe and the USA. Additionally, a new tariff with a contractually guaranteed uptime for mission-critical applications is being launched. In return, five major European corporations are committing to multi-year purchases, which are expected to create up to one gigawatt of computing power in Europe by 2030.
Customers will choose processing region and service level
With the Regional Endpoints now generally available, Mistral sends customer requests to either a European or a US address. The processing remains in the chosen region, with the exception of narrowly defined, contractually secured transfers to individual subcontractors. For companies with strict data protection or export regulations, this choice is now technically verifiable for the first time, not just contractually promised.
At the same time, the Priority Tier is being launched as a beta version: a tariff with reserved processing quotas and a contractual availability guarantee of 99.5 percent, which mathematically allows for about three and a half hours of downtime per month. Mistral positions itself as the only European AI provider that offers both regional choice and a contractually binding availability guarantee. Additionally, the open model GLM-5.2 from the Chinese provider Z.ai will also run on Mistral’s infrastructure – under the same regional rules as its own models.
Five corporations secure multi-year computing power
For capacity expansion, Mistral is establishing so-called European Compute Units. Companies pay in advance for a multi-year share of future computing power, thus securing access before new data centers are even built. According to Mistral, five companies have joined the coalition as the first members: the travel technology company Amadeus, the Dutch chip equipment supplier ASML, the consulting firm Capgemini, the French state bank Caisse des Dépôts, and the shipping company CMA CGM.
Amadeus CEO Luis Maroto justified the move by stating that capacity and control over operations are becoming increasingly important. ASML CEO Christophe Fouquet explained, according to Mistral, that his company wants to build the capability to develop and operate AI applications on its own terms. Overall, the coalition is expected to help Mistral’s European computing capacity grow to up to one gigawatt by 2030 – a target figure that the company itself mentions and which is not independently verified.
EU sovereignty debate remains controversial
According to a report by Fortune from August 5, Mistral CEO Arthur Mensch has given Europe a time window of about two years to build its own AI infrastructure before dependence on US technology becomes too great. Critics counter that Mistral continues to operate its models predominantly on Nvidia chips and partly in US cloud infrastructure. True independence has not been achieved in this way. Just in July, Mistral entered into a billion-dollar partnership with Microsoft for additional data center capacity in Europe – a deal that remains controversial as a sovereignty project due to its proximity to the US corporation.
The new coalition is part of a broader European race for computing capacity. In parallel to the private initiative, the EU is tendering for up to seven AI gigafactories as part of EuroHPC, for which German locations are also applying. The German government has recently increased its own funding for a German application to one billion euros. Whether private capacity commitments like those from Mistral will compete with or complement state-funded gigafactories for the same chips and skilled workers remains open.
It will be crucial whether companies from regulated industries actually adopt the EU endpoints and the Priority Tier in productive systems or whether it remains at pilot projects. The first test will come with the EU tender for the gigafactories, with the application deadline ending on November 12, 2026: only then will it become clear how much of the promised European computing power will actually be built – and who will ultimately operate it.
