A consortium of French companies has placed a roughly $10 billion bid to host one of the European Union's five planned AI gigafactories, according to a report from Bloomberg. The consortium, named AION, is led by Iliad's cloud computing subsidiary Scaleway and proposes a 200-megawatt facility centered on next-generation GPU clusters equivalent to more than 288,000 current-generation Nvidia H100s. This bid is the largest single-country offer disclosed since the European Commission opened its gigafactory selection process under the InvestAI Facility.
Key Facts of the AION Bid
The AION consortium includes a near-complete roll-call of the French AI ecosystem. Key partners consist of GPU and chip-design specialists VSORA and SiPearl; model labs Kyutai and H Company; model-distribution platform Hugging Face; IT-services group Sopra Steria; consultancy Artefact; Atos's compute subsidiary Eviden Bull; and developer-tooling company ZML. Operational support comes from GENCI and Inria, co-leaders of the existing AI Factory France EuroHPC project, with hosting through Opcore, Iliad's data-centre joint venture.
The proposed facility would require a capital commitment of $10bn, matching Iliad chair Xavier Niel's longstanding assertion that France must outspend, not merely match, on AI infrastructure to keep pace with US and Chinese developments. Iliad has invested €20bn in European infrastructure over the past decade, so the AION figure represents roughly half that commitment dedicated to a single facility. The 288,000-H100-equivalent target is positioned as the largest single GPU cluster outside the US hyperscalers and the Microsoft-OpenAI Stargate footprint.
EU Gigafactory Program Context
The bid sits within the InvestAI Facility, a €20bn envelope announced earlier in the year to underwrite up to five gigafactories across the European Union. The European Commission received 76 expressions of interest in the initial sounding round. Among the member states co-financing the program are Spain, Germany, the Netherlands, Italy, Finland, and Portugal. Telefonica is preparing the final Spanish bid, while German and Dutch consortia are also in advanced stages. The formal call window was deferred from late 2025 to the first half of 2026 to give consortia time to assemble multi-billion-euro capital structures.
AION's bid positions France as a single-country contender against these multi-state proposals. The strategic context leans into ongoing pressure on European AI sovereignty. GPU-as-a-service offerings from US providers have dominated European frontier-AI procurement, and OpenAI's pause on its UK Stargate site over energy costs and regulatory uncertainty has created a window for a French-only proposition. France's low-carbon grid offers available power capacity, and the consortium boasts a sovereign software stack combining SiPearl and Eviden hardware with Hugging Face and Kyutai software.
Scaleway CEO Damien Lucas framed the pitch by stating that 'Europe can no longer afford to outsource the foundations of its AI future.' This sentiment underscores the consortium's emphasis on open-source and public-private partnership. The involvement of GENCI and Inria positions the facility as part of European public-research compute infrastructure, in contrast to the more commercially-driven parallel program led by MGX, Bpifrance, Nvidia, and Mistral.
Comparison with Other French AI Infrastructure Projects
AION is not the only French-flagged AI infrastructure project. The MGX-Bpifrance-Nvidia-Mistral 1.4GW Paris-area campus, announced in 2025, represents a separate initiative. Mistral is also separately raising debt and equity for its own Sweden and Paris data-centre footprint. AION's stated differentiator is the open-source-and-public-private framing, aiming to serve the broader European research community rather than solely commercial interests.
The $10bn commitment is notable even in the context of global AI infrastructure spending. Comparable US announcements include the Google-Blackstone $25bn TPU-cloud joint venture and various hyperscaler expansions. The question remains whether Europe's gigafactory program will ultimately operate at the same order of magnitude as the US private-sector build-out.
Next Steps and Unknowns
Several details remain undisclosed. The Bloomberg report did not name the specific French site under consideration, the capital-stack breakdown between Iliad equity, EU grants, member-state co-financing, and private debt, nor the construction timeline. The formal procurement-decision date has not been set by the European Commission. The EuroHPC Joint Undertaking, which runs the selection process, has not yet publicly named the bidder pool or set a final decision date.
The next visible milestone will be the EuroHPC JU's shortlist announcement, expected before the end of the year according to the formal-call cadence published earlier. The winning bids will be selected based on criteria including technical capability, financial viability, strategic alignment with EU AI goals, and potential for open research access. AION's bid emphasizes its alignment with EU sovereignty objectives, but it faces stiff competition from other member states.
Beyond the immediate selection process, the AION bid highlights broader trends in global AI infrastructure. The race to build large-scale compute facilities is accelerating, with governments and private consortia vying to host the next generation of AI training clusters. Europe's approach, through the InvestAI Facility, aims to pool resources across member states to create competitive offerings. However, the relatively small number of gigafactories (five) means that only a few sites will benefit from this EU funding, intensifying competition among member states.
The AION consortium's emphasis on French sovereignty and open-source principles may resonate with EU policymakers concerned about dependency on US technology. The involvement of public research institutions like GENCI and Inria suggests that the facility would be available for academic and public-sector AI research, not just commercial use. This could be a decisive factor given the EU's push for 'AI made in Europe' that prioritizes transparency and trust.
At the same time, the $10bn price tag raises questions about return on investment. While the facility would be a major economic driver for the host region, the commercial viability of such large GPU clusters depends on sustained demand for AI compute. The market for AI training and inference is growing rapidly, but it remains concentrated among a few major players. The consortium's ability to attract both public and private tenants will be critical to its long-term success.
In summary, the AION bid represents a significant commitment by French industry and public institutions to secure a leading role in European AI infrastructure. The consortium's proposal is bold in scale and ambition, but it must navigate a competitive selection process and address remaining uncertainties around financing and site selection. The outcome will have implications not only for France's AI ecosystem but also for the broader European strategy to build sovereign AI capabilities.
Source: TNW | Eu News