On May 30, 2026, SoftBank turned France’s annual Choose France summit into a hard infrastructure test for Europe’s AI ambitions: €45 billion for 3.1 GW of AI data center capacity by 2031, with a stated path to €75 billion and 5 GW across France.
That timing matters because the announcement was not just another foreign direct investment trophy for Paris. It was framed as Europe-scale AI infrastructure, anchored in northern France and tied directly to President Emmanuel Macron’s push to make France a serious AI compute hub, according to CryptoBriefing.
May 30: SoftBank Turns Choose France Into an AI Compute Pitch
SoftBank’s first phase targets the Hauts-de-France region, with sites named in Dunkirk, Bosquel, and Bouchain. The first facilities are expected to come online by 2028, with full first-phase completion targeted for 2031.
The official SoftBank announcement described the broader plan as up to €75 billion for 5 GW of capacity, calling it the group’s largest AI infrastructure investment in Europe. The company said the project is designed to expand access to high-performance compute in France and support demand from AI companies, cloud providers, enterprises, public institutions, and research organizations.
“AI is entering a new era, and the countries that build the infrastructure for this transformation will shape the future of technology, industry and society,” said Masayoshi Son, Chairman and CEO of SoftBank Group Corp.
MLXIO analysis: the real signal is control. OpenAI Codex and white-collar work show why AI models are only as useful as the compute available to train and run them. By putting tens of billions into French capacity, SoftBank is positioning itself not just as an investor in AI companies, but as an owner of the physical layer those companies need.
2028 Becomes the First Test of the €45B Timeline
The headline figure is €45 billion, but the operational target is the harder number: 3.1 GW of AI data center capacity in roughly five years.
That kind of buildout means more than shells, servers, and land. AI data centers require dense power delivery, cooling systems, networking, security, and hardware cycles that age quickly as newer compute arrives. The sources do not break down SoftBank’s spending across chips, construction, power infrastructure, or operations, so any exact allocation would be speculative.
Still, the power requirement alone explains why France won this round. CryptoBriefing points to France’s relatively competitive electricity costs and nuclear-heavy grid. SoftBank’s own release cites France’s grid infrastructure, industrial land availability, engineering talent, and national AI commitment.
| Milestone | Source-backed detail |
|---|---|
| May 30, 2026 | SoftBank announces France AI data center commitment |
| First phase | €45 billion for 3.1 GW in Hauts-de-France |
| First operations | Facilities expected from 2028 |
| Phase-one completion | Targeted for 2031 |
| Full ambition | Up to €75 billion for 5 GW across France |
The risk sits in execution. CryptoBriefing flags permitting delays, construction bottlenecks, and grid connection timelines as potential obstacles. The 2028 opening target will show whether this is a funded industrial program or an oversized political announcement.
Northern France Gets the Power, Land, and Policy Stack
SoftBank’s chosen geography is not incidental. Dunkirk, Bosquel, and Bouchain point to a strategy built around industrial land, grid access, and local political support.
The Bouchain site also brings EDF into the story. In SoftBank’s release, EDF Chairman and CEO Bernard Fontana said the project selected for Bouchain shows France can host large-scale digital infrastructure supported by “competitive, sovereign and low-carbon electricity.”
That phrasing matters. France is not selling itself only as a place to park servers. It is selling power quality, sovereign industrial policy, and lower-carbon baseload as a package. For AI infrastructure, where electricity is not a side cost but a core input, that package is the pitch.
SoftBank will also partner with Schneider Electric in Dunkirk. The plan includes two facilities: one operated by SoftBank to manufacture enclosures, and one operated by Schneider Electric to integrate data center power modules. That adds a manufacturing angle to what would otherwise be a pure compute-capacity story.
For readers tracking the difference between endpoint hardware and the infrastructure behind it, this sits far above consumer and enterprise device rollouts such as MLXIO’s coverage of the 96GB ThinkPad P16s Gen 5 landing early in Europe with AMD or the €49.90 Casio W-738H rollout across the EU. SoftBank is betting on the layer those devices and applications ultimately connect to.
SoftBank’s AI Strategy Moves From Stakes to Steel, Power, and Sites
SoftBank’s France plan fits a broader AI infrastructure push. Fortune reported that the French commitment follows SoftBank’s March announcement of a large-scale data center project in Ohio, potentially channeling $500 billion to install 10 GW of capacity. Fortune also cited SoftBank’s work with OpenAI, Oracle Corp., and Abu Dhabi’s MGX on the $500 billion Stargate initiative in the U.S.
The France project therefore does not look isolated. It looks like part of Masayoshi Son’s attempt to secure AI compute bases in major markets.
CryptoBriefing also says SoftBank’s annual net profit reportedly quadrupled to over $32 billion by mid-May 2026, driven largely by AI-related investments. That gives context to the timing: SoftBank is leaning into AI while its AI-linked gains are visible.
But infrastructure changes the risk profile. Venture-style AI exposure can rise or fall with valuations. Data centers lock capital into land, power contracts, construction schedules, and hardware refresh cycles. MLXIO analysis: this is a heavier, slower, more political strategy than backing software companies. It may produce durable capacity, but only if demand, financing, and execution line up.
Fortune also reported questions over whether Son can tap enough financing to realize all of his AI ambitions, including a scaled-back margin loan plan tied to SoftBank’s OpenAI stake. That does not invalidate the France plan, but it makes funding structure a key watch item.
Governments, Utilities, and AI Buyers Will Not See the Same Deal
For the French government, the SoftBank deal validates Macron’s pitch that France can attract major AI infrastructure capital. The official statement by Roland Lescure, Minister of Economy, Finance, Industrial, Energy, and Digital Sovereignty, emphasized jobs, digital infrastructure, fast-track procedures, and digital sovereignty.
For AI companies and enterprise buyers, the potential gain is more local high-performance compute capacity. SoftBank says the data centers will serve AI companies, cloud providers, enterprises, public institutions, and research organizations. That could matter for buyers that want capacity closer to European customers and institutions.
For utilities and local authorities, the pressure point is power. 5 GW of capacity is a huge electricity commitment. The sources point to France’s nuclear-heavy grid and grid access as advantages, but they do not provide details on water use, cooling requirements, local environmental reviews, or household-power tradeoffs.
Local communities will likely judge the project on a narrower question: does the industrial activity justify the land and energy footprint? The sources mention job creation and local industrial development, but they do not quantify expected jobs.
By 2031, France Either Converts Capacity Into AI Output—or Exposes the Limits of Buildout Politics
The strongest version of this story is clear: by 2031, France has become one of Europe’s central AI compute locations, SoftBank has anchored a major infrastructure base, and local companies, researchers, cloud providers, and public institutions gain access to capacity that would otherwise be harder to secure.
The weaker scenario is just as clear. Permitting slips. Grid connections lag. Financing tightens. Hardware cycles move faster than construction. AI revenue fails to justify the capital intensity. In that case, the project becomes a costly test of whether political ambition can keep pace with compute economics.
The first evidence will come before 2031. Watch the 2028 facility target, the buildout progress at Dunkirk, Bosquel, and Bouchain, and the Schneider Electric manufacturing cluster. If those pieces move on schedule, SoftBank’s France bet starts to look like infrastructure strategy. If they stall, the €45 billion headline will read more like an option than a commitment.
Impact Analysis
- SoftBank’s €45B commitment could make France one of Europe’s most important AI compute hubs.
- The project ties AI competitiveness to physical infrastructure, especially access to large-scale data center capacity.
- The 2028 launch target will be an early test of whether Europe can execute fast enough on AI infrastructure.
