SoftBank’s €75 Billion French Data Center Bet Signals Europe’s Entry Into Global Compute Arms Race

SoftBank announced a €75 billion ($87 billion) commitment to develop 5 gigawatts of AI data center capacity in France, the largest single foreign infrastructure investment in European history and a signal that compute sovereignty has become a strategic priority worth hyperscale capital deployment. The May 30 announcement, timed within 48 hours of Alphabet’s $80 billion equity raise for AI infrastructure, marks an inflection point: Europe is no longer a secondary market for compute, but a contested battleground where sovereign capacity commands US-level investment.

The Power Grid Calculus

SoftBank’s decision hinges on a structural advantage: France draws roughly 70% of its power from nuclear reactors, making it the world’s largest net electricity exporter with industrial power prices less than half of Britain’s, per Tom’s Hardware. European industry broadly pays double the electricity rate of US counterparts, but France’s nuclear baseload creates an anomaly that CEO Masayoshi Son cited as “absolutely decisive” in selecting the location.

The first phase commits €45 billion to deliver 3.1 GW of capacity in the Hauts-de-France region by 2031, according to SoftBank’s official announcement. The project will convert EDF’s Bouchain thermal plant and construct greenfield facilities in partnership with Schneider Electric and French grid operator EDF. Yet France’s transmission operator RTE assumes only 35% of the 14 GW currently reserved for Data Centers will actually connect due to grid constraints—a disconnect between political ambition and physical infrastructure that will test whether SoftBank’s timeline survives contact with reality.

Compute as Strategic Asset

The timing is not coincidental. Five US cloud and AI infrastructure providers—Microsoft, Alphabet, Amazon, Meta, and Oracle—collectively committed $660-690 billion in capex for 2026, nearly doubling 2025 levels. Alphabet revised its 2026 capital expenditure forecast upward to $180-190 billion in April, then days ago announced an $80 billion equity raise including $10 billion from Berkshire Hathaway to fund the expansion. Google Cloud revenue surged 63% year-over-year to $20 billion in Q1 2026, with backlog nearly doubling to over $460 billion.

SoftBank’s shares jumped 14% on May 31 and have risen more than 70% in 2026, according to CNBC, reflecting investor recognition that infrastructure—not models—is the binding constraint in AI scaling. The capital allocation shift is structural: more than 60% of the big four hyperscalers’ $320+ billion collective 2026 AI spending goes to power infrastructure, cooling, and construction rather than compute hardware itself.

The Sovereignty Premium

The United States produced more than 50% of all significant AI models in 2024; Europe produced just 6%. American and Chinese companies control more than 90% of the world’s AI-specialised data centers, per DeepTech.Build. President Emmanuel Macron has championed sovereign AI infrastructure as a matter of technological independence, backing local players like Mistral AI while pressing for European compute capacity that keeps data and processing under EU jurisdiction.

The EU Council approved €37 billion in gigafactory investments across ten facilities in seven countries in January 2026, establishing a regulatory framework for compute sovereignty. SoftBank’s commitment exceeds that entire pan-European programme by more than 2x, concentrated in a single country. The scale signals that private capital views European AI infrastructure as commercially viable, not merely a political subsidy play.

France’s Minister of Economy Roland Lescure called the investment “testament to President Macron’s ambition to position France as a leading destination all along the AI value chain,” framing it explicitly as geopolitical positioning rather than pure cost arbitrage.

Competitive Response Pressure

SoftBank’s move creates immediate pressure on US hyperscalers. Microsoft, Amazon, and Google have each announced European data center expansions over the past 18 months, but none at this scale or with this level of sovereign partnership. If France successfully delivers grid capacity and regulatory streamlining, it becomes a proof point that could pull hyperscaler capex away from US domestic builds.

Amazon CEO Andy Jassy said in the company’s Q1 shareholder letter: “We’re not going to be conservative in how we play this—we’re investing to be the meaningful leader, and our future business, operating income, and FCF will be much larger because of it.” That stance implies willingness to match competitive infrastructure builds wherever compute demand concentrates, including Europe if sovereign customers prefer local capacity.

The semiconductor supply chain adds friction. TSMC, Samsung, and Intel are all capacity-constrained on advanced nodes needed for AI accelerators, and lead times for power infrastructure components (transformers, switchgear, cooling systems) have stretched beyond 18 months in some cases. SoftBank’s 2031 timeline for initial capacity reflects these bottlenecks—this is not rapid deployment, but a five-year construction programme that assumes supply chains normalise.

What to Watch

RTE’s grid connection timeline will determine whether SoftBank’s 2031 target is realistic or requires staged rollout beyond the decade. If France cannot deliver promised transmission capacity, the investment case weakens and other European markets (Spain, Poland, Nordic countries with surplus hydro) become attractive alternatives. Watch for Microsoft, Amazon, and Google to announce European capacity expansions in Q3-Q4 2026—competitive dynamics suggest they cannot cede this market without response.

Semiconductor supply remains the wildcard. If TSMC’s Arizona and Germany fabs ramp as planned, accelerator availability improves by 2028-2029. If geopolitical disruptions (Taiwan, trade restrictions) constrain supply, even completed data centers may sit partially idle waiting for chips. Energy costs matter more than capex efficiency now; any material increase in French industrial power rates would erode SoftBank’s structural advantage and slow follow-on investment.

The political sustainability of concentrating this much power demand in a single region is untested. Local opposition to data center energy consumption has already emerged in Ireland, Netherlands, and parts of Germany. If Hauts-de-France experiences grid strain or local energy price increases, the project becomes a political liability rather than an economic asset. SoftBank is betting that France’s nuclear surplus can absorb 5 GW of new demand without triggering backlash—a hypothesis that will be tested as construction begins.