SoftBank announced it will invest up to €75 billion in AI data centers across France, with the first €45 billion committed to three sites in the country's northern Hauts-de-France region by 2031. The deal was struck through personal diplomacy between SoftBank founder Masayoshi Son and French President Emmanuel Macron, who met during Macron's visit to Japan earlier this year.
The scale is easier to understand with one comparison. France had roughly 2 gigawatts of installed data center capacity at the start of 2025. SoftBank's plan adds up to 5 gigawatts of new capacity on top of that, more than tripling the country's current total.
French industrial group Schneider Electric will partner on designing and supplying equipment, including a manufacturing facility at the Port of Dunkirk. State-owned energy company EDF is providing the site of a former power plant in Bouchain for redevelopment. These are not passive spectators. They are the supply chain behind the build.
The reason France attracted this investment over other European locations is simple: electricity. France generates around 70% of its power from nuclear plants, making it one of the cheapest and most stable electricity markets in Europe. EDF already offers long-term pricing contracts tied to nuclear production costs, which gives large industrial customers predictability that gas-dependent markets simply cannot match. For data centers, where electricity can account for 30 to 50% of total operating costs, that gap is decisive.
This matters beyond France's borders. High electricity prices in Germany and the UK have already caused major AI companies to pause or redirect infrastructure plans. France, with its nuclear base and long-term pricing stability, sits in a different position. France's electricity prices have largely decoupled from neighboring countries, with particularly wide gaps compared to Germany and Italy, according to the country's national grid operator.
For business operators across Europe, this shift has a practical implication. Where AI infrastructure gets built will increasingly determine where AI services are cheapest, fastest, and most data-secure to run. Companies that process sensitive data under EU rules, or that care about the carbon footprint of their AI usage, will find France-hosted services a more attractive option as this capacity comes online.
The broader question is whether SoftBank can actually fund all of this. SoftBank's total debt stands at roughly $135 billion. S&P Global Ratings downgraded SoftBank's credit outlook from stable to negative in March 2026, citing concerns about liquidity and asset credit quality. The company also recently had to scale back a planned $10 billion loan backed by its OpenAI stake to around $6 billion after lenders pushed back on the difficulty of valuing a private company at its current price. The France commitment sits on top of a $40 billion bridge loan taken out earlier this year and SoftBank's involvement in the Stargate data center initiative in the United States.
The ambition is real and the rationale is sound. France's nuclear grid is a genuine structural advantage, and the pipeline of projects already in planning shows demand is there. But SoftBank is building its financing on a tower of assumptions: that OpenAI's paper valuation holds, that lender appetite remains strong, and that the returns from all this infrastructure materialize before the debt comes due. Whether €45 billion in concrete gets poured by 2031 depends on whether that tower stays standing.