SoftBank will invest up to €75 billion to build Europe’s largest AI hub in France

SoftBank has fully entered the global race for Artificial Intelligence infrastructure and picked France as the stage for one of the most ambitious projects on the planet. The Japanese group, led by Masayoshi Son, has committed to invest up to €75 billion to build a massive network of AI computing clusters that is set to become the largest data center complex of its kind in all of Europe.

This project is not just another tech announcement; it is being treated as a strategic move at a time when Europe is trying to catch up with the United States, China and even some Middle Eastern countries in the race for compute capacity. While AI model development gets all the spotlight, people behind the scenes know that without heavy infrastructure – energy, data centers, networks and chips – none of these systems can truly scale.

The move also delivers a major political win for French president Emmanuel Macron. The package was lined up right before Choose France, the annual event where the government brings together global executives and investors to attract large industrial and tech projects to the country.

How the deal was born: Macron, Son and the AI bet

According to people close to the talks, the agreement between SoftBank and the French government gained momentum after a dinner in Tokyo in early April between Emmanuel Macron and Masayoshi Son. In that meeting, Macron highlighted three key points to convince SoftBank:

• France’s strong use of nuclear energy in its power mix, with low carbon emissions;
• a fast-track permitting process for AI facilities and data centers;
• France’s clear ambition to become Europe’s main AI infrastructure hub.

The pitch worked. Son publicly stated that SoftBank is proud to make a commitment of this size in France and stressed that the country offers a rare combination of:

• solid industrial capacity;
• highly qualified technical talent;
• a long-term national vision for AI and digital infrastructure.

In practice, this means the French project moves to the forefront of SoftBank’s strategy to position itself at the heart of the AI revolution, competing at scale with enormous projects on other continents.

The size of the project: 5 GW of capacity by 2031

One of the most impressive aspects of this plan is the scale of the infrastructure. The initial commitment calls for SoftBank to lead a €45 billion investment to build 3.1 gigawatts (GW) of AI computing capacity in the Hauts-de-France region, in northern France, by 2031. After that, an expansion of another 2 GW is planned, reaching a total of 5 GW.

To get a sense of the scale:

• 5 GW is roughly equivalent to the output of around five medium-size nuclear power plants;
• it is similar to the peak power demand of New York City at certain times.

If the complex is completed at this scale, SoftBank’s total investment in France is expected to reach €75 billion, roughly US$ 87 billion according to the group’s estimates.

Dunkirk as an AI and robotics hub

One of the highlights of the plan is a large facility in Dunkirk, a city in a strategic position in northern France. There, SoftBank will partner with Schneider Electric to build a hub for:

• large-scale AI infrastructure;
• robotics solutions manufacturing and related technologies.

This location is seen as ideal for serving clients in London, Brussels and Amsterdam, thanks to geographic proximity and network connections. The idea is to cut latency and offer a fast access route for major financial and corporate centers across Western Europe.

Why Europe is racing to catch up in AI infrastructure

Despite having a strong scientific base in many countries, Europe has fallen behind in building mega data centers and AI-focused clusters when compared to:

• the United States, driven by massive investments from big tech firms;
• China, which pours resources into strategic infrastructure;
• and regions in the Middle East, such as the United Arab Emirates, which invest heavily using cheap energy and abundant capital.

Many large data center investments have flowed into areas with:

• cheaper and more stable energy;
• quick access to the power grid and fewer basic infrastructure bottlenecks;
• more flexible regulation on issues like urban planning, data use and emissions.

SoftBank’s move is an attempt to shift this balance a bit. By anchoring a gigantic project in France, the group helps pull Europe to a new level of AI-focused compute capacity, with cutting-edge data centers operating under European privacy and data protection rules.

SoftBank and its global AI strategy

The mega project in France is not a standalone initiative. It is part of a broader SoftBank strategy to become a central player in the world’s Artificial Intelligence infrastructure.

Some of the most relevant recent moves include:

• a 10 GW data center project in Ohio, in the United States, announced in March by authorities connected to the US government;
• more than US$ 60 billion already committed to investments in OpenAI, the maker of ChatGPT;
• plans to take its robotics and energy businesses public in the US;
• strengthening semiconductor capabilities around Arm, the UK-based chip design firm considered the group’s crown jewel.

At first, Son outlined the idea of a consortium called Stargate, with around US$ 500 billion in large-scale dedicated compute infrastructure, mainly for OpenAI’s own use. Over time, that vision has evolved into multiple data center projects around the world, and France is now emerging as one of the most visible pillars of this shift.

Cost and funding model

Market analysts estimate that each 1 GW of AI infrastructure can cost around US$ 50 billion, including:

• land acquisition;
• civil works and construction;
• power connection and supply;
• IT equipment such as GPUs, servers and networks;
• cooling and security systems.

With a total plan of 5 GW in France, it is clear that SoftBank will not foot the entire bill alone. The typical model for this kind of project involves:

• a smaller slice of equity capital put in by SoftBank and close partners;
• most of the amount coming from project finance, with debt financing tied to the assets being built.

So far, full details on financial partners have not been disclosed, and the final hardware and systems suppliers for the French data centers have also not been officially named.

Abu Dhabi, US, France: a global AI network

Beyond France and the United States, SoftBank is part of a consortium planning to deploy 5 GW of AI infrastructure in Abu Dhabi. This group includes heavyweight names such as:

• G42;
• OpenAI;
• Oracle;
• Nvidia;
• Cisco.

This setup shows that SoftBank is not just betting on a single project in France, but rather building a kind of global mesh of AI mega data centers, connecting regions with high demand and different energy and regulatory profiles.

Not every announcement becomes reality: the risk of mega projects

Even with hundreds of billions being announced worldwide for AI capacity expansion, not every project comes to life. A frequently cited example is OpenAI’s own plan to build a major facility in northeastern England, celebrated by the UK government in September. That project has since been put on hold indefinitely.

This underscores an important point: announcing big numbers is one thing; executing, building and operating in practice is another. Factors such as:

• environmental permits;
• local pressure over water and energy use;
• cost overruns versus initial budgets;
• regulatory changes;
• and even the evolution of AI chips and models themselves

can reshape the viability of a large project over time. In France’s case, political alignment and Macron’s decision to prioritize the topic increase the odds of execution, but the risks are not completely off the table.

Macron, elections and the Choose France showcase

Since taking office in 2017, Emmanuel Macron has built up an image as a pro-business, pro-technology leader, trying to position France as a destination for large industrial and digital investments. The Choose France event, held annually at the Palace of Versailles, has become the main showcase for this agenda.

Tools we use daily

Translation

Text Inspection & Clipping

Social Media

Search & Research

Recruitment

Programming

Productivity & Organization

Operations

Marketing & Sales

Finance & Investment

Education

Design & Media

Data & Analytics

Content & Writing

Automation

In recent editions, the event has leaned even more heavily into AI, data centers, chips and cloud technologies. The message is clear: France wants to be seen as:

• a country with low-carbon electricity, thanks to extensive use of nuclear power;
• a regulatory environment willing to fast-track critical infrastructure projects for AI;
• a stable base for global companies that want to operate under European rules.

This new SoftBank announcement lands at a politically sensitive moment. There is less than a year to go before the next French presidential election, with an uncertain outlook and the far right leading some polls. Macron cannot run again, but he wants to leave a strong mark on the country’s economy and industrial policy, and the AI bet is a direct part of that legacy.

Energy, carbon footprint and the sustainability debate

A central piece of this story is the link between large-scale AI and energy consumption. Data centers for training and running advanced models need huge amounts of electricity and cooling, which has put the industry under pressure to respond on its environmental impact.

In France’s case, the competitive edge lies in an electricity mix with a heavy share of nuclear power, which is traditionally associated with low operational carbon emissions. This allows the government and SoftBank to pitch the project as part of a more climate-aligned AI infrastructure, especially when compared with facilities that depend directly on fossil fuels.

At the same time, the debate is far from settled. Many argue that, in the long run, the ideal setup would include a stronger mix of renewables such as solar and wind, along with hardware and software efficiency improvements to reduce consumption per unit of delivered compute. If the project in France moves forward as planned, it is likely to become one of the main case studies in the ongoing discussion about how to balance:

• AI capacity growth;
• energy demand;
• and environmental responsibility.

What this means for Europe’s AI ecosystem

If SoftBank’s French plan materializes, the impact on Europe’s AI ecosystem could be significant. Some likely consequences include:

• More top-tier compute capacity inside the European Union, allowing banks, hospitals, governments and major industries to run advanced models without moving sensitive data outside the bloc;
• a new layer of competitiveness for AI startups, which will be able to access high-performance clusters without relying solely on infrastructure on other continents;
• stronger bargaining power for Europe in negotiations involving chips, frontier models and global cloud platforms.

For companies of any size working with AI, the presence of a mega hub like this in France could change system architecture decisions, especially when it comes to:

• where to store and process large volumes of data;
• how to deal with latency between end users and servers;
• which cloud regions to use in order to comply with local rules.

In the end, SoftBank’s announcement is not just about a massive check. It is about how physical and energy infrastructure is becoming a central part of the conversation when we talk about Artificial Intelligence at global scale – and now, about France’s push to claim a leading role in that game.