Europe’s AI Infrastructure Race Turns Transnational

“National sovereignty and European competitiveness are not the same thing.”

Three material developments today: French AI infrastructure expands into Sweden and Denmark, Milan gains a fully operational sovereign-cloud region, and Italy's Gigafactory coalition considers a broader industrial and cross-border architecture.

The underlying shift is consequential. Europe's AI competition is no longer simply about which country secures a Gigafactory. It is increasingly about who controls the computing platform, who attracts the physical investment, and who captures the economic returns.

On October 13, French cloud provider Scaleway announced two significant AI infrastructure expansions.

Sweden: Scaleway will deploy an AI supercomputer in Stockholm powered by AMD Instinct MI355X accelerators. Bull will integrate and maintain the system, supported by Giga Computing, HPE, VAST Data and Motul.

Denmark: Scaleway announced a new Copenhagen GPU Zone powered by NVIDIA HGX B300 systems, with several thousand GPUs planned.

Across Paris, Stockholm and Copenhagen, Scaleway's longer-term infrastructure ambition extends to tens of thousands of GPUs, tens of exaflops of computing power and tens of megawatts of capacity.

These are announced deployments, not disclosed financial closings. Individual site investment amounts and binding power commitments have not been published.

Why this matters. France is beginning to separate ownership of the AI platform from the physical location of computing infrastructure. French companies can retain control of cloud services, customer relationships and systems integration while deploying computing capacity in Nordic locations offering favorable electricity and cooling conditions.

This could produce a new European model: FRENCH CLOUD CONTROL → NORDIC POWER → GLOBAL SEMICONDUCTORS → EUROPEAN ENTERPRISE DEMAND.

The semiconductor dimension is especially important. Scaleway's decision to operate both AMD and NVIDIA architectures potentially reduces dependence on a single supplier and allows customers to optimize workloads across different hardware and software ecosystems. AMD's ROCm ecosystem also provides an alternative to NVIDIA's proprietary CUDA software environment.

Second-order effect: Europe could develop a more competitive market for accelerated computing even without producing its own frontier GPUs. For investors, the opportunity extends beyond data centers to systems integration, networking, cooling, storage and GPU utilization. For policymakers, the strategic objective should be European control over computing services and industrial applications—not necessarily European ownership of every semiconductor.

A second announcement from Scaleway is particularly consequential for Italy. Milan now has a complete three-Availability-Zone cloud region. The company announced two additional Milan zones, bringing its Italian infrastructure to three geographically separate, independent operating locations.

Scaleway also confirmed plans for three Availability Zones in Stockholm and three additional zones in Frankfurt in 2027. The company highlighted the industrial significance of its Qarnot acquisition and the existing cooperation between Qarnot and A2A in Brescia, where computing infrastructure can supply recoverable heat to the district-heating network.

The Italian opportunity. This is an important development for Industrial AI. Italy's Gigafactory strategy has concentrated heavily on assembling large industrial partners. Scaleway demonstrates a complementary approach: build resilient computing infrastructure close to industrial customers and connect computing activity to the broader energy system.

The potential Italian economic model becomes: COMPUTE → INDUSTRIAL APPLICATIONS → PRODUCTIVITY → RECURRING REVENUE. With an additional opportunity: COMPUTE → WASTE-HEAT RECOVERY → ENERGY EFFICIENCY.

The distinction is significant. A Gigafactory creates computing capacity. An integrated Industrial AI platform creates the possibility of monetizing that capacity across manufacturing, engineering, energy, mobility and infrastructure.

For CEOs and boards: the priority should be industrial workloads with measurable operating improvements and revenue potential. For investors: evaluate the relationship between computing utilization, energy costs, customer retention and returns on invested capital. For EU policymakers: energy efficiency and compute infrastructure should increasingly be evaluated together.

The new Milan cloud capacity is operational progress, but it should not be confused with approval or construction of Italy's proposed EuroHPC Gigafactory.

Italian reporting on October 13 provides additional indications of the proposed national Gigafactory architecture. Palazzo Chigi is considering a consortium built around Eni, TIM and Leonardo, potentially expanding to Fincantieri, CDP, Enel, Saipem, Snam and FiberCop.

The reported initial financing structure envisages the Italian government at €200 million, the European Union at €200 million and private investors at €800 million, for an initial phase of €1.2 billion. The longer-term investment could exceed €5 billion. The report also describes discussions about possible Greek participation representing approximately 25% of consortium computing capacity.

These are proposals and reported discussions, not newly confirmed financing commitments or a finalized Greek agreement. The initial €1.2 billion framework had already been reported earlier in October; today's incremental information concerns the potential industrial and cross-border configuration.

Why this matters. The possible involvement of Fincantieri would strengthen the industrial-demand dimension, adding shipbuilding, complex systems engineering and defense-related applications. Greek participation could broaden the prospective customer and institutional base.

However, Italy's central financing challenge remains unresolved. A broad consortium does not automatically produce a bankable infrastructure asset. The decisive next disclosures must concern committed equity, secured power, final sites, binding customer demand and governance. The reported financing framework is potentially credible, but its effectiveness will depend on converting corporate interest into legally enforceable commitments.

Today's developments point toward a more differentiated competitive structure.

Five Key Takeaways

what to carry forward

• FRANCE IS EXTENDING ITS AI REACH. Scaleway's Nordic expansion strengthens the ability of French-controlled cloud infrastructure to serve European customers beyond France.
• THE NORDICS ARE MOVING UP THE VALUE CHAIN. They are attracting not only data-center investment, but increasingly sophisticated AI computing platforms.
• ITALY HAS AN INDUSTRIAL AI OPPORTUNITY. Milan's completed cloud region improves the enabling infrastructure. The national Gigafactory proposal still needs binding financing and execution commitments.
• HARDWARE COMPETITION IS BECOMING STRATEGIC. AMD's growing role offers European computing operators greater architectural choice, potentially improving price competition and bargaining power.
• CAPITAL FORMATION MUST BECOME TRANSNATIONAL. Europe's strongest AI investment opportunities may combine capital, computing, energy and demand across several member states rather than within one national consortium.

Europe's AI Gigafactory race is beginning to expose a fundamental distinction. National sovereignty and European competitiveness are not the same thing. A country can host substantial computing infrastructure without capturing much of its economic value. Conversely, a European company can control a computing platform while distributing its physical infrastructure across several countries.

France's latest expansion demonstrates the second model. Italy's proposed Gigafactory still needs to demonstrate how its industrial customer base will translate into sustainable demand. The Nordics are demonstrating why electricity and physical infrastructure remain powerful determinants of capital allocation.

The next competitive frontier is not the Gigafactory as an isolated national asset. It is the integrated European AI production system.

Europe's long-term AI advantage will depend less on where computing capacity is physically installed than on who can integrate that capacity with industrial demand and convert it into sustainable economic returns.