Europe’s AI Capital Race Enters the Credit Divide

Three consequential developments since October 18 reveal an accelerating divergence in Europe's AI infrastructure race. Germany is mobilizing institutional capital for energy and digital infrastructure, Siemens is advancing its Industrial AI strategy, and Spain is confronting the financial risks of concentrated AI-compute customers.

The Big Take: Europe's AI Supercycle is entering a more selective phase of capital formation. Investors are beginning to distinguish between industrial infrastructure supported by productive demand and highly leveraged computing businesses whose expansion depends on continued access to credit.

The fundamental question is shifting: Which European AI ecosystems can convert infrastructure spending into recurring industrial cash flow without accumulating disproportionate financial risk?

On October 19, Spanish financial newspaper Cinco Días reported that CoreWeave accounts for approximately 90% of Merlin Properties' current data-center rental revenues.

The exposure represents approximately 8.6% of Merlin's total rental income.

CoreWeave's deteriorating credit-market indicators have raised concerns about the sustainability of heavily leveraged AI infrastructure expansion. However, market-implied default probabilities are estimates of risk, not evidence that a default has occurred.

Merlin has defended its position, emphasizing contractual protections, the high utilization of CoreWeave's leased capacity and its intention to diversify future customers.

Its development programme targets approximately 724 MW of IT capacity by 2032, although not all phases are fully financed.

The hidden pattern. The financial risk of the AI Supercycle is beginning to migrate from computing operators toward their infrastructure landlords.

On October 19, Reuters reported two significant developments.

First, the Made for Germany initiative, representing 147 companies and investors, identified up to €4.1 trillion of potential investment by 2035, across sectors including artificial intelligence, energy, infrastructure and defense. Its members reported plans to invest more than €850 billion through 2028.

Second, Brookfield announced its intention to invest tens of billions of euros in Germany over five years, identifying energy and digital infrastructure as strategic priorities.

These figures cover multiple sectors. They are not new financing commitments to Germany's proposed EuroHPC Gigafactory, and the €4.1 trillion estimate is potential investment rather than contracted capital.

Why this matters. Germany is assembling potentially complementary components of an Industrial AI investment system.

On October 19, Siemens executive board member Cedrik Neike outlined an ambitious Industrial AI strategy at the company's Erlangen electronics plant.

Siemens aims to raise factory automation from approximately 30–40% today to 70–80%, using AI, digital twins and advanced manufacturing technologies. The company intends to invest €500 million in expanding the Erlangen facility by 2030.

Neike argued that manufacturers in high-cost economies need annual productivity improvements of approximately 7–9% to remain competitive.

These automation and productivity figures are targets and management assessments, not achieved results.

The second-order effect. This development provides a more concrete economic rationale for Industrial AI computing investment. Gigafactory infrastructure becomes strategically valuable when industrial customers can demonstrate that additional computing capacity improves operational performance.

No new EuroHPC Gigafactory tender award, finalized site selection or binding Gigafactory-specific financing agreement was verified in today's review for France, Italy or the Nordics.

Five Key Takeaways

The selective phase of Europe's AI capital formation

• AI credit risk is becoming a European infrastructure issue. Spain's Merlin–CoreWeave exposure illustrates how financial stress among computing customers can affect infrastructure valuations.
• Germany is building a broader capital-formation ecosystem. Brookfield's investment intentions strengthen the potential connection between energy infrastructure, institutional finance and digital investment.
• Industrial AI provides an alternative growth engine. Siemens demonstrates how computing demand can be connected to operational productivity rather than speculative capacity expansion.
• The next market rotation may favor productive infrastructure. Grid assets, electrical equipment, automation and industrial software could prove more resilient than highly leveraged computing operators, although valuations and project economics remain decisive.
• European policymakers must improve capital productivity. Public funding should support commercially executable projects with transparent financing, credible electricity delivery and measurable economic benefits.

Germany is seeking to mobilize capital around industrial infrastructure and productive applications. Spain is confronting the financial risks created when infrastructure expansion depends heavily on a small number of leveraged computing customers.

The distinction is becoming increasingly consequential. The Great Split in European AI will not necessarily occur between countries with and without Gigafactories. It may emerge between ecosystems that generate productive industrial cash flow and those that accumulate capital-intensive computing capacity faster than they monetize it.

For investors and corporate boards, that distinction should increasingly govern capital allocation.

Europe's next AI winners will be determined not by how much capital they attract, but by how effectively they convert that capital into industrial productivity, sustainable cash flow and durable competitive advantage.