SAP CEO Christian Klein believes Europe should avoid replicating the U.S.’s large-scale AI infrastructure expansion. In his view, a European version of the massive “Stargate” initiative wouldn't offer as much value as focusing on the practical deployment of AI in key industries.
Klein shared his thoughts during a recent media interview, responding to European proposals to build five AI “gigafactories.” These large data centers, which many cities are now competing to host, aim to bolster the continent's AI capabilities. But Klein remains unconvinced that this kind of buildout is necessary.
The Commoditization of AI Models
Klein argues that large language models (LLMs), despite their high training demands, are rapidly becoming commoditized. He cites the surprising success of Chinese company DeepSeek, which launched a model that rivals offerings from OpenAI and Google — and did so without massive infrastructure.
Instead of matching U.S. spending, Klein suggests that Europe should focus on industry-specific AI applications, particularly in sectors like automotive and chemicals. According to him, targeted innovation would deliver far more value than building an expensive AI superstructure.
Fragmented Progress Already Underway
This approach is already taking root across Europe, though in a fragmented way. For example:
Juvoly is advancing AI speech recognition in Dutch.
Silo AI, a Finnish company recently acquired by AMD, is building multilingual LLMs for the Nordic region.
Cradle, a startup in the biotech space, is developing tools to make biology programmable through AI.
However, in the automotive sector — a priority area for Klein — Europe appears to be falling behind. Klein may well prefer to see the EU’s €20 billion AI infrastructure fund redirected toward innovations in battery technology or autonomous vehicles, though these are also areas where Europe is struggling to keep pace.
The Investment Gap Is Stark
The contrast in ambition between Europe and the U.S. is hard to ignore. American tech giants have laid out plans to spend up to $500 billion on the “Stargate” project alone. Meanwhile, Europe’s commitment stands at just €20 billion for its proposed five AI centers — a 20-fold difference.
This gap was highlighted again during Nvidia CEO Jensen Huang’s recent visit to Europe. Huang warned that Europe’s AI progress is being throttled by limited computing power and announced plans to expand Nvidia’s hardware footprint across the continent. Naturally, Nvidia stands to benefit from such expansion — the demand for GPUs, the "pickaxes" of the AI gold rush, drives their value sky-high.
The Case for Strategic Autonomy
Even so, should Europe ignore Huang’s warnings? Not entirely. At the very least, it would be prudent for Europe to ensure it can meet rising demand for AI applications domestically — even if that means relying on U.S.-based cloud services and chips.
Klein’s viewpoint aligns with the broader debate over Europe’s digital sovereignty. Earlier this year, he called plans to build European cloud infrastructure to compete with U.S. hyperscalers “nonsensical,” citing high energy costs that make such ventures economically unattractive. That challenge still looms, even as Europe aspires to take a leading role in AI.
In short, Klein’s argument is not against AI — it’s about being smart with how Europe invests in it. Rather than duplicating U.S.-style infrastructure, he believes the continent should play to its strengths, channeling funds into real-world applications that offer immediate and strategic value.
