To avoid being cut off from frontier AI, non-US countries can offer US hyperscalers incentives like subsidized energy for building data centers locally. In return, they can demand contractual guarantees for frontier model access, creating leverage against future US government-imposed restrictions.

The European Commission is linking AI infrastructure directly to energy systems and national sovereignty. This approach merges AI regulation, data center planning, and grid sustainability into a single strategic problem, viewing compute capacity as a critical national resource intertwined with energy policy.

Mistral's $1.4B investment in Swedish AI infrastructure is more than an expansion; it's a political move. By building a "fully European AI stack," Mistral is positioning itself as the regional alternative to US tech giants, capitalizing on growing desires for data sovereignty amid fraying political ties.

A key strategy for middle powers is to offer fast, efficient data center construction to leading US AI labs. In return for alleviating the labs' 'inference crunch', these nations can negotiate guaranteed access to new frontier models at the same level as the US commercial market, ensuring they aren't left behind.

The global supply chain for cutting-edge AI chips is a major chokepoint, ideal for governance. Three companies design them, one (TSMC) manufactures over 90%, and one Dutch firm (ASML) makes the essential machinery. This concentration makes tracking and controlling compute resources feasible for a global coalition.

For Europe to compete in AI, it must overcome its aversion to large-scale energy projects. The winning strategy is to co-locate massive compute infrastructure in areas with cheap, abundant energy, like Norwegian wind farms. Without this, Europe risks becoming a 'tourist economy' built on past glories.

Reid Hoffman advises Europe against trying to replicate US hyperscalers. Instead, governments should offer streamlined access to energy and data center permits to US tech giants in exchange for compute resources, enabling European companies to build competitive AI applications.

The abstract race for AI superiority is now grounded in physical reality. Control over electricity grids, cooling, and land for data centers has become as strategically important as semiconductor supply chains, shaping who can scale frontier AI.

The prospect of moving AI compute to space creates a hard deadline for the geopolitical strategies of middle powers. Strategies like Europe's 'compute for access' rely on leverage from controlling terrestrial data center construction. Once compute moves to space, this leverage disappears, forcing these countries to develop new long-term plans.