NATIONS ARE BUILDING AI

Governments have decided that advanced artificial intelligence is too important to be left entirely to commercial markets. The result is a wave of national strategies that treat AI capability as critical infrastructure comparable to semiconductors, energy, or advanced manufacturing.

These strategies combine public investment, regulatory direction, and restrictions on technology flows. They are already influencing where models are trained, what data is available, and which companies can participate at the frontier — context that sits underneath everything in our AI coverage.

How is the geography of development changing?

The United States has used export controls on advanced chips — administered through the Commerce Department's Bureau of Industry and Security — to limit the scale at which certain countries and entities can train frontier models. At the same time, it has supported domestic semiconductor manufacturing through the CHIPS and Science Act and backed large-scale compute initiatives. We covered the mechanics of the restrictions in chip controls are reshaping AI.

China has responded with its own substantial investment in domestic chip design and fabrication, alongside policies to increase the volume and quality of training data available to its developers. Progress is real but remains constrained relative to unrestricted access to the highest-performance hardware.

Other countries and regions are pursuing their own variants. The EU has paired the AI Act's regulatory framework with public compute programs. Some states are aligning with one of the two major blocs; others are attempting to maintain access to both while building limited domestic capacity. The result is a more fragmented landscape than existed even three years ago.

What does state involvement do to private development?

Commercial AI companies now operate in an environment where government policy directly affects their access to compute, data, and talent. Decisions about where to locate research teams, where to train models, and which markets to prioritize are no longer driven solely by cost or talent availability.

Alignment with national priorities can unlock funding, data access, and regulatory support. Divergence can trigger restrictions or loss of market access. This changes the risk and return profile of different AI business models depending on their geographic footprint and technology dependencies.

The effect is most visible at the frontier, where the largest models require resources that are now subject to explicit state control in multiple jurisdictions — the same choke points that define the AI chip supply chain. For narrower applications, the impact is smaller but still growing as governments extend their focus beyond the absolute cutting edge.

How should investors read the fragmentation?

For investors, the rise of sovereign AI strategies creates both new risks and new opportunities for capital allocation. Companies whose technology or data advantages depend on cross-border flows face higher political risk. Companies that can operate effectively within a single bloc, or that control resources valued by multiple blocs, may benefit.

Infrastructure that supports domestic or allied compute capacity has clearer policy support. Research and application development that aligns with stated national priorities in areas such as defense, healthcare, or critical infrastructure can attract non-dilutive capital or preferred regulatory treatment.

The long-term question is whether the fragmentation increases or decreases the overall rate of progress. Parallel development tracks can produce redundancy and faster diffusion within blocs, but they also reduce the benefits of global collaboration and specialization.

The Bottom Line

AI development is no longer a purely commercial domain. States have inserted themselves as both funders and gatekeepers, and the effects are visible in hardware access, data availability, and the geography of talent and investment.

Builders and operators need to treat geopolitical alignment and domestic capability as first-order variables in their strategy. Investors need updated frameworks for assessing which AI businesses are structurally advantaged or disadvantaged by the new policy environment.

The companies and jurisdictions that can reliably secure the inputs required for advanced AI under these constraints will capture a larger share of the value than those that cannot.