21 November 2026

Halting new data centers won’t stop AI’s gains from concentrating in a few firms, so Congress should focus instead on spreading AI adoption across the economy.

National Interest | Bruno S. Sergi

In the wake of the 2026 US midterm elections, public opposition and potential regulatory moratoria threaten 45 percent of planned American data center capacity expansion, driving cloud infrastructure investment toward Southeast Asia. Google committed $15 billion to Malaysian data centers while Microsoft expanded enterprise cloud capacity across Malaysia and Indonesia.

Local restrictions on computing infrastructure fail to curb market concentration among frontier model developers. Server relocation does not reduce market power. A March 2026 Gallup survey revealed 71 percent of Americans oppose local AI data center construction, exceeding opposition to nuclear power plants. Economic gains depend on tool adoption across broader domestic sectors rather than server location. US Census Bureau data from November 2025 to January 2026 shows only 18 percent of domestic business firms use AI in any capacity. Furthermore, 57 percent of those adopting firms utilize the technology across three or fewer internal business functions.

Comment

Offshoring high-density AI compute nodes to Southeast Asia exposes critical digital infrastructure to foreign jurisdiction risks, mirroring legacy supply chain dependencies created under the CHIPS and Science Act of 2022. While enterprise hyperscalers maintain model ownership, physical server distribution in Malaysia or Indonesia places data transit routes under foreign host-nation regulatory frameworks. This geographic decoupling separates algorithm development from physical processing assets.

As a second-order consequence, sovereign compute degradation directly impacts defense industrial base capability, where secure processing relies on domestic compute clusters. Federal procurement protocols, such as FedRAMP High baselines, strictly limit secure workloads to domestic soil. Shifting commercial AI infrastructure abroad reduces the domestic capacity available for Defense Logistics Agency supply chain modeling during wartime mobilisation.

Strategic Question for Discussion
If commercial hyperscalers shift planned domestic compute capacity to Southeast Asia, how will federal programs requiring FedRAMP High baselines maintain resilient access to dual-use model training facilities?
The trajectory indicates that federal defense agencies will face higher procurement premiums to isolate dedicated domestic compute enclaves as commercial infrastructure migrates abroad. Defense Logistics Agency modeling and classified dual-use applications cannot legally leverage foreign host-nation servers without compromising security compliance. Consequently, public sector requirements will increasingly rely on subsidized, defense-only data centers to preserve sovereign processing capacity.
Share your assessment in the comments below.
πŸ’¬
Ask me
Ask Strategic Study India New chat ×