Morgan Stanley warns $1.6 trillion AI infrastructure investment faces multiple risks: US still short 32 gigawatts of power by 2028.

Beating AI News Flash: Morgan Stanley has summarized the core risks facing U.S. AI infrastructure development as "people, power, and politics," believing that local governments tightening approvals, supply-demand reversal leading to simultaneous price and volume declines, and power shortages could collectively impact up to $1.6 trillion in AI infrastructure investment plans.

On the political front, data center site selection, grid connection applications, electricity pricing mechanisms, and tax incentives are mainly influenced by state and local governments. As U.S. midterm elections approach, restrictions on data center construction could become an important factor affecting AI industry sentiment. The report notes that eliminating related tax incentives has become a bipartisan issue, and regulatory decisions by public utility commissions could also have long-term impacts on project implementation.

On the supply-demand side, Morgan Stanley warns that the current AI computing power shortage is driving companies to purchase in advance and accept premiums, but once supply catches up with demand or demand growth slows, upstream products such as chips and memory could face simultaneous declines in both prices and shipments, putting pressure on profit margins. Even if cloud providers' capital expenditure merely enters a plateau period, it could weaken suppliers' pricing power.

Power is the most direct constraint at present. Morgan Stanley's model shows that even accounting for power supply solutions such as nuclear energy, gas turbines, and fuel cells, the U.S. could still face a net power gap of 32 gigawatts by 2028, with the gap potentially expanding to approximately 72 gigawatts by 2029. Power delivery delays could prevent GPUs procured by cloud providers from being put into use in time, resulting in equipment hoarding and inventory buildup.

Morgan Stanley currently maintains a constructive view on AI infrastructure, believing that power shortages are more likely to cause project delays rather than demand completely disappearing. However, the bank warns that if political resistance, power bottlenecks, and supply-demand reversal compound with each other, it could trigger obstacles to chip procurement, declining upstream profit margins, and even reverse the entire AI capital expenditure cycle.