Morgan Stanley Flags Labor, Power and Political Risks to AI Buildout

Morgan Stanley sees shortages of skilled workers, limited electricity access and political resistance as risks to the pace of AI data center construction alongside its $1.583 trillion investment projection for 2027.

The projection covers global AI computing investment by hyperscalers and neocloud companies. Of the projected $1.583 trillion, $1.388 trillion is cash capital spending and $195 billion is lease investment. It is not an estimate of spending on every part of AI infrastructure.

The three constraints concern how quickly planned facilities can be built. Morgan Stanley expects the buildout to continue, even as labor availability, power access and politics present obstacles to construction.

The bank’s earlier projection of global AI computing investment puts the total at $978 billion in 2026 and $1.812 trillion in 2028. Its model also projects that the investment cycle will contribute about 24% of S&P 500 revenue growth in 2027.

Those figures cover a broader group of companies and include leases, making them different in scope from Morgan Stanley’s July outlook. For the five largest U.S. technology companies, the July outlook projected nearly $800 billion in 2026, roughly $1.2 trillion in 2027 and $1.4 trillion in 2028. The two sets of figures cannot be directly compared.

The construction risks leave a distinction between projected investment and the pace at which new AI facilities can be completed. Morgan Stanley’s investment model projects continued growth through 2028, while its assessment of labor, power and politics identifies constraints on carrying out the buildout.