Five companies, $4.2 Trillion and a lot of debt: how the AI buildout is concentrating risk while creating 750,000 jobs in the US

A small group of technology companies, Microsoft, Meta, Alphabet, Oracle, and Amazon, is expected to spend USD 4.2 trillion over the four years through the end of 2029, according to FactSet estimates.

That figure is the centerpiece of a new study by economist Stijn Van Nieuwerburgh, published by the Brookings Institution, and it is also, in the paper's telling, one of the primary sources of risk in the American economy's fastest-growing investment cycle. The concern is not just the size of the spending but how it is being financed.

A handful of hyperscalers and a borrowing boom

Hyperscalers are increasingly turning to debt, and Van Nieuwerburgh notes that technology companies are using off-balance-sheet entities and other financing structures to borrow from banks and private credit firms. These arrangements, he writes, can make the underlying exposures difficult to assess.

The Brookings paper identifies specific structures in use: joint ventures, private credit, securitizations, and special-purpose vehicles. What worries the economist is not the technology sector alone. If AI fails to generate enough revenue to support the debt raised to build data centers and related infrastructure, the effects could reach into the broader financial system.

The study frames the question as this: not whether AI infrastructure is already a systemic risk, but under what circumstances losses at individual projects could become correlated and spread across companies and financial institutions.

The biggest infrastructure cycle in American history

The spending on those financing structures is, by the study's projection, of historic scale. Total investment in data centers and related AI infrastructure is projected to reach USD 10.3 trillion between 2025 and 2032, an annual average equivalent to 3.6% of U.S. GDP.

According to reporting by The Wall Street Journal, that would make AI the largest infrastructure investment cycle in U.S. history relative to the size of the economy, surpassing the great buildouts of the past: canals, railroads and electrification. A comparison from Goldman Sachs puts the current moment in context.

AI investment in the United States is estimated to amount to 1.9% of GDP in 2026, a record, and still barely half the projected average annual pace through 2032. The gap between those two figures illustrates how far spending would have to climb to meet the Brookings projection.

Labor, land and power under strain

While the money flows through financial markets, the physical buildout is reshaping resource markets on the ground. Data centers are consuming an increasing share of the workforce and the electricity supply, potentially raising costs for other businesses.

The Federal Reserve Bank of Richmond has reported that data center construction is straining labor supplies in its region.

Land is another contested resource. In areas where large numbers of data centers are being built, demand for land can push up prices and compete with other industries, including manufacturing. In communities hosting large data center projects, the increased demand for electricity is contributing to higher power costs.

The pressures are visible in financial conditions as well. Federal Reserve Governor Kevin Warsh has cited borrowing by hyperscalers as one factor contributing to higher long-term interest rates, a shift that can make homeownership less affordable.

The jobs boom on the other side of the ledger

The same construction wave that is straining labor supplies is also generating employment at a remarkable pace. LinkedIn estimates that more than 750,000 AI-related jobs were created in the United States between 2023 and 2026, a figure that helps offset some of the disruption AI is causing in the broader labor market. These positions pay well.

The median advertised salary for AI-related jobs on LinkedIn is about USD 180,000, compared with USD 80,000 across other roles. The demand extends beyond software engineers.

In the Washington, D.C., area, the number of unionized electricians has risen from about 9,000 to 17,500 in recent years as data center construction has accelerated, evidence that the boom is pulling in skilled trades alongside technical talent.

A USD 63 trillion wealth effect

Financial markets have registered the cycle too. According to Federal Reserve data, U.S. stock and mutual fund holdings totaled USD 63 trillion in the second quarter, nearly double the level at the end of 2022. The surge in AI-related stocks has generated a dramatic increase in financial wealth, though much of that wealth is concentrated among households that already own large amounts of equities.

Unresolved questions about a downturn

Taken together, the figures describe an investment cycle that is simultaneously creating hundreds of thousands of jobs, driving up stock-market wealth, and straining labor markets, land prices, electricity supplies, and interest rates. What remains unresolved, per the Brookings study, is whether the financial system can absorb a downturn in the cycle without correlated losses spreading across companies and institutions.

The answer hinges on factors the study does not resolve: whether AI generates sufficient revenue to service the debt raised to build the infrastructure and under what conditions losses at individual projects could propagate through the banking system and private credit markets.

What the available numbers do make clear is the magnitude of what is under construction: USD 10.3 trillion in projected investment, USD 4.2 trillion of it from five companies, financed increasingly through structures that make exposures difficult to assess, all against a backdrop of rising power costs, tighter labor supplies, and record equity wealth.