The scale of Big Tech’s AI investment is starting to reshape the US economy in ways once reserved for national priorities. Combined capital spending on AI by Alphabet, Amazon, Meta, Microsoft and Oracle is projected to reach roughly 3.2% of US gross domestic product by 2027, according to a Morgan Stanley analysis built on Congressional Budget Office data. If that holds, annual AI capital expenditure would surpass US national defense spending, estimated at about 2.6% of GDP, for the first time in history.
The ramp is steep. AI capital spending by the five companies sat near 1.45% of GDP in 2025 and is set to climb to about 2.5% this year, nearly matching defense, before the crossover in 2027. In dollar terms, their combined outlay is projected to top $800 billion in 2026, a figure Morgan Stanley recently revised upward from $765 billion, and to reach a record $1.1 trillion by 2027. Nearly all of it flows into AI infrastructure: data centers, GPU clusters, custom chips and the power and leases to run them. White House AI adviser David Sacks has estimated the buildout is adding more than 2.5 percentage points to GDP growth this year.
Why It Matters
The comparison to defense is imperfect but revealing. Data centers do not build submarines or fund troops, so measuring the two as a share of GDP speaks to the scale of capital, not military power.
Yet the connective tissue is real, since AI now underpins intelligence, cyber operations and logistics, and every hyperscaler holds defense contracts. That a handful of companies will soon direct more capital toward AI than the country spends on its military underscores how central private compute has become to the economy.
Reasons for Caution
The spending is increasingly funded by debt and rests on returns that remain unproven. Analysts warn that capex is consuming close to all of these firms’ operating cash flow, squeezing the free cash that funds buybacks and dividends, with some projecting sharp declines this year.
Much of US GDP growth is now propped up by the buildout, which Deutsche Bank has cautioned carries “no guaranteed return.” Power availability is emerging as a binding constraint, and the central question, whether AI revenue can grow fast enough to justify a trillion-dollar annual habit, is still unanswered. Should demand disappoint, an economy leaning this heavily on AI investment would feel the pullback quickly.
