Five companies are about to outspend the Pentagon. Let that sit for a second.

Morgan Stanley projects that combined capital expenditures from Alphabet, Amazon, Meta, Microsoft, and Oracle will hit roughly $1.1 trillion by 2027, representing approximately 3.2% of projected US GDP. To put that in perspective, the annual US defense budget, long considered the ultimate expression of national spending priorities, would be smaller than what five tech CEOs decide to pour into AI infrastructure.

The numbers behind the AI arms race

Morgan Stanley’s 2026 forecast for combined capex across these five companies was recently revised upward to around $805 billion, up from a prior estimate of $765 billion.

Most of this capital is flowing into AI-related infrastructure. Data centers, GPU clusters, and the leases required to house them are consuming the lion’s share of these budgets.

David Sacks, the tech investor and current White House AI and crypto czar, has estimated that AI capex will provide a tailwind of more than 2.5% to GDP growth this year. By 2027, he projects that figure climbs above 3%.

For historical context, the late-1990s telecom boom saw inflation-adjusted spending of approximately $200 billion. The current AI investment cycle is set to dwarf that by a factor of five.

What this means for investors

The most obvious play is the companies doing the spending themselves. Alphabet, Amazon, Meta, Microsoft, and Oracle are essentially betting that their AI infrastructure investments will generate returns that justify historically unprecedented capital deployment.

The second-order effects may be more interesting. Companies supplying the picks and shovels for this buildout, think chip designers, data center REITs, power generation companies, and cooling technology providers, stand to benefit from a demand curve that shows no sign of flattening. The revised 2026 forecast from Morgan Stanley, moving from $765 billion to $805 billion, suggests analysts are still catching up to the actual pace of spending commitments.