- Meta, Amazon, Microsoft, and Alphabet are expected to spend roughly $750 billion on capital expenditures this year, mostly on AI data centers.
- That’s more than the entire GDP of Ireland or Belgium, and over half of what all S&P 500 companies spent combined last year.
- Short-term memory prices used in data centers are more than 5 times higher than a year ago.
- Oracle invoked force majeure on a New Mexico data center last month after a key gas pipeline hit regulatory delays.
What Happened
The four largest US cloud companies — Meta, Amazon, Microsoft, and Google parent Alphabet — are expected to spend roughly $750 billion on capital expenditures this year, primarily on AI data centers, Bloomberg reported on October 12, 2026. That figure exceeds the entire GDP of Ireland or Belgium and is more than half of what every S&P 500 company spent combined last year. “We’ve never seen it,” said John McWilliams of real estate firm Cushman & Wakefield’s research group.
Why It Matters
Cushman & Wakefield’s cost breakdown shows the spending isn’t just about scale — unit costs are rising too. Trade wars, fuel costs, and geopolitical conflict have pushed up steel and cement prices, while the AI boom itself has driven up demand for chips, copper, and the short-term memory used inside data centers, now priced at more than five times what it cost a year ago, per Bloomberg Intelligence data. That cost backdrop sits behind moves like Amazon’s $1 billion pledge to communities near its data centers.
Technical Details
Demand for power-generation and transmission equipment is so high that supplier backlogs now stretch beyond 2030. Local opposition and regulatory setbacks are adding further costs, forcing developers toward pricier alternatives. Oracle issued a force majeure notice last month on a large New Mexico data center after a gas pipeline central to the project ran into regulatory delays.
Who’s Affected
Developers face what McWilliams calls “extremely expensive facilities to develop,” with nearly every major cost category — materials, components, power, land — rising at once. AI startups without a big-tech guarantor are increasingly locked out of the priciest capacity as cloud giants absorb the rising costs themselves.
What’s Next
Watch whether memory and power-equipment prices ease as new supply comes online, and whether more developers follow Oracle in invoking force majeure as regulatory delays collide with construction timelines.
