Goldman says hyperscaler AI spend could jump to $1.2 trillion next year

The headline numbers

Goldman Sachs' team led by Ryan Hammond projects that Amazon.com Inc., Alphabet Inc., Microsoft Corp., Oracle Corp. and Meta Platforms Inc. will take combined AI-related capital outlays to $1.2 trillion next year. That mark sits above a widely cited 2027 capex consensus of $1.1 trillion. The firms are tracking toward $800 billion of AI buildout this year.

As the strategists put it, "Based on consensus estimates, capex in 2027 is on track to reach a larger share of GDP than any technological investment cycle since the railroad build-out in the late 1800s." To make those investments pay, Goldman estimates the companies would need roughly $300 billion a year in AI revenue over the coming years.

Growth trajectory and financing

Goldman sees the surge in spending moderating. After roughly doubling this year, their model points to growth stepping down to 54% in 2027 and then to 12% in 2028, when total hyperscaler capex is projected to reach $1.4 trillion. The team also flagged that outlays have pushed past cash generated from operations, which raises the odds of more borrowing and equity issuance.

"The rate of both capex growth and upside surprises to consensus estimates will diminish relative to recent quarters," the note said. Potential speed bumps include limited power availability, tight labor and memory supply, and restrictions on where new data centers can be built.

Big ideas change the economy, so steady planning protects your financial future. Join Briefs Finance CEO Jaspreet Singh on September 29th for a FREE live investor workshop, How to Profit From A Dollar That's Losing its Value, where he shows how we're spotting investment opportunities as the dollar falls. Save your spot.

Revenue, valuation and market signals

Goldman estimates that customers would need to buy on the order of $1 trillion per year of AI application software for hyperscalers to earn attractive returns and for app makers to keep healthy margins on compute costs. For context, total global software outlays are pegged at about $1.5 trillion in 2026. Even so, the firm says current AI revenue estimates remain below break-even needs, although they are climbing fast. Growth in cloud sales at Amazon, Alphabet, Microsoft and Oracle quickened to 48% in the latest second quarter from 25% in 2024, and Amazon, Alphabet and Microsoft together have disclosed a $1.7 trillion cloud revenue backlog.

On pricing, skepticism has crept in. Investors currently assign the median AI infrastructure stock a forward P/E of 22x; in April, that figure was 32x. For the hyperscalers, valuation multiples have fallen to levels not seen in more than ten years, and the premium they command over the typical S&P 500 company is at a record low. For your wallet, that mix of heavy spend, rising sales and cooler multiples means the AI buildout is real, but the payoff timeline is doing most of the talking.

When innovation reshapes industries, disciplined saving and smart decisions keep wealth growing. Our CEO Jaspreet Singh is hosting a FREE live investor workshop, How to Profit From A Dollar That's Losing its Value, on September 29th. Sign up free to join him live.