Goldman Sachs Projects $1.2 Trillion Big Tech AI Infrastructure Spend in 2027

Goldman Sachs has delivered its most bullish forecast yet on Big Tech’s AI infrastructure spending, projecting that Amazon, Alphabet, Microsoft, Oracle, and Meta will collectively pour $1.2 trillion into AI infrastructure in 2027—a 50% increase from roughly $800 billion in 2026.

The forecast, which sits above Wall Street’s consensus of $1.1 trillion, positions the coming investment cycle as the largest relative to GDP since the 19th-century railroad construction boom. The five hyperscalers have been locked in an unprecedented capital expenditure race, each building out massive GPU clusters to train and serve frontier models.

Decelerating growth, still massive scale. While the headline number is staggering, the growth rate is actually slowing: from approximately 100% in 2026 to 54% in 2027, and just 12% in 2028. This moderation suggests the industry is approaching an inflection point where infrastructure buildout begins to outpace demand.

The bank estimates the group needs to generate approximately $300 billion in annual AI-related revenue to justify the spending—a steep hill climb even for the world’s most valuable companies. However, cloud revenue growth is accelerating: from 25% in 2024 to 48% in Q2 2026, providing some validation for the capex surge.

Treasury yields add pressure. The investment thesis faces headwinds from rising borrowing costs. The 10-year Treasury yield now sits near 5.17%, up roughly one percentage point since January. For debt-heavy operators like CoreWeave, every 100-basis-point rise adds approximately $30 million to annual interest expenses on floating-rate debt. Oracle’s quarterly interest expense has jumped 55% to $1.43 billion, and the company’s shares are down about 30% year-to-date.

JPMorgan estimates $4.1 trillion in AI-related debt will be issued through 2030, making the debt cost dynamics increasingly material to the investment case.

The Goldman forecast underscores a central tension in the AI infrastructure story: the capital requirements are unprecedented, the competitive dynamics demand continued spending, and the timeline for returns remains uncertain—even as the hyperscalers show no sign of pulling back.