Hyperscalers need $300 billion in annual AI revenue to break even: Goldman

Goldman Sachs estimates that the top U.S. AI hyperscalers must generate approximately $300 billion in annual AI revenues to break even on their substantial expenditures. Analyst Ryan Hammond highlighted that the hyperscalers are set to invest $800 billion in capital expenditures by 2026, with consensus forecasts projecting this figure to reach $1.1 trillion in 2027.

Hammond's projections suggest that hyperscaler cloud revenues have surged by roughly $70 billion above the pre-AI trend in Q2 2026, with announced revenue backlogs surpassing $1.5 trillion. To earn solid returns, AI users would need to allocate around $1 trillion annually to AI applications, in comparison to the current $1.5 trillion global software spending.

Despite skepticism surrounding the long-term earnings of AI infrastructure stocks, Goldman asserts that the hyperscalers are trading at low multiples and favorable signs of returns on investment, coupled with a deceleration in spending growth, should bolster share prices. Goldman also anticipates the AI software and services sector to continue delivering both successful and unsuccessful stocks.

While enterprise adoption remains in its nascent stages, Hammond notes that the accelerated pace of enterprise AI spending indicates that the impact of AI on corporate earnings will become increasingly apparent in upcoming quarters.