Goldman says AI hyperscalers need $300 billion revenue to break even

Goldman Sachs estimates the largest U.S. AI hyperscalers need about $300 billion in annual AI revenue to break even as 2026 capital expenditures head toward $800 billion.

Consensus estimates call for hyperscaler capital expenditures to increase to $1.1 trillion in 2027. Goldman’s baseline has spending exceeding that forecast, although the firm expects both the pace of growth and the scale of upside surprises to slow.