Half of S&P 500 earnings growth traces to AI data centre spending, Goldman estimate shows supplier revenue booked years before buyers book the expense, hyperscaler capex growth halves then slows to 17 percent by 2028

Images

zerohedge.com

Almost half of this year's earnings growth in the S&P 500 traces back to companies buying chips, servers, power gear and concrete to build artificial intelligence, according to a Goldman Sachs estimate published on September 23. The spenders themselves are not the source of the profit — the money lands as revenue at chip designers, memory makers, transformer and cooling suppliers, contractors, electricians and local utilities, each of whom books it immediately.

The accounting asymmetry is the whole story. When a hyperscaler commits ten billion dollars to a data centre, the seller records the revenue in the current year, while the buyer capitalises the purchase and expenses it through depreciation over five or six years. In the year the money changes hands, the economy records the supplier's profit without recording the buyer's cost. Lance Roberts, writing via RealInvestmentAdvice.com, calls this a golden window: a period in which supplier earnings and buyer expenses are deliberately out of phase.

The underlying identity is old. Jerome Levy worked out the profit equation early in the last century and Michal Kalecki formalised it in the 1930s: for the economy as a whole, profits equal investment minus household, government and foreign saving, plus dividends. Investment generates the profits that investment then claims to validate. The capex cycle Roberts describes runs in three stages — rising spending lifts supplier revenue, margins and hiring; flat spending freezes supplier revenue while the buyers' depreciation keeps climbing; falling spending turns the whole arrangement into a drag.

The numbers now on the table are large by any standard. Goldman puts hyperscaler capex at roughly 800 billion dollars in 2026, up 94 percent from about 412 billion in 2025, heading to 1.2 trillion in 2027 and 1.4 trillion in 2028. FactSet's July tally lands near the same 800 billion figure for calendar 2026. The absolute spending keeps setting records; the growth rate does not, falling from 94 percent to 50 percent to 17 percent across the three-year stretch.

FactSet expects calendar 2026 S&P 500 earnings growth of 32.8 percent, with information technology at 65 percent. Goldman's own sensitivity rule holds that a 250-billion-dollar surprise in 2027 capex, in either direction, moves S&P 500 earnings growth by roughly six percentage points. The base case assumes capex rises 50 percent to 1.2 trillion and delivers about 11 percent earnings growth. Hold capex flat at 800 billion instead, apply the same sensitivity, and 2027 earnings growth lands near 1.4 percent.

Most investors, Roberts notes, read the Goldman estimate as a story about the spenders. The estimate is equally a story about the suppliers, whose revenue this year is financed by depreciation schedules that have not yet begun.