Amazon, Alphabet, Microsoft, Oracle and Meta will expand their data center footprints and energy procurement to support new machine learning workloads. Goldman Sachs strategists forecast these firms will spend $1.2 trillion on AI infrastructure in 2027, a 54% increase from 2026, and $1.4 trillion by 2028. This figure for 2027 is above the Wall Street consensus target of $1.1 trillion, though the annual growth rate is expected to slow from nearly 100% this year to 12% by 2028.

The cloud operators must generate roughly $300 billion in annual AI revenue to break even on these investments, according to the firm. Goldman estimates that spending surprises of $250 billion in either direction next year could shift S&P 500 earnings growth by about 6 percentage points. Total investment across compute, data centers, and power may reach $7.6 trillion by 2031 as the primary bottle neck shifts from chip production to available power and physical infrastructure.