Goldman Sachs projects that the five largest US hyperscalers — Amazon, Alphabet, Microsoft, Oracle, and Meta — will spend approximately $1.2 trillion in AI infrastructure capex in 2027, a 50 to 54 percent jump from an estimated $800 billion in 2026, exceeding Wall Street’s consensus forecast. Cumulative investments could reach $7.6 trillion between 2026 and 2031, with an upside scenario of $1.4 trillion in 2027. Goldman Sachs estimates these five companies will need roughly $300 billion in annual AI-related revenue just to break even on their infrastructure investments. The key bottlenecks identified are energy supply, labor availability, and memory chip shortages, particularly high-bandwidth memory. The growing reliance on debt financing increases sensitivity to interest rates and raises the risk that AI revenue growth may not keep pace with spending.
Source: Read the original article
