Amazon, Alphabet, Microsoft, Oracle and Meta will spend a combined $1.2 trillion on AI infrastructure in 2027, Goldman Sachs estimates. That would rise more than 50% from roughly $800 billion projected for 2026 and exceed Wall Street's $1.1 trillion consensus.
Strategist Ryan Hammond's forecast, reported by Bloomberg, would make the investment cycle the largest since nineteenth-century railroad construction when measured against gross domestic product. Growth would then slow sharply: Goldman projects the pace falling from nearly 100% in 2026 to 54% in 2027 and 12% in 2028.
The commercial hurdle is large. Goldman calculates that the companies would need about $300 billion in annual AI revenue to recover the spending. Cloud revenue growth accelerated from 25% in 2024 to 48% in the second quarter of 2026, but current earnings still do not settle whether demand can support the buildout.
Financing and construction could constrain the plan before demand does. Spending now exceeds cash generated from operations, increasing reliance on debt, while power, skilled labor and memory chips are potential bottlenecks. The $1.2 trillion figure is a bank forecast rather than announced company budgets.