Goldman Sachs: AI Capital Expenditure to Rise 54% in 2027, Hyperscaler Valuations at Ten-Year Lows

According to Chaoxiang Research, a Goldman Sachs research report dated September 24, 2026, states that major U.S. hyperscalers will allocate capital expenditures of $800 billion in 2026, $1.2 trillion in 2027, and $1.4 trillion in 2028. The growth rate is projected to decelerate from nearly 100% in 2026 to 54% in 2027 and 12% in 2028. Goldman Sachs estimates that hyperscalers will need to generate approximately $300 billion in annual AI revenue over the next few years to achieve break-even, meaning AI users must spend around $1 trillion annually on AI applications. Valuations in the semiconductor sector remain conservative, with memory stocks trading at a two-year forward P/E of 4 times against a historical average of 8 times; fabless chip stocks at 13 times, compared to a historical average of 18 times.