Bank of America Research Says AI Capital Expenditure Is 50% Above Expectations

Bank of America research shows that AI capital expenditure this year is 50% higher than forecast, while revenue is only 20% higher, and this mismatch runs through its subsequent analysis. The research also points out that the weighted average cost of capital for hyperscalers has risen from 4% in 2023 to 10%, and the risk-free rate is more than 4% above the pandemic low. The research estimates that capital expenditure by the top five hyperscalers plus neoclouds will be $956 billion in 2026, $1.32 trillion in 2027, and $1.5 trillion in 2028, with growth rates of 38% and 14%, respectively; demand estimates suggest that 212 GW of new data center capacity will be needed from 2026 to 2030, equivalent to about $9.5 trillion, which is $2.1 trillion higher than the current capital expenditure forecast of $7.4 trillion.

AI-generated from public reports

1. Oct 20· 1 reports

   Bank of America Research Says AI Capital Expenditure Exceeds Expectations by 50%

   Source:Rohan Paul