Here’s something economists don’t see every day: companies spending more money precisely when things get more expensive. Morgan Stanley’s Andrew Sheets, the firm’s Global Head of Fixed Income Research, laid out the case in a May 11 podcast that AI infrastructure spending is fundamentally rewriting how the US economy responds to price signals.
The core observation is striking. Copper prices are up 40% year-over-year. Gas turbines have climbed 50%. Memory chips have surged between 150% and 300% in just twelve months. And demand for all of it keeps rising.
The $800 billion question
Morgan Stanley now anticipates $800 billion in AI-related capital expenditure from major US technology firms in 2026. That figure is nearly double the previous year’s estimates and roughly triple what was spent in 2024. The 2027 projection climbs even higher, to $1.1 trillion.
Sheets described this spending pattern as “uniquely price insensitive.” In plain English: companies building AI infrastructure don’t flinch at higher costs. They just keep buying chips, data centers, and power equipment regardless of what the invoice says.
