Inflation Data Splits AI Hardware Commitments from Software Valuations

The divergence follows sustained producer-price increases in semiconductors and power equipment. Primary records from company 10-Q filings show hyperscale capex guidance unchanged at $200 billion-plus for 2024, while software multiples have compressed on higher discount rates. Treasury yield curves above 4.5% on the 10-year have raised the hurdle rate for non-revenue-generating AI models.

US export controls on advanced chips to China create an additional bifurcation: domestic leaders capture restricted demand while non-strategic software firms face global competition without comparable barriers. Federal Reserve minutes from March document explicit concern that AI-related capital formation could add to demand-pull inflation, shifting the reaction function toward tighter policy.

Competing interests are clear in the ledger. Hardware suppliers gain from sovereign AI funds in the US, EU, and Gulf states; they absorb higher input costs but lock in multi-year orders. Software platforms lose on valuation resets and face delayed enterprise budgets as borrowing costs rise. Primary earnings transcripts confirm that only firms with contracted silicon allocations are maintaining guidance.

Next data points are May PCE and June FOMC dot plot revisions; sustained core services inflation above 3% would extend the split into H2.