AI-related chip demand may be a longer-lasting inflation pressure than the Federal Reserve usually expects, San Francisco Fed president Mary Daly told Axios. She says some businesses are already considering forward contracts for memory chips and redesigning products to use fewer chips, signs that the squeeze could reach beyond data centres.
Watch Desk analysis
What happened
Daly told Axios that companies in her district are preparing for tighter chip supplies. Some, she said, are seeking forward contracts to secure memory chips; others are reworking products to rely less on chips. She pointed to the risk that demand for AI hardware could compete with chips used in cars, appliances and other goods.
Daly said the Fed often thinks of supply shocks as fading within one to three years, but this one may take longer to ease. She also cautioned that AI demand is growing, and that it could compound with energy costs and tariffs.
Why it matters
If demand for AI equipment tightens supply of chips used elsewhere, businesses outside the data-centre boom could face higher costs. That would make this more than a question of who gets the latest accelerators: it could affect product design and prices across the wider economy.
There is a monetary-policy wrinkle, too. Daly said higher interest rates may do less to slow the hyperscalers driving much of the AI build-out, although borrowing costs can still affect other businesses and the broader inflation outlook. Her account points to a risk the Fed is watching, not proof that a general chip shortage or fresh wave of price rises has arrived.
Our read
The revealing detail is not another large forecast for AI spending. It is that, according to Daly, some firms are already looking for ways to secure memory or use less of it. That is a useful early signal, but it is still an account of what she is hearing from business contacts, not a measured map of shortages. Keep an eye on what happens to chip prices and supply beyond the data-centre market; the boom’s bill may not stay neatly in one corner.
What to watch
- Whether companies outside the AI sector report tighter memory-chip supplies or higher prices.
- Whether forward contracts and product redesigns become more common.
- Whether energy costs, tariffs and AI-related demand ease separately or compound.
- How the Fed assesses persistent supply pressure alongside its other inflation and employment concerns.
Discussion spark: If AI infrastructure demand starts competing with chips for cars and appliances, should governments prioritise supply for the wider economy, or let the market decide who gets the chips?
Sources and evidence
- SF Fed president: AI demand could extend energy shock (6 October 2026, 15:55 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.