The machines we're building to make us richer are making money itself more expensive.
The Summary
- Bond market indicators suggest AI investment is intensifying inflationary pressures just as Kevin Warsh takes over the Fed amid war-driven price shocks and rising yields
- BNP Paribas analysts say AI is worsening the Fed's inflation problem at precisely the wrong moment for monetary policy
- The capital flooding into AI infrastructure—data centers, chips, power grids—is competing for resources, driving up costs faster than productivity gains can offset them
The Signal
Kevin Warsh inherits the Federal Reserve chair on Friday facing a textbook policy nightmare: simultaneous supply shocks from war and demand shocks from the largest infrastructure buildout since rural electrification. Bond yields are climbing as markets price in sustained inflation, but the source isn't just geopolitics. It's also the hundreds of billions pouring into AI compute.
The infrastructure demands of frontier AI are colliding with an already-strained economy. Data centers need power, cooling, and land. Chip fabs need specialized construction and rare materials. Every hyperscaler racing to 100,000 H100 clusters is bidding up the same constrained inputs: electrical engineering talent, transformer substations, construction crews already booked solid through 2027.
"The AI boom is only worsening Kevin Warsh's inflation problem."
This creates a perverse dynamic for monetary policy. Normally, the Fed would raise rates to cool demand and tame inflation. But AI investment isn't discretionary consumer spending that rates can easily suppress. These are multi-year capital commitments from companies with fortress balance sheets. Microsoft and Google aren't canceling data center projects because the 10-year Treasury moved 50 basis points.
James Egelhof, Chief US Economist at BNP Paribas, points to bond market indicators that capture this tension: inflation expectations embedded in Treasury Inflation-Protected Securities spreads are rising even as the Fed maintains restrictive policy. The market is telling us that AI-driven demand is proving stickier than traditional economic cycles would predict.
Here's what makes this different from past tech booms:
- Physical infrastructure requirements are front-loaded and massive, not gradual
- Productivity gains from AI arrive later, after the investment wave drives up input costs
- Competition for scarce resources (power, specialized labor) is concentrated in narrow bottlenecks
The Implication
Warsh walks into a job where his two mandates—stable prices and maximum employment—are being pulled apart by the very technology promising to supercharge future growth. If he tightens too much, he risks strangling the AI buildout that could eventually ease inflation through productivity. If he stays loose, he validates market fears that the Fed has lost control of prices.
Watch the composition of inflation data over the next six months. If construction costs, electrical equipment, and specialized labor categories stay elevated while consumer goods cool, that's the AI boom's fingerprint on monetary policy. The bond market is already pricing it in. The question is whether Warsh can navigate it without breaking something.