AI Is Repricing the Cost of Money

“The cost of building AI is arriving before the cost savings from using it.”

The central thesis is uncomfortable but simple: AI is already tightening the monetary and financial environment — before its full productivity dividend is visible.

The mechanism runs in two phases. First, the cost: AI capex drives power, compute and infrastructure demand, which creates financing pressure, higher real yields, higher hurdle rates and less room for easing. Only later comes the dividend: AI diffusion lifts output, lowers unit costs, delivers disinflation and restores room for easing.

Three central-bank signals now support that reading.

1. 01 Fed — Rate Channel

   Cleveland Fed President Beth Hammack points to higher real yields as heavy technology investment intensifies financing demand. That demand is adding pressure while real yields remain elevated. signal: Heavy technology investment is lifting financing demand into an environment of elevated real yields.

2. 02 Fed — Systemic-Risk Channel

   Kansas City Fed President Jeff Schmid is asking whether the AI ecosystem itself could become systemically important, as it grows more concentrated, leveraged and interconnected. signal: The AI ecosystem is concentrating, levering up and interconnecting — a systemic-risk question in the making.

3. 03 ECB — Macro-Scale Channel

   The ECB sees AI effects across semiconductor constraints, investment and corporate financing — evidence the demand is visible at macro scale. signal: AI demand shows up in chip constraints, investment and corporate financing across the euro area.

Board Takeaways

govern the capex

• Raise the hurdle rate.
• Force CapEx accountability.
• Secure energy, compute, data and skills.
• Stress-test higher-for-longer real yields.
• Govern the conversion from AI spending to competitive advantage.