The 10-year U.S. Treasury yield has reached its highest level in nearly two decades, and for the AI capex cycle, that number is not a background condition. It is the price at which the entire buildout is being financed.
A higher 10-year yield means a higher discount rate, which means a lower present value for every future cash flow that an AI company is promising to its investors. The hyperscalers that are spending the most on data centers and chips are the ones most exposed, because their returns are back-loaded into the 2030s. A two-decade-high yield compresses the window in which those returns have to materialize, and it raises the cost of the debt that funds the buildout.
The second-order effect is on the banks that underwrite the capex. A higher yield environment is not a bad environment for a bond bank. It is a good one. But it is a bad environment for the equity multiple that the AI stocks are currently trading at, and the gap between the two is where the next correction lives. The market is pricing in growth. The bond market is pricing in cost of capital. Those two prices are now diverging, and the divergence is the risk.
For the next quarter, the question is whether the yield curve will stay elevated long enough to force a repricing of the AI stocks, or whether the growth narrative will outrun the discount rate. The 10-year yield is the single number that will decide, and it is at a level that the last generation of investors has not seen in twenty years.
Image source: i.ibb.co