The AI boom has been the driver of returns since 2022, making equities, the credit market and the US economy reliant on the AI theme. Thursday was a perfect example of how intertwined the AI theme is across all three.
What happened?
On Wednesday the market was positive as the election cleared a major event risk, and investors bet the Trump administration would be reined in. We saw this play out as oil dropped from $100 to $94 per barrel. On Thursday, however, investors reassessed the wider implications for data-center expansion, as Democrats campaigned on its impact on power and utility costs and local restrictions gained traction. As a result, the market repriced a delay in the build-out, which pushes out 2027-2028 earnings. While this was more pronounced on Thursday, the delay was already underway, as semiconductor companies had flagged possible build-out delays on their earnings calls.
What made it so violent?
Retail investors started the selloff with their heaviest net single-stock selling since the Covid crash, and the size of the move triggered rule-based sellers.
- Options dealers: clients bought index puts and dealers took the other side, hedging by shorting S&P futures. As the market fell, the puts became more sensitive to the market, forcing dealers to short more futures, which added to the pressure.
- Volatility-control funds: these funds cut equity exposure when volatility rises. With the VIX jumping from 17 to 28, they had to sell roughly $61bn.
- Levered ETFs: funds that promise 2x or 3x the daily return must rebalance at the close. After a big down day, they have to sell to bring leverage back in line.
- CTAs: trend-following funds flipped to sell as the S&P and Nasdaq broke below key price levels, keeping pressure on the downside.
What does this mean for the economy and rates?
Most of the repricing happened in October, and Thursday was the confirmation. We do not believe this is the end of the AI build-out, but rather a delay. Before Thursday, the market was rotating out of AI into other sectors. This time, however, selling broadened beyond AI and money moved into Treasuries. The 10-year fell from 4.95% to 4.86% by Friday, a sign the economy may slow, which goes hand in hand with delayed earnings given how exposed the economy is to AI. In the rate-sensitive middle of the curve, 2-, 3- and 5-year yields fell about 10bp, suggesting the market has pulled back its expectations for a hawkish hiking cycle.
Now what?
The AI story is delayed, not over. In equities, the AI infrastructure build out now trade at 13.1x forward earnings, so much of the delay is priced in. However, the group likely needs clarity on data-center policy or capex guidance to re-rate. Companies with high debt or elevated earnings expectations remain the most exposed to a slower build-out.
For credit, the delay is actually a positive. Hyperscalers have been the main source of pressure on spreads this year. A slower pace of spending means less borrowing and gives cash flows time to catch up. As such, what is negative for AI equity earnings is a positive technical for credit. We saw this on Thursday when IG spreads widened only 2bp while the S&P fell 4.4%.
