By most measures, the U.S. economy is doing fine. Third-quarter GDP grew at a 3.4% annualized pace, corporate profits are near records, and the job market is cooling but hasn’t cracked. So why did the Nasdaq 100 slide into correction territory at the start of November, falling 10% from its peak?
The short answer is that investors have stopped asking whether artificial intelligence is real and started asking who is going to pay for it.
For two years the story was simple. The largest technology companies were spending staggering sums on chips, data centers and power, and they were paying for most of it out of cash flow. That is no longer the case. Hyperscaler bond issuance reached roughly $220 billion by mid-August, compared with about $12.5 billion over the same stretch a year earlier. In early November, Oracle announced plans to raise $10 billion in new stock and another $10 billion in bonds to keep its buildout going. The shares fell hard, and that reaction tells you most of what you need to know about where the market’s head is.
Demand for AI isn’t the problem. This earnings season has largely confirmed that customers want more computing power than the industry can supply. The problem is the cost of building it, and that cost just went up. The Federal Reserve raised rates in September for the first time since 2023, held at 3.75% to 4% in late October with two members pushing for another hike, and markets are bracing for a December increase. Elevated oil prices tied to the conflict with Iran are keeping inflation sticky and the Fed on edge, even as hopes for a diplomatic resolution come and go. When money is cheap, investors will fund a decade-long bet largely on faith. When money costs 4%, they want to see the returns.
That shift is showing up beneath the surface. At one point in early November, the S&P 500 fell on a day when most of the stocks in it rose. Money is leaving the most crowded corner of the market and finding its way into healthcare and other areas that had been left behind. Bond investors, meanwhile, are demanding the widest premium to lend to tech companies in more than two decades.
We don’t read this as the end of the AI cycle. We read it as the market growing up. Companies with real earnings, strong cash generation and the discipline to match spending to revenue should come through this fine, and probably stronger. Those financing ambition with borrowed money and diluted shareholders will have a harder time.
For our clients, this is why we have spent the past year trimming concentrated tech exposure, rebalancing toward parts of the market the index had left behind, and adding high-quality bonds that now pay a real yield. A balanced portfolio feels dull in a melt-up. It earns its keep in weeks like this one.
With the midterm elections now behind us, expect the market to spend some time sorting out what the results mean for specific industries. Elections can shift that outlook at the margins, but they rarely change the underlying math. Right now, that math is about the price of money and what it buys.