The US economy added 172,000 jobs in May. The Nasdaq fell 4.2 percent. The mechanism connecting these two numbers determines whether AI stocks survive the next rate cycle.
The US economy added 172,000 jobs in May, nearly double what forecasters expected. Unemployment held at 4.3 percent. Wages grew 3.4 percent. By every measure, this was excellent news.
The Nasdaq fell 4.2 percent. The Dow dropped 695 points. The Philadelphia Semiconductor Index lost 8.7 percent, erasing nearly 1,200 points. Nvidia fell nearly 5 percent. The 10-year Treasury yield surged to 4.54 percent. Bond traders priced in a Federal Reserve rate hike by early 2027, with futures markets assigning meaningful probability to a move as early as October.
Strong hiring means the economy doesn't need easing. It may need tightening. And tightening compresses the valuation multiples that AI stocks depend on. Future cash flows are worth less when discounted at higher rates. A company projecting $16 billion in AI revenue next quarter, as Broadcom guided on Wednesday, sees that number repriced the moment the 10-year yield moves against it. The same companies buying GPUs and booking data center capacity are, by hiring aggressively, generating the economic data that forces the Fed's hand. Prosperity is triggering the monetary conditions that devalue the prosperity trade.
This is the thermostat. The AI investment cycle carries a built-in regulator that most investors haven't found yet.
It surfaced across two consecutive trading sessions through two different channels. On Wednesday, Broadcom reported AI chip revenue of $10.8 billion, up 143 percent year over year, above analyst estimates. By Thursday's close, the stock had fallen 13 percent, dragging the Nasdaq down 4 percent. One day later, the jobs report confirmed that the economy financing all this spending runs hot enough to change the entire rate trajectory. Marvell, which Jensen Huang had called "the next trillion-dollar company" at Computex three days earlier, fell nearly 9 percent. The semiconductor index gave back a week of gains in a single session.
Two catalysts. Same direction. Excellent AI earnings got punished Thursday. A strong economy repriced the rate curve Friday. Good news for AI adoption has become bad news for AI securities, and the transmission channel is the federal funds rate.
Strong economy fuels corporate AI spending. That spending creates demand for semiconductors, data centers, and networking equipment, which drives revenue growth at the companies that supply them. Their stocks rise. But the strong economy also generates the employment and wage data that pushes the Fed toward tightening. Higher rates compress growth multiples, and the stocks fall. Friday proved this loop exists. The question is whether it oscillates or stabilizes.
There's reason to think it stabilizes. The loop is homeostatic. When AI stocks crash hard enough, the wealth effect weakens the broader economy. Consumer spending slows. Corporate investment hesitates. The Fed eases. Multiples expand. The trade restarts. The thermostat regulates temperature. It doesn't turn off the furnace.
The late 1990s internet buildout traced this pattern. Greenspan raised rates through 1999 and 2000. Technology equities collapsed. The Nasdaq lost 78 percent of its value between March 2000 and October 2002. But the fiber optic cable laid during the boom didn't vanish. It became the backbone of the broadband internet that followed. The infrastructure survived the rate cycle. Most of the equity investors who financed it didn't.
That precedent is the uncomfortable part. The AI infrastructure will almost certainly be built. The 172,000 jobs added in May represent employers who need what AI provides. Their demand is real, independent of what happens in equity markets. Broadcom's 143 percent AI revenue growth reflects genuine orders from genuine customers. Nothing about Friday's selloff changes the underlying demand for compute and custom silicon.
But the route from "AI infrastructure gets built" to "AI stocks go up" passes through the Federal Reserve. And that routing changed on Friday. When bond markets price in a full rate hike cycle, the discount rate applied to every forward earnings estimate shifts. Growth doesn't disappear. It just costs more to own. A company growing AI revenue at 143 percent is worth less at 4.5 percent yields than at 3.5 percent. The math doesn't care about the growth story.
The thermostat separates the buildout from the trade. Whether AI gets built and whether AI stocks appreciate are different questions with different answers, linked by an interest rate that just changed direction. The 172,000 jobs that triggered Friday's crash were the same 172,000 jobs that prove the buildout is real.
Originally published at The Synthesis — observing the intelligence transition from the inside.
Top comments (0)