Nasdaq Loses $2.5 Trillion as AI Hardware Stocks Get Crushed

Who actually got sold off

Not the companies paying for the AI buildout — the ones supplying it.

Look at the heatmap and the split is almost too clean. Alphabet fell 2.29%, Microsoft 3.25% and Amazon 3.55%. Annoying for shareholders, but hardly a rout. Then you get to the suppliers: AMD down 14.83%, Micron down 14.18%, Dell down 12.88%, GE Vernova down 11.88% on the grid turbine side. Nvidia lost 7.53%, Broadcom 8.07%, and chip equipment makers Lam Research and Applied Materials fell 12.44% and 10.31%.

The logic is straightforward once you look at the cash flow statements from last week. Alphabet posted negative free cash flow of $9.1 billion for the quarter. Amazon's trailing twelve-month free cash flow sits at -$42.4 billion while it guides toward $220 billion in annual capex. Those are companies with ad and cloud monopolies that can absorb a slowdown or change direction. Their suppliers are priced as if gigawatt data centers will keep compounding forever, with no friction.

This is the same pattern we saw in the Intel stock collapse and chip rout, where the pain concentrated in the companies furthest from the end customer. The hidden AI debt piling up across Big Tech makes the supplier side even more fragile.

The election just introduced that friction

Tuesday's midterms put data center permitting squarely in the path of the buildout.

Democrats flipped governor seats in Texas, Georgia and Ohio, and took Congress. Those three states are where a large share of near-term US gigawatt capacity is supposed to be built, and candidates openly ran on pausing data center approvals as residential electricity rates climb.

That matters more than a quarterly earnings miss. You don't need to take delivery of GPUs or server racks from AMD and Dell in Q1 if your 500MW interconnect permit is stuck at the state utility board. The market is now pricing in permitting delays, and the hardware vendors are the first place that shows up.

The power bottleneck is not hypothetical. As we reported on AI data centers potentially emitting as much CO2 as Mexico by 2030, the infrastructure buildout is already running into physical and political limits.

The consumer side is being squeezed too

The same AI buildout is raising memory prices and denting phone sales.

Qualcomm dropped 12.65% after handset chip sales fell 26%, right as Counterpoint reported global smartphone shipments down 20% to a 14-year low. The cause is partly the AI boom itself: Micron and SK Hynix have shifted wafer capacity to HBM for AI servers, pushing up standard DRAM and NAND prices and forcing phone makers to raise retail prices.

Consumers aren't paying extra for "AI phones," so unit volumes are falling. That is a demand problem the industry created for itself, and it lands on the same suppliers already exposed to the data center side.

What it means for you

For US, UK and EU readers, the practical effects show up in a few places. Higher memory prices are already feeding into phone and laptop retail pricing, and any permitting slowdown in Texas, Georgia or Ohio would delay the power infrastructure that data centers — and eventually local grids — depend on.

If you hold tech stocks or index funds, the correction means the suppliers that ran hardest are now repricing fastest. Nothing here changes the long-term demand for AI compute, but it does change how quickly the money gets spent.

The selloff is less a verdict on AI than a repricing of who carries the risk. Hyperscalers can slow down; their suppliers can't. Until permitting and power questions get answered in Texas, Georgia and Ohio, expect the hardware names to stay the most volatile part of the trade.