Nasdaq Falls 5.5% Into Correction as Midterms Hit AI Chip Stocks

On Thursday 5 November 2026 the Nasdaq Composite fell 5.52% to 24,315.00, its worst day since April 2025, and closed more than 10% below its October record, which puts it in a correction, according to CNBC's market wrap. The selling started in AI chips. AMD and Micron each lost more than 17% and Nvidia fell 9.26% to $181.60, per StockTitan's closing data. The trigger was the 3 November midterm elections. Democrats won both chambers of Congress, and local voters came out strongly against new data centers. The market read that as a threat to the AI buildout schedule that every chip valuation now assumes.

What happened

The week started the other way. On Wednesday 4 November the Dow rose more than 1,500 points, its best day since April 2025. Thursday took all of that back. The Dow lost 1,833.06 points (3.44%) to 51,523.06, the S&P 500 fell 4.36% to 7,353.73, and all three major averages had their worst session since April 2025, CNBC reported. The iShares Semiconductor ETF had its worst day since March 2020. The VIX touched 28.09 during the session.

The election result mattered because of what was on the local ballots, not only who won Congress. Ballotpedia's count found that voters favoured limiting data center development in 40 of 41 local measures on 3 November. They approved 27 restrictive measures, including five full bans, twelve partial bans and three six-month moratoriums in Maine, and they defeated all 13 measures that would have helped development. The single exception was Harwood, North Dakota. Reuters, as cited by StockTitan, reported that UBS strategists now see about 30% of the data center additions expected in 2028 as at risk.

Executives started saying the same thing out loud. Meta President Dina Powell McCormick said building data centers is "going to be harder" after the elections, CNBC reported. CNBC's live blog also noted that Sen. Bernie Sanders called for a national data center moratorium.

Friday brought a partial bounce. Nvidia rose 2.78% to $186.64 and AMD recovered to $449.04, but both stayed well below their October peaks, Decrypt reported via Yahoo Finance. Asia took longer to recover. South Korea's KOSPI fell 6% on Friday, Taiwan's benchmark lost more than 8% over the week, and Reuters reported that Taiwanese officials asked state-owned banks to support prices. On Monday 9 November Nvidia fell another 4.27% after its board opened an independent review following reports that its chief executive was the target of a foreign intelligence effort. That is a separate story, covered by CNBC.

The numbers

Why it matters

Chip valuations are a bet on construction. Nvidia guided to $108.0 billion of revenue for the quarter that ends in October, up from $96.2 billion in the quarter before, according to its second-quarter fiscal 2027 release. Revenue like that only arrives if somebody can pour concrete, connect power and get permits. Thursday was the first time this cycle that voters, rather than lenders or power utilities, looked like the constraint.

The selloff also spread past chips. CNBC reported that CoreWeave, Nebius and Iren all fell sharply. Those neoclouds borrow heavily against future GPU rental income, so a slower buildout hurts them twice: fewer new contracts, and more expensive refinancing.

What sceptics say

Here the sceptics are the bulls, because the strongest pushback against calling this a burst bubble is in the earnings. Qualcomm chief executive Cristiano Amon told CNBC, "We don't have a demand issue; it's a supply issue." AMD said its data center sales more than doubled in its latest quarter. Nvidia's last reported quarter showed revenue up 106% year on year. A 10% fall from a record is a correction, not a crash. For scale, the Nasdaq lost roughly 77% between March 2000 and October 2002, as Decrypt's piece points out.

The bear case is that profits are being spent faster than they arrive. CNBC noted that Meta's free cash flow turned negative in its latest quarter, after Alphabet and Amazon did the same in the prior period. If local permitting now slows the buildout as well, the industry gets the cost of the capex without the matching growth.

My take, as opinion: I would not read one ugly day as a verdict. I would read 40 of 41 ballot results as a durable change in the cost of building. That change shows up in your cloud bill eventually, whatever the share price does.

What it means for engineering and DevOps teams

• Capacity planning. If new data center capacity slips in contested US states, GPU and high-memory instance availability in those regions gets tighter. Keep at least one alternative region or provider qualified for your inference workloads, and test it rather than just listing it in a doc.
• Vendor risk on neoclouds. If you rent GPUs from a provider whose shares and bonds are swinging 10% a day, read the contract for termination, data export and notice terms now. The economics behind that market are in my GPU cloud breakdown.
• Cost. A slower buildout and higher financing costs mostly lead to firmer prices, not cheaper ones. Budget your AI spend per unit of work, as in FinOps for AI token costs, so a price change shows up quickly.
• Power. Most of the local opposition is about electricity and water. The background is in AI power in 2026.

FAQ

Did the AI bubble burst on 5 November 2026?

Not by any standard definition. The Nasdaq entered a correction, which means a fall of at least 10% from a recent high. The dot-com crash was a fall of roughly 77% over two and a half years. One bad day, with a partial bounce the next, does not settle the question.

Why did the midterm elections hit chip stocks?

Investors read the results as a sign that data center construction will slow. Democrats won both chambers of Congress, and Ballotpedia found that voters favoured limiting data center development in 40 of 41 local ballot measures. Fewer data centers means fewer GPUs and memory chips sold.

What are the next catalysts?

CoreWeave reports third-quarter results after the close on 10 November, and Nvidia reports on 17 November. Both will show whether customers are still signing for capacity at current prices.

Related: GPU Clouds in 2026: H100s at $2.59 an Hour, $104B of CoreWeave Backlog on $35B of Debt, and a Six-Year Bet on Silicon, The 2026 SaaS Selloff: IGV Fell 36%, Then Rose 45%. Revenue Grew 11% to 37% the Whole Time and AI Power in 2026: 485 TWh, a 0.3 GW Stargate, a 116 GW Turbine Backlog, and 1,800 MW That Dropped Off the Grid in Seconds.