LIVE TICKER
[Live wire]Connecting to the newsroom…
WorldPublished 7 November 2026 • 9 min read • 2,155 words • 8 cited sources

AI stock selloff: Nasdaq falls 5.5% after Democratic midterm sweep

The Nasdaq fell 5.5% and lost an estimated $2.5tn in market value as Democrats' midterm sweep raised fears of limits on AI data centres. Markets partly rebounded on Friday.

AI stock selloff: Nasdaq falls 5.5% after Democratic midterm sweep
BBC News / Bloomberg via Getty Images — view at source ↗

Key facts at a glance

Nasdaq one-day fall
5.5% on Thursday 5 November 2026, its worst session in 19 monthsThe Guardian
Market value erased
About $2.5tn, according to Dow Jones dataBBC News
Public opposition to local data centres
71% of Americans, including 48% strongly opposedGallup
US data centre pipeline
3,041 operational, 499 under construction and 1,603 plannedData Center Map via BBC Verify
Expected AI investment in 2026
About $1tn, according to Goldman Sachs researchBBC News

What happened in the AI stock selloff?

The selloff started in New York on Thursday 5 November 2026 and rolled through Asia and Europe before Wall Street steadied on Friday. The trigger was political rather than corporate. Democrats won control of both chambers of Congress in the US midterms, and investors read the result as a threat to the data-centre building boom behind the AI trade.

The Nasdaq Composite fell 5.5% on Thursday, its worst session in 19 months, according to The Guardian. BBC News, citing Dow Jones data, estimated that the index lost about $2.5tn in market value. It closed more than 10% below its recent high, which puts it in correction territory.

Wall Street's fear gauge moved faster than the index. The Vix jumped more than 60% to its highest level since the early weeks of the Iran war, The Guardian reported. Investors also moved into government bonds, pushing the 10-year Treasury yield from 4.95% to about 4.85%, according to the BBC.

Friday brought a partial recovery rather than a full one. The Nasdaq closed up 1.1%, the S&P 500 added 0.3% and the Dow Jones Industrial Average slipped 0.4%, the BBC reported. Even after the fall, the Nasdaq remained higher than at the start of 2026, so the sell-off did not erase the year's gains.

Donald Trump, who had campaigned on accelerating AI, responded on social media. He described data centres as “gold” and wrote that only “foolish leadership” would turn down “the riches they bring”, the BBC reported. He has also pushed for the technology to be called “super intelligence” rather than artificial intelligence, and has dismissed safety fears as a “hoax”.

Futures had pointed the other way before the US open. Nasdaq futures were up 1.68% on Friday morning, the BBC reported, a sign that the sell-off reflected positioning and political uncertainty as much as a settled view of AI's economics.

Why did the US midterms trigger a tech selloff?

Data centres became one of the most visible issues of the campaign. Democratic candidates campaigned against the facilities' water and electricity demands, their noise and land use, and the effect on household bills, according to the BBC and The Guardian. Voters passed dozens of local ballot measures aimed at restricting projects.

The polling explains why the issue travelled so far. A Gallup survey conducted in March and published in May found that seven in ten Americans (71%) opposed building an AI data centre in their local area, including 48% who were strongly opposed. Barely a quarter favoured the idea, and only 7% were strongly in favour.

Opposition to data centres ran higher than opposition to nuclear power. In the same survey, 53% said they would oppose a nuclear plant in their area, against 71% for a data centre. Gallup said the results represented the first time it had asked about data-centre construction.

The backlash also crossed party lines. Majorities of every party group opposed a local data centre, although Democrats were more likely to be strongly opposed (56%) than Republicans (39%). Opposition was strongest in the Midwest (76%) and the South (75%), and weaker in the West (63%) and the East (68%).

The reasons for local opposition are concrete. In the same survey, 46% of Americans said they worried a great deal about the environmental impact of AI data centres, and another 24% worried a fair amount. Among opponents, half cited excessive use of resources, with 18% naming water and 18% naming energy. Supporters mostly cited economic benefits, including jobs.

Investors did not wait for legislation. They sold the companies most exposed to construction delays, then bought some of them back when no policy had actually changed. Dan Coatsworth of AJ Bell told the BBC that investors worry about the money being spent with “no clear idea of when, or whether, there will be payback”.

That worry predates the election. The BBC reported that concerns about returns had been building for months, and Goldman Sachs research cited by the BBC expects AI investment to reach $1tn in 2026 alone. The vote gave those concerns a date and a cause, which is often enough to force a repricing.

The boom itself began with ChatGPT's release in late 2022. As the BBC described it, investors then poured money into chips, networking equipment, data centres and the energy needed to run them, and the value of technology companies rose sharply. That concentration is why a political signal about data centres can move the whole market.

Which AI and chip stocks fell the most?

Chipmakers bore the brunt. Nvidia fell more than 9% on Thursday, according to the BBC, while AMD and Micron each dropped more than 17%. The Philadelphia Semiconductor Index, which tracks 30 large chip companies, fell more than 12%.

The losses were concentrated because the index is top-heavy. The BBC reported that the so-called magnificent seven technology companies, which include Apple, Nvidia and Tesla, shed a trillion dollars in value in a single trading day. When a handful of firms account for that much of an index, a political shock in one sector becomes a market-wide event.

Storage and memory names fell further. SanDisk tumbled 22% and Western Digital 17%, The Guardian reported. Both sit in the supply chain that has benefited most from AI servers, which made them easy targets when the spending outlook wobbled.

Tesla fell nearly 11% and SpaceX dropped 8%, together wiping more than $300bn off the two companies, the BBC reported. Elon Musk had spent nearly $450m backing Republican campaigns in the midterms, including at least $45m for Ken Paxton's Texas Senate bid. Texas nonetheless turned blue.

The damage was global. Japan's Nikkei 225 closed about 6% lower. South Korea's Kospi fell more than 8% at one point, triggering a circuit breaker, before closing down about 6%, with SK Hynix down more than 10%.

Japan's fall was not limited to index heavyweights. The BBC reported that shares in chip companies Kioxia and Tokyo Electron saw big falls as the Nikkei closed down 6%. Memory prices and AI server demand are closely linked, so the same fear travelled through the same supply chain.

In Europe, the FTSE 100 fell as much as 2.35% to 10,528 during morning trading before closing about 1% lower, The Guardian reported. Germany's Dax fell roughly 2%, France's Cac 40 lost 0.8% and Spain's Ibex 0.6%.

Friday's rebound was uneven. Nvidia rose about 7% in early trading, AMD 12% and Micron 8%, while the Philadelphia Semiconductor Index gained about 5%, the BBC reported. Tesla recovered only about 3% after losing a tenth of its value the day before.

Can Democrats actually stop data centre construction?

Congress does not issue building permits. Data-centre siting is decided largely by state and local governments, through zoning, utility approvals and land-use reviews. Democratic candidates also campaigned to end tax deals for technology companies, the BBC reported. The practical effect of a sweep therefore depends on where those decisions are made, not only on who controls the House and Senate.

Some newly elected Democrats have promised to use those levers. Texas governor-elect Gina Hinojosa said she would halt new data-centre projects until protections are in place. “There are no rules and laws to protect Texans,” she said on CNN, according to The Guardian.

Senator Bernie Sanders called for an immediate national moratorium. Representative Yassamin Ansari of Arizona called for a pause on advanced AI development, regulation of the industry and a ban on artificial superintelligence. New Hampshire senator-elect Chris Pappas praised voters for “opposing big data centers that harm our communities”.

Not every winning Democrat ran against AI itself. James Talarico, who won a Senate seat in Texas, framed the issue more carefully. He said he wanted to protect US leadership in AI while acknowledging that “we are still grappling with how it will affect everything from how we work to how our children learn”.

The rhetoric has already reached boardrooms. Meta platforms president Dina Powell McCormick told The Wall Street Journal it is “going to be harder” to build data centres after the vote, The Guardian reported. The White House's National Economic Council director, Kevin Hassett, countered on CNBC that local ballot measures would largely not affect existing projects.

The measures on the ballot were local, but their cumulative effect is national. The Guardian reported that opposition to data centres figured heavily in campaigns across the country and that voters passed dozens of local ballot measures restricting projects. For developers, that means a project-by-project fight over water, power and noise, even in states whose governors are friendly.

Industry analysts make a similar distinction between permits and plans. Dan Ives, a Wall Street tech investor, told the BBC that US data-centre capacity is set to double by the end of next year and that roughly a third of it is in Texas. Projects that already hold permits and are under construction, he argued, will not be stopped.

That leaves a narrower risk than the sell-off implied. New projects face longer odds in hostile states, while existing ones face protest and political pressure. The difference matters for companies that have already booked the power, the land and the construction contracts.

Is the AI data centre buildout slowing down?

The pipeline is still large. Figures from Data Center Map, shared with BBC Verify, count 3,041 operational data centres in the United States, 499 under construction and 1,603 planned. Virginia has the most existing and proposed sites, with 710, followed by Texas with 612.

Individual projects show the scale at stake. The Stargate campus near Abilene, Texas, will have eight buildings and more than four million square feet of internal space, with a planned capacity of 1.2 gigawatts, about the output of a nuclear reactor.

Meta's Hyperion campus in Louisiana sits on a site more than three times the size of Central Park, according to BBC Verify. Neither project depends on a vote in Congress, which is why analysts expect construction to continue even if the political weather changes.

Opinion among investors is split. Mohit Kumar of Jefferies told The Guardian he did not see the sell-off as “the start of the unwind of the AI trade”. AI investment has driven US growth, he argued, and Democrats would be reluctant to lose the jobs that data centres create. Chris Beauchamp of IG warned that more selling could follow.

Ives put the case for the industry in blunt terms. Even if 10% of the 800 to 1,000 data centres currently planned in the US are never built, he told the BBC, the remaining sites will be able to charge more for their capacity. He said the setback “does not cancel Christmas” and that the buildout would continue, only more slowly.

The more durable shift may be in how investors value the sector. Katy Kaminski of AlphaSimplex told the BBC that investors are trying to work out which companies will actually be profitable, rather than simply buying anything linked to AI. Tudor Brown, a UK tech founder, described “jitters” around the idea of infinite AI growth.

What should investors watch next?

The first test comes from company results and capital-spending guidance. If the largest cloud buyers reiterate their construction plans, the sell-off will look like a repricing of political risk. If they trim budgets, the market's fear will have been early rather than wrong.

Congress is the second test. A moratorium call from Sanders is not a bill, and federal legislation would take time even with a new majority. Hearings, federal permitting rules and the tax treatment of data centres are the more likely vehicles, and each creates headlines before it creates law.

State governments are the third. Hinojosa's pledge in Texas and similar promises from newly elected governors in other data-centre states matter more for 2027 projects than anything said in Washington this month. Utility commissions, which decide grid connections, may matter more still.

Interest rates remain the backdrop. The Bank of England raised interest rates and signalled that another increase could follow, the BBC reported. In the US, the fall in Treasury yields after the vote offered some support to valuations, even as tech shares were being sold.

Outside the US, the politics look familiar. Activists breached the site of a proposed Microsoft data centre in Amsterdam as a “Stop Dirty Data Centres” campaign began, the BBC reported. UK Prime Minister Andy Burnham said AI would stay “on the top of the agenda” and called for “a global code” to capture the benefits while guarding against the risks.

The UK is trying to sit on both sides of the argument. Burnham said there was a case for AI but that ministers had to be “very clear eyed” about the risks, the BBC reported. Last month's Budget included more than £600m for AI-related projects, including an extra £120m for the AI Security Institute and £100m for a national AI Biosecurity Lab.

For now, the sell-off has changed the terms of the debate rather than the direction of spending. The buildout still has more than 1,600 planned US sites, about $1tn of expected investment this year and a White House determined to defend it. What it no longer has is the assumption that local consent is automatic.

Books behind the context

Sources used for this dispatch

  1. BBC News - Global stocks plunge as fears over US AI slowdown spread to Asia and Europe
  2. BBC News - AI stocks climb after trillions wiped from market values
  3. BBC News - Why the US midterms sparked an AI stock selloff
  4. BBC News - US tech stocks edge back up after AI slowdown fears sent value of firms plummeting
  5. The Guardian - Global stocks fall amid 'big worries over AI' after Democrats' midterms triumph
  6. The Guardian - Several Democrats renew calls for datacenter pauses after blue wave
  7. TechCrunch - From the front row to the hot seat: what the midterms mean for Big Tech
  8. Gallup - Americans Oppose AI Data Centers in Their Area

This article is original analysis and synthesis. Facts, figures and quotations are attributed to the sources above. Images remain the property of their publishers and are credited in the caption.

Questions readers ask

Why did AI stocks fall after the US midterms?

Democrats won control of both chambers of Congress after campaigning against data-centre expansion. Investors feared slower permitting, new regulation and local opposition would delay the AI infrastructure buildout, so they sold the companies most exposed to it.

How much did the Nasdaq lose?

The Nasdaq Composite fell 5.5% on 5 November 2026 and lost an estimated $2.5tn in market value, according to Dow Jones data cited by the BBC. It was the index's worst session in 19 months.

Did AI stocks recover the next day?

Partly. The Nasdaq closed up 1.1% on Friday, and Nvidia rose about 7% in early trading. The rebound recovered only a fraction of Thursday's losses.

Have data centre projects actually been stopped?

No national pause has been enacted. Some newly elected Democrats and state governors have called for moratoriums, but projects already permitted and under construction are expected to continue.

Is this the end of the AI boom?

Most analysts cited by the BBC and The Guardian describe a correction and a repricing of political risk, not a reversal. More than 1,600 US data centres are still planned and AI investment is expected to reach about $1tn in 2026.

Which companies were hit hardest?

Chip and storage makers led the fall. AMD and Micron each dropped more than 17%, SanDisk fell 22% and Western Digital 17%, while Nvidia fell more than 9%.

Topics:AI stocksNasdaqUS midtermsdata centresNvidiaAI investment

More from the research desk