Morning Edition · Sunday, November 8, 2026Published at 12:20 AM EST · New York
Nvidia fell 9.3% and the Nasdaq Composite fell 5.5% on Thursday, while about 2,000 people marched against data centers in Dublin and 1,200 in Helsinki.

Voters, and not only interest rates, are now limiting artificial intelligence (AI) spending. In the US midterms on Nov. 3, Democrats won both chambers of Congress. The Associated Press counted 54 Senate seats and 254 House seats for the party, with some races still uncalled. Two days later, AI and chip shares fell sharply. Nvidia dropped 9.3%, SanDisk lost 20%, Western Digital and AMD each lost 15%, and Micron and Seagate fell about 13%. The Nasdaq Composite fell 5.5% and entered a correction. The largest technology companies together lost about $1 trillion in market value.
The Financial Times summarized the political problem in its headline "Data centres rebuked", with the subhead "Capex, meet the electorate." NBC described the result as a wave of victories by anti-AI candidates. The same arguments are now heard in Europe. Brazil's InfoMoney reports that about 2,000 people marched in Dublin and another 1,200 in Helsinki against new data centers. The two cities host large clusters of the energy- and water-intensive facilities.
Not every government is retreating. Dan Tomlinson, Britain's growth minister, said the UK will not slow down on AI despite the US backlash. He called the current pace of investment "too low for what we need to stay at the forefront." London's position puts it in competition for data-center projects that may struggle to get permits in US states or Irish counties.
From an Austrian perspective, the AI build-out is a large, rate-sensitive capital project that depends on low-cost power, fast permits and cheap financing. All three inputs are now at risk at once. Rising short-term rates make the debt-financed portion more expensive, and a hostile Congress raises the regulatory risk on the remainder. The memory and storage makers fell furthest because they sit at the end of the order chain, where lower spending by the largest buyers shows up first.
Part of a tracked trend
AI Spending Meets Funding Limits
As AI capital spending shifts from internal cash flow to debt and private funding, each new spending increase will be answered with lower supplier valuations and higher financing costs until spending growth slows to what measured returns can support.
If the election caused the sell-off, Democratic lawmakers and local opponents of data centers gain leverage over siting and power prices. Jurisdictions that court capacity, such as Britain, gain an argument for investment that US states may lose.
The closing figures hold up, with the S&P 500 down 4.36% and Nvidia down about 9.3%. But chip stocks were already falling before the vote count, and Nvidia had lost about 7% in the five sessions to Nov. 4. The 2,000 Dublin marchers is a larger number than the several hundred organizers expected, and no independent report confirms the Helsinki march.
An open-source-intelligence read of how likely this story is true with its real nuance, not a judgment of any outlet. It assesses the claim, weighing independent and adversarial reporting. How we label confidence.
What this means
Synthesized from: Financial Times Markets · InfoMoney · Financial Times Technology
Start a discussion in Townsquare.
More from this edition
The pressure works through three channels at once: local permitting and power-price politics, the cost of the debt financing the build-out, and supplier margins at chip, memory and storage firms. The companies most exposed are the hardware suppliers and the private credit lenders that have financed data-center construction, while hyperscalers with large cash reserves can delay projects without breaching any loan terms. Markets had partly priced slower AI spending growth, but they had not priced a Congress elected partly on opposition to data centers. Where the consensus may be wrong is in treating this as a US-only cost. The Dublin and Helsinki protests show that siting opposition has spread across rich democracies, which favors jurisdictions that court capacity, such as Britain and the Gulf states. The desk's read is that over the next two to three quarters, hyperscalers will slow US project announcements and shift new capacity toward jurisdictions that offer power and permits, while suppliers' guidance weakens. This would prove wrong if the largest buyers raise their 2027 capital spending guidance in January earnings and the new Congress does not move AI or energy-siting legislation.
What to watch
Observations to monitor, not financial advice.
Comments
1Nov 8, 5:20 AM · edited
SanDisk, Western Digital, Micron and Seagate fell 13% to 20% against 9.3% for Nvidia, which suggests investors see memory and storage suppliers as most exposed to slower data center buildouts.