Wednesday felt like relief. After the Democratic midterm sweep, US stocks rose broadly — the Dow about 3%, the S&P 500 about 1.58%, and the Nasdaq about 1.44%. Thursday took most of that back. On 5 November 2026, US stocks tumbled sharply, led by AI and semiconductor stocks, in the market’s worst day since the tariff announcements of April 2025, according to AP. And this time the closing numbers are confirmed: the Nasdaq finished at 24,712.60, down 3.9% on the day and more than 10% below its record close — a correction, by the standard definition.
The confirmed close — and why intraday numbers differ
The Nasdaq Composite fell nearly 4% to 24,712.60 on 5 November, more than 10% below its record closing high of 27,599.89 on 6 October — a correction by the standard 10% definition, and the Nasdaq’s second correction this year, according to Reuters. Business Today reported the same closing picture: Nasdaq down 3.9%, the S&P 500 down 2.8%, and the Dow down 1,153 points, or 2.2%.
Figures captured during the day looked different, and that is worth explaining rather than hiding. AP, at around 11 a.m. ET, showed the S&P 500 down 3.2%, the Nasdaq down 5%, and the Dow down 1,146 points (2.1%). The National, at 10:43 a.m. ET, reported the Nasdaq down 4.17%, the S&P down 2.68%, and the Dow down 880 points (1.65%). Those were intraday snapshots, taken while trading was still under way. The closing figures above are the settled end-of-day numbers; the intraday figures should be read as ranges showing how deep the fall looked at different points in the session, not as competing closes.
What is clear from every source is the direction and the scale: a broad, sharp sell-off, concentrated in the very stocks that had led the market higher all year. Thursday’s slide also erased Wednesday’s post-midterms gains — the Dow had surged more than 1,500 points on Wednesday.
What “correction” actually means
The word sounds technical, but the definition is simple: a correction is a decline of at least 10% from a recent high. The Nasdaq has now fallen more than 10% below its record close of 27,599.89 on 6 October, putting it in correction territory. The Nasdaq 100 had already crossed that line, falling more than 10% from its 6 October peak by 3 November — its third technical correction of the year, on that measure.
A correction is not a crash, and it is not a prediction. It tells you how far prices have come down from the top. It does not tell you where they go next. The Nasdaq shed more than $2.5 trillion in market value in October alone, according to Reuters, before Thursday’s fall — a reminder that this correction was building before the midterms, not created by one day’s trading.
The stocks that fell hardest
The pain was concentrated in AI and chips. Nvidia and Broadcom fell about 6% each, and Oracle fell about 6%. Qualcomm fell 11% despite beating earnings expectations, and Tesla fell 9%. CoreWeave fell about 9%, Nebius about 13% and Iren about 14%, according to Business Today. Earlier in the day, reports captured at different times put some moves in ranges — Nvidia down roughly 6% to 8%, Broadcom roughly 6% to 8%, Oracle 6% to 7.6%, Tesla 9% to 9.5%, and the chip ETF SOXX down 8% — because prices were still moving when those reports were filed. The closing picture is the one in the signals above; the ranges show how volatile the session was.
The fall was not confined to the United States. South Korea’s Kospi fell 5.3% in the earlier session, and in Asian trading after the US close, the Kospi fell a further 4.63% to 5,420.77, the Nikkei fell 1.42% to 61,306.77 and Taiwan fell 2.54%, while Hong Kong’s Hang Seng rose 3.09% and Australia’s ASX rose 2.36%. India’s Nifty was up 1.70%. Yet it was not everything falling at once in the US either: roughly 2 in 5 S&P 500 stocks still rose, according to AP — the sell-off was concentrated in AI and tech, not a blanket rejection of every company.
The two worries driving the sell-off
Why did investors sell? No single cause has been proven, but three explanations were widely reported on the day. The first is spending versus profit. Companies have poured enormous money into AI — data centres, chips, power, infrastructure. JPMorgan estimates global AI-related spending at around $900 billion this year, potentially $5.5 trillion cumulatively by 2030, as reported in a Reuters analysis, and big tech has borrowed heavily to fund it. The worry, as reported, is whether that spending turns into profit quickly enough. Results and outlooks from Alphabet, Amazon and Intel disappointed investors, according to the same analysis, and Alphabet and Meta earnings last month failed to reassure on AI-spending returns.
The second worry is regulation. With a Democratic-controlled Congress after the midterms, investors expect greater scrutiny of the AI data-centre buildout, according to Reuters. No new rules have been written — the market is pricing in the possibility of a tougher environment.
The third is Qualcomm’s weak smartphone-chip outlook. Its fiscal Q4 handset revenue fell 26% year over year to about $5.14 billion, below the $5.25 billion analysts expected — a signal that can make investors question demand across the whole chip chain. These are reported explanations, not established causation. Markets often fall for a mix of reasons.
Why one day is not the whole story
Thursday did not come out of nowhere. The Nasdaq 100 had already fallen about 3.5% around 30 October, and the S&P technology sector about 6.8% in October. The AI trade had been wobbling before the midterms, rallied on Wednesday, then fell again on Thursday — a market that rises 1.44% one day and falls roughly 4% the next has not made up its mind.
A correction is also not a verdict on whether AI itself works. It is a measure of price — how far the Nasdaq has fallen from its October peak. The underlying question is simpler: how fast does hundreds of billions of dollars of AI spending turn into real earnings? That question will not be answered by one day’s trading, in either direction.
What to watch next
No one can say where the market goes from here, and this article will not try. But there are a few things investors will be watching closely. First, earnings — especially from the big AI spenders and the chipmakers, and whether their outlooks calm or deepen the spending-versus-profit worry. Second, data-centre spending: whether companies slow down, keep building, or explain more clearly how the investment pays off. Third, Congress: what the new Democratic-controlled Congress actually says and does about AI and the data-centre buildout, rather than what the market fears it might do. US October nonfarm payrolls are also due on Friday 6 November, a scheduled data point that could move markets again.
This is not investment advice. It is a record of a sharp, confusing day — when the stocks that had carried Wall Street higher suddenly carried it lower, a day after a rally that had suggested the opposite.
What is the Nasdaq correction?
The Nasdaq closed at 24,712.60 on 5 November 2026, down 3.9% on the day and more than 10% below its record close of 27,599.89 on 6 October — a correction by the standard 10% definition.
Which stocks fell hardest?
AI and chip stocks led the fall: Nvidia and Broadcom about 6% each, Qualcomm 11%, Oracle about 6% and Tesla 9%, on the confirmed closing figures.
Is this investment advice?
No. This article records what happened on 5 November and what was reported as driving it. It is not investment advice.