Level 1 — Absolute Beginner

Thursday was a very bad day for the stock market. Share prices fell a lot. People call this a big drop.

The Nasdaq index fell 5.5 percent. The S&P 500 fell 4.4 percent. The Dow Jones lost 1,833 points. These are three important American indexes.

Technology companies fell the most. Nvidia lost 9.3 percent of its value. AMD lost 17.3 percent. Both companies make computer chips for artificial intelligence.

Investors are worried. Companies spend a lot of money on artificial intelligence. Investors ask a simple question: will this money come back? Nobody knows the answer yet.

the stock market

A place where people buy and sell parts of companies.

a share

A small part of a company that you can buy.

to fall

To go down. Prices fall when they get smaller.

an index

A number that shows how a group of companies is doing.

an investor

A person who puts money into companies to make more money.

a chip

A very small computer part inside a phone or a machine.

worried

Feeling that something bad may happen.

value

How much money something is worth.

Level 2 — Elementary

On Thursday, November 5, 2026, American stock markets had their worst day in about eighteen months. The Nasdaq Composite, which is full of technology companies, dropped 5.5 percent and closed at 24,315 points. That left the index almost 12 percent below its record high, which means it has entered what traders call a correction.

The other big indexes also suffered. The S&P 500 fell 4.4 percent to 7,353.73, its worst session since the tariff turmoil of April 2025. The Dow Jones Industrial Average lost 1,833 points, or 3.4 percent, and finished at 51,523.06.

Chipmakers took the hardest hit. Nvidia, the company that supplies most of the processors used to train artificial intelligence systems, fell 9.3 percent. Its rival AMD dropped 17.3 percent, even though AMD had reported very strong results only two days earlier, with revenue up 45 percent to 13.38 billion dollars.

Analysts gave three main reasons for the selling. First, investors are no longer sure that huge spending on AI will produce profits. Second, many technology shares look expensive compared with their earnings. Third, voters in the November 3 midterm elections supported dozens of local measures that restrict new data centres, which are the buildings AI needs to run.

a correction

A fall of about 10 percent or more from a recent high point in a market.

a record high

The highest level that something has ever reached.

a chipmaker

A company that designs or builds computer processors.

revenue

The total money a company receives from selling its products.

earnings

The profit a company makes after it pays its costs.

a data centre

A large building full of computers that store and process information.

a session

One trading day on a stock market.

a rival

A company or person that competes with another.

Level 3 — Intermediate

Thursday, November 5, 2026, will be remembered as the day the market's faith in artificial intelligence cracked. The Nasdaq Composite tumbled roughly 5.5 percent to close at 24,315, leaving it about 11.9 percent beneath its record and formally in correction territory. The S&P 500 slid 4.4 percent to 7,353.73, its ugliest session since the tariff turmoil of April 2025, while the Dow Jones Industrial Average surrendered 1,833 points, or 3.4 percent, to 51,523.06.

The damage was concentrated exactly where the AI story has been loudest. Nvidia, whose processors underpin almost every large AI model in commercial use, shed 9.3 percent. AMD cratered 17.3 percent, a punishing result for a company that had reported third-quarter revenue of 13.38 billion dollars just two days earlier, up 45 percent year on year, with data centre sales nearly doubling to 8.5 billion dollars. When a stock falls that hard on good news, it usually means investors have stopped rewarding growth and started questioning the price they paid for it.

Several threads came together at once. Japanese newspaper Nikkei calculated that the largest technology firms carried roughly 3.2 trillion dollars of off-balance-sheet AI commitments at the end of September, a figure that had grown about 90 percent in six months. Alphabet alone accounted for about 1.05 trillion dollars and Meta about 810 billion dollars. Bloomberg separately reported that Microsoft was tightening its oversight of smaller cloud partners, and shares in several of those firms dropped more than 13 percent.

Politics added the final push. In the midterm elections held on November 3, voters approved 33 of 45 local ballot measures that limit data centre development, according to Redbud Consulting. An NBC Election Day poll found that 76 percent of voters opposed data centres in their own communities. Investors who had assumed AI infrastructure could be built anywhere, quickly and cheaply, suddenly had to price in local resistance. Some analysts insisted the panic was overdone, but on Thursday almost nobody was buying that argument.

to tumble

To fall suddenly and steeply.

correction territory

The condition of a market that has dropped roughly 10 percent or more from its peak.

to underpin

To form the essential support or foundation for something.

to crater

To collapse dramatically in value.

year on year

Compared with the same period twelve months earlier.

off-balance-sheet

Describing financial obligations that do not appear directly in a company's main accounts.

oversight

Careful supervision or monitoring of something.

a ballot measure

A proposal that voters decide on directly in an election.

Level 4 — Advanced

For the better part of three years, the artificial intelligence trade functioned as a one-way bet: capital expenditure was treated as evidence of ambition rather than risk, and any company able to articulate a credible AI strategy was rewarded accordingly. On Thursday, November 5, 2026, that reflex finally broke. The Nasdaq Composite fell approximately 5.5 percent to 24,315, roughly 11.9 percent below its all-time high and squarely in correction territory. The S&P 500 surrendered 4.4 percent to close at 7,353.73, its most severe single-session decline since the tariff convulsions of April 2025, while the Dow Jones Industrial Average gave up 1,833 points, a 3.4 percent retreat to 51,523.06.

The distribution of the losses is more instructive than their magnitude. Nvidia, the indispensable supplier of the accelerators on which contemporary model training depends, declined 9.3 percent. AMD fell a remarkable 17.3 percent, notwithstanding third-quarter results published only forty-eight hours earlier that showed revenue of 13.38 billion dollars, a 45 percent annual increase, and data centre revenue of 8.5 billion dollars, up 97 percent. A stock that is punished on unambiguously strong numbers is telling you something about the multiple rather than the earnings: the market had ceased to debate whether these firms could grow and had begun to interrogate what that growth was worth.

Three structural anxieties converged. Nikkei's tally of off-balance-sheet AI commitments across the largest technology companies reached roughly 3.2 trillion dollars at the end of September, an increase of some 90 percent in six months, with Alphabet responsible for about 1.05 trillion and Meta for about 810 billion. Such obligations, by their nature, sit awkwardly outside the headline accounts that investors scrutinise most closely. Simultaneously, Bloomberg reported that Microsoft had declined an option to expand a Texas arrangement with the cloud provider Nscale and was tightening oversight of several neocloud partners; shares in Nebius and IREN fell more than 13 percent, and CoreWeave declined by double digits. The implication was unwelcome: the hyperscalers themselves may be growing more selective about where compute capacity gets built.

The final catalyst was political, and it is the one least amenable to a quarterly fix. In the midterm elections of November 3, voters approved 33 of 45 local ballot measures constraining data centre development, per Redbud Consulting, and an NBC Election Day survey registered 76 percent opposition to data centres in respondents' own communities. Compute has always been modelled as a function of capital and silicon; it is now plainly also a function of municipal consent, water rights and electricity tariffs. Google's announcement that same week of a 17 billion dollar facility in Laramie County, Wyoming, anchored by a 702 megawatt fuel cell project, illustrates both the scale of the ambition and the scale of the permitting problem. Some strategists characterised Thursday's repricing as an overreaction to a political signal rather than an economic one. That may prove correct. What is no longer arguable is that the market has begun to apply a discount rate to AI enthusiasm, and discount rates, once applied, are seldom withdrawn quickly.