Thursday Wiped More Than $2 Trillion Off the Market — and Friday's Chip Bounce Took Back a Fifth: the Halal Investor's Read (2026)
Thursday 5 November 2026 was the day the artificial intelligence trade broke, at least for a session. The Nasdaq Composite fell 5.52 percent, its worst day since April 2025, the S&P 500 fell 4.36 percent, the Dow Jones Industrial Average lost 1,833 points, and the Philadelphia Semiconductor Index fell 12.25 percent, its biggest one-day drop since March 2020. More than $2 trillion of market value was wiped out. The selling had no single cause: a Democratic sweep of both chambers of Congress in the midterm elections, fears that new data centers will be slower and costlier to build, a paused Micron memory plant, a weak Qualcomm outlook, and forced selling by funds that had bought the AI trade with borrowed money all hit the same tape on the same day. Then Friday brought a partial bounce, led by the very names that had fallen hardest: AMD rose 11.04 percent, Micron rose 5.12 percent, Nvidia rose 2.78 percent, and the Nasdaq gained 1.09 percent, helped by a jobs report that came in nearly three times stronger than expected. The bounce recovered only about a fifth of Thursday's fall, and the week still ended as the worst for the S&P 500 since April 2025. This article reads the two days the way a halal investor should read them: what the numbers actually said, why the market fell and then partly recovered, what the screen says about the companies at the centre of it, and the framework for holding through a rout without letting a headline do the deciding.
Updated November 2026 · Reviewed by the Canadian Halal Investor team
Thursday's numbers: the worst day since April 2025
The scale of Thursday is best read index by index, because the damage was concentrated where the AI trade lives. The Nasdaq Composite closed at 24,315, down 5.52 percent, its worst session since April 2025 and more than 10 percent below its October record, which is the textbook definition of a correction. The Nasdaq 100 fell 6.36 percent, its worst day since 16 March 2020. The S&P 500 fell 4.36 percent to 7,353.73, its worst day since 4 April 2025. The Dow Jones Industrial Average fell 3.44 percent, down 1,833.06 points to 51,523.06. The information technology sector fell 7.3 percent, its worst day since March 2020, and the Philadelphia Semiconductor Index plunged 12.25 percent, its biggest one-day drop since March 2020. Roughly a dozen large AI stocks lost about $1.4 trillion of value between them by the close, and Nvidia alone lost more than $440 billion of market value in a single session.
The individual moves were larger than the indexes, because the selling went straight at the suppliers of the AI buildout. Nvidia fell 9.26 percent to $181.60. AMD fell 17.26 percent, from $488.78 to $404.40, two days after reporting a record quarter. Micron slid 17.46 percent. SanDisk cratered 22.26 percent. Western Digital fell 17.35 percent. Tesla fell 11 percent. The week as a whole was worse for the memory names than the single day suggests: over the five sessions to Friday's close, Micron fell 22.01 percent, SanDisk 27.44 percent, Western Digital 27.95 percent and Seagate 28.48 percent. The selling was not only institutional. Citadel reported that individual investors recorded their most aggressive selling in the first 30 minutes of a trading session since 7 April 2025, ending 22 consecutive sessions of net buying by retail investors. A trade that had been bought on every dip for weeks finally met a day when the buyers stepped aside, and then joined the selling.
Two details keep the day in proportion. First, the market had been nervous before Thursday: the semiconductor index had been heading steadily downward since its late-June record, and the week had already carried the midterm results, a Bank of England rate hike to 4.00 percent, and a Federal Reserve that had held rates at 3.75 to 4.00 percent on 28 October while the 10-year Treasury yield sat near 4.9 percent. Second, Thursday was not a crash in the sense of a broken market. It was a repricing of one trade, the AI infrastructure trade, on a day when several separate worries arrived at once. The distinction matters, because a repricing has a cause that can be examined, and the cause is where the halal read begins.
Why it fell: three causes hit at once
There was no single cause, and any article that picks one is simplifying. The reporting converges on three forces that hit the same tape on the same day: politics, the economics of AI spending, and borrowed money. Each is worth taking separately, because each teaches a different lesson.
Politics came first. On Tuesday 3 November, Democrats won both chambers of Congress in the midterm elections, and the market spent Thursday pricing in what that means for the data-center buildout that the AI trade depends on. Hyperscalers need more data centers to keep training and serving models, and data centers need permits, power and water. Governors-elect in states where new projects are planned have pledged to slow them: in Texas, Gina Hinojosa pledged an immediate moratorium on new data centers, and in Ohio, Amy Acton backs a conditional one. UBS strategists estimated that around 30 percent of expected 2028 data-center additions could face delays. The logic is simple even where the politics are messy: if Congress and state governments slow permits, power deals or AI rules, the spend cycle that has carried Nvidia, AMD and the AI builders gets a haircut. That is the fear that hit the tape on Thursday, not a single company guide. It is also worth noting what did not happen: no law was passed, no permit was revoked and no project was cancelled on Thursday. The market priced a risk, not an event, and a priced risk can be repriced again.
The second force was the economics of AI spending itself. Investors have spent the whole earnings season asking whether companies are spending more on AI than AI earns, and Thursday added two company-specific answers that fed the doubt. Micron paused a planned $4.5 billion memory plant in Taiwan amid a labour dispute, which read as a crack in the memory supercycle that had carried the stock up 141.76 percent in 2026 before the rout. Qualcomm's weak outlook for handset chips, reported on Wednesday evening, added to the chip selloff. Oracle and Marvell fell on plans to fund AI buildouts. The pattern of the season is now familiar: Amazon fell 7 percent on a $77 billion AI spending quarter, Alphabet fell 10 percent on $51.4 billion of quarterly capital spending, and Intel plunged 25 percent on a weak outlook despite an earnings beat. The market is no longer rewarding spending as growth. It is asking for the revenue that spending is supposed to produce, and on Thursday it asked all at once.
The third force was borrowed money, and it is the one that matters most for a halal reader. Many funds had bought AI stocks with borrowed money. When prices fall, lenders demand more collateral, and anyone who cannot pay is forced to sell. That selling pushes prices lower still, which triggers more margin calls, which forces more selling. It is a cascade, and it explains why the falls were so large in a single session: the selling was not only investors changing their minds, it was investors being made to sell. The retail data points the same way. Individual investors had bought through every earlier bout of weakness this year, 22 straight sessions of net buying, and on Thursday they sold more aggressively in the first half-hour than on any day since 7 April 2025. A market built on leverage falls further and faster than the news alone would justify, because the leverage has to come out. That is not a market judgment. It is arithmetic.
Friday's bounce: chips led, the Dow did not follow
Friday brought a partial bounce, and chips led it. The Nasdaq Composite rose 1.09 percent to 24,579.42, up 264.43 points. The S&P 500 rose 0.31 percent to 7,376.86. The Dow Jones Industrial Average, though, kept sliding, down 0.46 percent, or 237.33 points, to 51,285.73, dragged by the old-economy names that did not share in the tech rebound. The Russell 2000 rose 1.0 percent. The semiconductor index rallied 4.2 percent, and the VanEck Semiconductor ETF jumped 3.62 percent. Memory stocks rallied about 10 percent intraday before closing lower than that peak, but the closes were still strong: AMD surged 11.04 percent to $449.04, clawing back a chunk of Thursday's loss. Nvidia rose 2.78 percent. Micron added 5.12 percent to $690.00. SanDisk rebounded 3.55 percent to $835.06. Intel rose 7.71 percent. Qualcomm rose 5.8 percent. Marvell rose 5.73 percent. Lam Research rose nearly 9 percent at midday after its board authorised a $25 billion share buyback. The bounce reached AI infrastructure names more broadly, with Sandisk, Marvell, Nebius and CoreWeave all among the gainers.
The jobs report helped the risk tone, and it is worth reporting precisely because it cut both ways. Employers added 165,000 jobs in October, against roughly 59,000 expected, and unemployment fell to 4.1 percent from 4.2 percent. A hot-enough jobs number usually leans hawkish for the Federal Reserve, and the rates market was already there: after the data, traders priced in about a 77 percent chance of a December rate hike, with the 10-year Treasury yield near 4.88 percent. So Friday's bounce happened on the same news that made a December hike more likely, which is the tension that runs through the whole week. The market bought the chips and sold the rate-sensitive old economy in the same session, which is why the Nasdaq rose and the Dow fell. Friday took back about a fifth of Thursday's fall. For the week, the Nasdaq fell 3.4 percent, its fourth straight weekly loss, the S&P 500 fell 2.7 percent, its worst week since April 2025, and the Dow fell 0.9 percent. The Nasdaq 100 fell 4.91 percent on the week to 26,989.99, closing 13.94 percent below its record of 6 October.
The two days together are the story of this earnings season in miniature. AMD reported a record quarter on Tuesday, with revenue of $13.377 billion, up 45 percent, data-centre revenue of $8.546 billion, up 97 percent, and a fourth-quarter guide near $17 billion. On Wednesday the stock barely moved. On Thursday it fell 17.26 percent. On Friday it rose 11.04 percent. The company did not change three times in three days. The price did, because the market's question changed three times: first about the quarter, then about the politics of data centers, then about whether Thursday had gone too far. A halal investor who reads the price as the verdict will be whipsawed by it. A halal investor who reads the screen and the business will notice that neither moved at all.
The average stock held up: the rout was narrow, not broad
One number from the week deserves more attention than it has received, because it changes what Thursday means. For the week, 261 of the S&P 500's 500 members rose, and the median member rose 0.27 percent. The index fell 2.7 percent while the average stock in it rose. That is possible only because the index is concentrated: Nvidia alone took 0.62 points off the S&P 500, and Nvidia, Micron, Tesla and Broadcom together took 1.38 points. The 500 members together lost about $1.75 trillion of market value on the week, but the losses sat in a handful of names. Eighteen of the twenty semiconductor members fell, and the memory and storage makers fell furthest: Seagate lost 28.48 percent on the week, Western Digital 27.95 percent and SanDisk 27.44 percent, while Lumentum, which makes optical parts for data centers, fell 35.94 percent, the most in the index. The power and construction names that had recovered the previous week gave it back, with Vertiv down 21.35 percent and Quanta Services down 12.86 percent.
The lesson is that an index move and a portfolio move are different things, and the difference is concentration. A halal portfolio built from the passing names on this site's screener hub is not the index, and it should not be read as the index. If it holds the chip names, it felt Thursday in full. If it holds the 261 names that rose, it did not. The point is not that one mix is better than the other. The point is that "the market fell 4.36 percent" is a statement about a weighted average dominated by a few very large companies, and a halal investor's first question after a rout should be which names they actually hold, not what the index did. The portfolio builder on this site builds only from passing screeners, which is the discipline that keeps the question answerable: if every holding has passed the screen, the rout is a price event, not a compliance event.
The screen results: every featured chip name still passes
None of Thursday's moves changed a single screen result, and that is the central fact of this article. The screen is measured on audited annual figures, not on prices, and no price, however large, changes a debt ratio or an income ratio. Every chip name at the centre of the rout still passes this site's October 2026 screen. Nvidia passes, with interest-bearing debt of 0.15 percent of market capitalisation and non-compliant income of 1.07 percent of revenue, both far below their ceilings. AMD passes provisionally, with debt of 0.32 percent of market capitalisation; its non-compliant income figure was undisclosed, which is why the pass stays provisional. Micron passes, with debt of 0.48 percent and non-compliant income of 0.64 percent. SanDisk passes, with debt of 0.08 percent and non-compliant income of 0.35 percent. Western Digital passes, with debt of 0.63 percent and non-compliant income of 0.39 percent. Seagate passes, with debt of 1.72 percent and non-compliant income of 0.25 percent. Intel passes, with debt of 7.4 percent and non-compliant income of 1.91 percent. Qualcomm passes, with debt of 7.93 percent and non-compliant income of 2.19 percent. Lam Research passes, with debt of 0.88 percent and non-compliant income of 0.84 percent. Marvell passes, with debt of 1.94 percent and non-compliant income of 1.54 percent. The full method, gate by gate, is set out in the screening methodology.
The numbers are worth reading as a group, because they show how wide the margins are. The debt ceiling is roughly 33 percent, with a reported disagreement between providers over whether it is 30 percent or about 33 percent, and every featured name sits below 8 percent, so the disagreement does not matter for any of them. The non-compliant income ceiling is 5 percent, and every featured name sits below 2.2 percent, so the disagreement does not matter there either. These are not close calls. They are comfortable passes, and a 17 percent or 22 percent fall in the share price does not move them, because the ratios are measured against market capitalisation and revenue, and the market capitalisation of a company that falls 17 percent is still far above the debt it carries. The one mechanical effect of a falling price is that the debt ratio rises, because the denominator shrinks, but at 0.08 percent for SanDisk or 0.48 percent for Micron, the ratio would have to rise by a factor of forty or more to reach the ceiling. That is not a rout. That is a different company.
The real screen thread from the week is not the rout. It is rates. Friday's jobs report pushed December hike odds to about 77 percent, and higher rates push interest income up for cash-rich companies, which is exactly the income the third gate measures. The names closest to the ceiling are not the chip names at all. Regeneron sits at 4.998 percent, a rounding error below the 5 percent ceiling, and Arista sits at 4.26 percent. On the failing side, Cloudflare fails at 6.05 percent and Palantir at 5.12 percent. A December hike does not change any of those ratios today, because the screen is measured on audited annual figures, but it raises the yield on the cash piles that produce the interest income, so the next refresh is where the pressure shows. That is the thread to watch: not whether the chips fell, but whether the cash-rich names near the line earn more interest while they wait.
The BDS flag and the moral screen, reported as context
One flag belongs in the article, reported as context and not as a screen change. Intel is a BDS boycott target: the company is named by the BDS movement as a priority boycott target, and the movement's guide notes that Intel is one of Israel's largest private employers and investors. That flag is informational only. It does not change the Shariah screening result, and it is not a religious ruling. Readers who follow the BDS movement's lists will weigh it for themselves; readers who do not will note it and move on. The other featured names, Nvidia, AMD, Micron, SanDisk, Western Digital, Seagate, Qualcomm, Lam Research and Marvell, do not appear on the BDS movement's published boycott, divestment or pressure lists. The site's moral screen, which checks documented business ties to the Israeli and United States militaries and to weapons manufacturing, publishes no findings for any of the featured chip names. The screen results stand on the financial gates alone: every featured name passes, and nothing else.
How a halal investor should approach a rout
A rout triggers a specific set of instincts, and they are worth naming before acting on any of them. The first is to sell everything, on the theory that a 5.52 percent day is the start of something worse. The second is to buy the dip, on the theory that a 17 percent fall in a good company is a bargain. The third is to do nothing and feel guilty about it, on the theory that doing nothing is not a decision. None of the three is a framework. A framework is a set of questions, answered in order, before any order is placed, and for a rout the questions are these.
First, screen before you react. Every featured chip name passes the screen, and that fact did not change on Thursday or Friday. A halal investor who sells a passing name because it fell 17 percent has let the price do the screening, and a halal investor who buys a failing name because it fell 22 percent has done the same thing in the other direction. The screen is the first question, and the price is not an input to it. Second, remember that you do not use margin. The forced-selling cascade that deepened Thursday is a leveraged phenomenon: funds that bought with borrowed money were made to sell when prices fell. Margin trading is not halal, and a halal portfolio carries no margin debt, which means no lender can liquidate it on a bad day. That is not a small protection. On Thursday it was the difference between choosing to sell and being made to sell, and the investors who were made to sell sold at the worst prices of the day. Third, separate the business from the price. AMD's quarter did not change between Tuesday and Friday: record revenue of $13.377 billion, data-centre revenue up 97 percent, a fourth-quarter guide near $17 billion. The stock fell 17.26 percent and then rose 11.04 percent. The business did not move twice. The price did, because the market's question changed. A company can be well run and fall 17 percent, just as a company can be poorly run and rise. The screen is not a quality rating, and the price is not a business judgment. They live on different axes.
Fourth, do not trade a headline. A 5.52 percent day is a repricing, not a trend, and the week it happened in was still a down week for the market after Friday's bounce. The disciplined position is to note the ratios as they stand, note that a single earnings beat or a single rout changes none of them, and recheck them at the refresh. Fifth, watch the ratios, not the ticker. The thread that matters from this week is the December hike, because higher rates raise the interest income of cash-rich companies, and the names near the 5 percent ceiling are the ones to recheck first. Sixth, remember the currency. The chip names are listed on the Nasdaq in United States dollars, so a Canadian buyer pays the broker's conversion spread on the way in and on the way out, and the screen result is the same in either currency. The framework is the position: screen first, no margin, separate the business from the price, do not trade a headline, watch the ratios, and wait for the refresh.
What to watch next
The next number that matters for the market is the October consumer price data, due Tuesday 10 November, which is the next input to the December Federal Reserve meeting on 8 to 9 December. A hot print strengthens the case for the hike the jobs report already made likely; a soft print weakens it. The Bank of Canada decides on 9 December, the same week. Nvidia's earnings report is the next big test for the AI trade itself: after a week in which the market punished spending and rewarded only the cleanest beats, the bellwether's numbers and guide will say whether the buildout is still being funded at the pace the chip trade assumes.
Closer to the politics that drove Thursday, watch whether the data-center moratorium pledges in Texas and Ohio become policy, and whether other states follow. The UBS estimate that around 30 percent of expected 2028 data-center additions could face delays is an estimate of a risk, not a count of cancelled projects, and the next few weeks will show whether it stays a risk or becomes a fact. For readers who hold any of the near-ceiling names, the practical next step is the same as it was after the jobs report: note the ratios as they stand, note that a rout changes none of them, and recheck them at the refresh. For readers who hold the chip names, the next step is the screen-first habit applied to whatever you do hold, because the next 5 percent day will not be Thursday's, and the framework that worked this morning is the one that will work then. Run it through the screening methodology, check the companies on the screener hub, and build the rest of the decision from there. A rout is a price reading. The screen is the house. Read the price, and keep the house in order.
FAQ
What happened on Thursday 5 November 2026?
The Nasdaq Composite fell 5.52 percent to 24,315, its worst day since April 2025, and more than 10 percent below its October record, which is the textbook definition of a correction. The S and P 500 fell 4.36 percent to 7,353.73, the Dow Jones Industrial Average fell 3.44 percent, down 1,833 points to 51,523.06, and the Philadelphia Semiconductor Index fell 12.25 percent, its biggest one-day drop since March 2020. Nvidia fell 9.26 percent, AMD fell 17.26 percent, Micron fell 17.46 percent, SanDisk fell 22.26 percent and Western Digital fell 17.35 percent, and more than 2 trillion dollars of market value was wiped out.
Why did the market fall so sharply?
There was no single cause. Three forces hit the same tape on the same day. First, politics: Democrats won both chambers of Congress in the midterm elections, and the market priced in slower and costlier data-center buildouts, with governors-elect in Texas and Ohio pledging moratoriums and UBS estimating that around 30 percent of expected 2028 data-center additions could face delays. Second, the economics of AI spending: Micron paused a planned 4.5 billion dollar memory plant in Taiwan, Qualcomm gave a weak handset outlook, and investors asked whether AI spending is producing enough revenue. Third, borrowed money: funds that had bought AI stocks with borrowed money were forced to sell as prices fell, which pushed prices lower still.
What happened on Friday?
Friday brought a partial bounce led by the same chip names that had fallen hardest. The Nasdaq Composite rose 1.09 percent to 24,579.42, the S and P 500 rose 0.31 percent to 7,376.86, and the Dow Jones Industrial Average fell 0.46 percent to 51,285.73. AMD surged 11.04 percent to 449.04 dollars, Micron added 5.12 percent to 690 dollars, Nvidia rose 2.78 percent and SanDisk rebounded 3.55 percent. The jobs report helped: employers added 165,000 jobs in October against roughly 59,000 expected, and unemployment fell to 4.1 percent. Friday took back about a fifth of Thursday's fall, and the week still ended down, with the Nasdaq off 3.4 percent and the S and P 500 off 2.7 percent.
Did any of the chip names fail the halal screen?
No. Every featured chip name still passes this site's October 2026 screen, because the screen is measured on audited annual figures, not on prices. Nvidia passes with debt of 0.15 percent of market capitalisation and non-compliant income of 1.07 percent of revenue. Micron passes at 0.48 percent and 0.64 percent, SanDisk at 0.08 percent and 0.35 percent, Western Digital at 0.63 percent and 0.39 percent, Seagate at 1.72 percent and 0.25 percent, Intel at 7.4 percent and 1.91 percent, Qualcomm at 7.93 percent and 2.19 percent, Lam Research at 0.88 percent and 0.84 percent, and Marvell at 1.94 percent and 1.54 percent. AMD passes provisionally at 0.32 percent debt, with its non-compliant income figure undisclosed. A 17 percent fall does not make a passing company fail.
Why does the rout matter for the screen at all?
The rout itself does not change any screen result, but the rates backdrop behind it does matter. Friday's jobs report pushed the odds of a December Federal Reserve rate hike to about 77 percent, and higher rates push interest income up for cash-rich companies, which is exactly the income the screen's third gate measures. The names closest to the 5 percent ceiling are not the chip names: Regeneron sits at 4.998 percent and Arista at 4.26 percent. The other lesson is about leverage. Margin trading is not halal, and a halal portfolio carries no margin debt, so no lender can liquidate it on a bad day. The forced selling that deepened Thursday was a leveraged phenomenon, and a halal investor is structurally outside it.
Is anything in this article a fatwa or a buy or sell call?
No. Everything here is facts and screening methodology only, with no fatwas and no personalised buy or sell calls. Rulings on your own situation belong with a qualified scholar, and investment decisions belong with advice suited to your circumstances. Not financial advice. Facts and screening methodology only, no fatwas.