Toronto · Independent Shariah screening · Updated daily
Market news · Researched from public market reporting · Content-only (no affiliate programs)

Anthropic's IPO Is Oversubscribed Several Times Over — at a $1.4 Trillion Valuation: the Halal Investor's Read (2026)

The largest initial public offering in history is oversubscribed before it has even priced, and the market is already arguing about what that means. Anthropic, the company behind the Claude chatbot, filed an amended prospectus on Tuesday 10 November 2026 offering 825 million shares at $70 to $78, and Bloomberg reported on Wednesday that the offering has drawn orders for multiple times the shares on offer, with the book covered on the first day of the roadshow and institutional demand running ahead of SpaceX at the same stage. The Financial Times, citing five people familiar with the matter, reported that the bankers are already weighing pricing the deal above the filed range, which would lift the valuation to $1.5 trillion or more and push total fundraising past $100 billion. The numbers behind the demand are the reason the deal matters to everyone else in the market: revenue of $33.2 billion in the nine months to 30 September, up 1,206 percent from a year earlier, a revenue run-rate of about $81 billion at the end of October, and operating losses that narrowed to $1.3 billion from $5.2 billion over the same period. The deal also matters to the companies a halal investor already owns or watches, because Google and Nvidia are each committing about $10 billion of a $26 billion concurrent private placement, and Microsoft and TSMC are buying in too. This article lays out what the offering actually is, why the valuation has two different numbers and both are right, who is buying and why, what the business behind the book really earns, why an oversubscribed book has failed investors before, what a weak debut would mean for the whole AI trade, and the part that matters to a halal investor: why there is no screen result for Anthropic yet, what the screen will ask once the shares trade, and how to think about an IPO that the market has already decided is a must-own before a single share has changed hands. This article was written on Wednesday 11 November 2026, before the pricing on 18 November and the first trade on 19 November. The numbers below are the offering as filed and as reported, not the result.

Updated November 2026 · Reviewed by the Canadian Halal Investor team

The deal: 825 million shares at $70 to $78, and a book covered on day one

Anthropic is a public benefit corporation that builds Claude, one of the two frontier AI models that define the current market, and its roadshow opened on Tuesday 10 November with the filing of Amendment No. 2 to its Form S-1. The offering is 825 million Class A shares at $70 to $78 each, which raises up to about $64.4 billion at the top of the range. The underwriters have an option to buy 123.75 million more shares, which would add about $9.7 billion at $78, and a concurrent private placement of non-voting stock to seven chip and technology companies adds $26 billion at the same price. Together, the public offering and the private placement could raise as much as $90.4 billion, which would make it the largest offering of all time, surpassing the SpaceX listing. The shares are expected to price on Wednesday 18 November and to begin trading on the Nasdaq on Thursday 19 November under the ticker ANT, with Goldman Sachs and Morgan Stanley leading the deal. About 20 percent of the offering is targeted at retail investors, and the chief executive, Dario Amodei, is scheduled to host a livestreamed investor event with JPMorgan chief executive Jamie Dimon on Friday. The original registration statement was filed on 26 October, and the earliest the prospectus allowed pricing was 13 November, so the current timetable is a choice, not a constraint: the bankers are taking an extra week because the demand is there to be harvested, not because the paperwork requires it.

The demand is the story of the week, and it is worth reporting precisely, because the words differ between accounts and the difference matters. Bloomberg, citing people familiar with the matter, reported that the offering has drawn orders for multiple times the shares on offer, driven by anchor-size indications of interest from large institutions. CNBC reported on Tuesday that the book was already covered at the $70 to $78 range on an allocable basis, and that institutional demand was running ahead of SpaceX at the same stage of its own process. Other accounts describe the book more cautiously, as covered rather than oversubscribed several times over, and the discrepancy is not a contradiction: indications of interest are not binding orders, allocations have not been made, and the final price will not be known until the evening of 18 November. What is not in dispute is the direction. A book that is covered on the first day of marketing, with anchor-size indications from the largest institutions, is a book the bankers can price up, and the Financial Times reported that they intend to try. The demand is also concentrated, which is the other detail worth naming. Anchor-size indications come from a small number of very large buyers, and a book built on a small number of very large buyers is a book that can change its mind quickly if the price moves too far or the market turns. The oversubscription is real, and it is not a guarantee of anything, and the next section explains why the valuation number itself is the first place that becomes visible.

There is a second fact about the deal that frames everything else, and it is the size of the company being sold. Anthropic's last private round valued it at $965 billion, so the $70 to $78 range already represents a step up from the private market, and the bankers' push to price above the range would be a second step up in the same week. The company has never been public, has never reported a quarterly earnings call, and has never had its shares trade on an open market, so the $70 to $78 range is the first price the public market has ever been asked to put on the company, and the book is the market's first answer. That answer is enthusiastic, and enthusiasm is not a valuation method. The valuation section that follows is the attempt to say what the price actually buys, and the business section after that is the attempt to say what the company actually earns, because the two have to be held together or the enthusiasm has nothing to hold on to.

The valuation: $1.4 trillion, or $1.02 trillion — both numbers are right

Every headline about the Anthropic offering carries a different valuation, and the confusion is not sloppy reporting. It is two different ways of counting the same company, and both numbers are correct on their own basis. At $78, the top of the filed range, Anthropic is worth about $1.02 trillion based on the shares that will actually be outstanding after the offering and the private placement, which is the figure Reuters used. On a fully diluted basis, counting every share including stock options, restricted stock units and shares reserved for employee plans, the same $78 a share comes to about $1.4 trillion, which is the figure Bloomberg and CNBC report. The Financial Times' $1.5 trillion, the number the bankers are said to be weighing, uses the same fully diluted basis, and it implies about $83 a share, roughly 7 percent above the top of the range, or about $1.09 trillion on shares outstanding. The filing itself implies a fully diluted range of about $1.26 trillion to $1.41 trillion, so the $1.4 trillion in the headlines is the top of the filed range on the fully diluted count, and the $1.02 trillion is the top of the filed range on the shares-outstanding count, and the difference between them is about 18 billion shares of options, restricted units and reserved employee stock. A reader who does not know which count a headline is using cannot tell whether the valuation went up or down, and a reader who does know can see that the bankers' push is a push for about $5 more a share on the fully diluted count, not a new company.

The third-party estimates around the deal are higher still, and they should be read as estimates, not as the deal. Morningstar put the estimated fair equity value at $1.7 trillion, and a pre-IPO contract on the prediction market Hyperliquid implied a valuation of about $1.8 trillion on Wednesday morning. Both are above the filed range, and both are the kind of number that appears when demand is strong and the price has not been set yet. The gap between the filed range and the outside estimates is the gap the bankers are trying to close by pricing above $78, and the gap is also the risk: a deal priced at the top of the outside estimates leaves no room for the first trade to go up, and a deal priced above the outside estimates leaves room for it to go down. The history of the valuation in a single week makes the point. The roadshow began with talk of a $2 trillion fair value, which slid to $1.5 trillion, then to $1.2 trillion, before the filing set the range that implies $1.26 trillion to $1.41 trillion fully diluted, and now the bankers are pushing back toward $1.5 trillion and beyond. A valuation that moves $800 billion in a week is not a valuation. It is a negotiation, and the negotiation is happening in public, with the bankers on one side and the buyers on the other, and the final price on 18 November will be the number both sides can live with.

For a halal investor the valuation question is not whether $1.4 trillion is too much or too little. It is what the valuation is made of, and the answer is that it is made of growth that has not yet turned into profit, which is the next section's subject. A valuation of $1.4 trillion fully diluted, against revenue of $33.2 billion in nine months and an operating loss of $1.3 billion over the same period, is a price on the future, not on the present, and the future being priced is a revenue run-rate of about $81 billion that is growing by about $10 billion a month. The price is the market saying it believes the run-rate will keep growing and the losses will keep narrowing, and the belief is the thing being bought. The screen does not measure belief. It measures the business, the debt and the interest income, and the valuation does not enter any of the three gates. A company can be worth $1.4 trillion and pass, or worth $1.4 trillion and fail, and the price alone says nothing about which one it is. The valuation matters to the screen in exactly one way, and it is indirect: the market capitalisation is the denominator of the debt ratio, so a higher price makes the debt ratio smaller and a lower price makes it bigger, and the price on 19 November will set the denominator for the first screen that can be run. That is the whole connection, and it is worth stating plainly, because the headlines will treat the valuation as the story and the screen will treat it as a denominator.

The buyers: Google and Nvidia are each putting in about $10 billion

The $26 billion private placement is the part of the deal that tells a halal investor the most, because it names the buyers and their reasons, and the buyers are the companies whose own screens are already known. Seven strategic investors will buy $26 billion of non-voting stock at the IPO price, including Alphabet, Nvidia, Microsoft and TSMC, and Google and Nvidia are each committing about $10 billion of the total. The placement is concurrent with the offering, at the same price, which means the strategic buyers are paying the same price as the public, and it is a placement, not a secondary sale, which means the money goes to the company. The strategic buyers are also the company's customers, suppliers and landlords in different combinations: Google is Anthropic's largest cloud provider and the company raised its minimum Google Cloud commitment to about $121.3 billion earlier this month, with Google also reported to be backstopping about $19 billion of Anthropic's TeraWulf data-centre leases; Nvidia supplies the chips that train and run Claude; Microsoft is both a customer and a competitor through its own AI business; TSMC makes the chips that Nvidia designs. A private placement to the companies that supply the business is a vote of confidence, and it is also a circular one, because the suppliers are buying the customer and the customer is buying from the suppliers, and the circle is the AI trade in miniature.

For a halal investor the placement matters because the three named buyers that this site has screened all pass. Alphabet passes this site's October 2026 screen, with debt at 1.23 percent of market capitalisation and non-compliant income at 1.08 percent of revenue, and Nvidia passes with debt of 0.15 percent and non-compliant income of 1.07 percent, even after falling 5.5 percent on Tuesday to $168.89 on the board probe into chief executive Jensen Huang, and Microsoft passes with debt of 1.06 percent and interest and dividend income of 0.99 percent. The placement does not change any of those results, and it does not transfer any of them to Anthropic, because a screen is a property of a company, not of its shareholders, and Anthropic's own screen has not been run yet. What the placement does say is that the three largest buyers of the deal are companies whose own balance sheets are clean on the debt and income gates, and that they are willing to put $10 billion each into a company that has never been public, at a price they have not yet seen, which is a statement about how much they value the customer relationship and the compute demand that comes with it. Google's $10 billion is also the largest single commitment to a company that has just raised its minimum Google Cloud commitment to about $121.3 billion, which is a customer that is also an owner, and the two roles will be hard to separate in the filings that follow.

There is a second way to read the placement, and it is about who is not buying. The seven strategic investors are all companies, not funds, and the placement is $26 billion of a deal that could total $90.4 billion, which means the majority of the money is coming from the public offering and the over-allotment, not from the strategics. The public book, which Bloomberg reports as oversubscribed several times over, is the money that will set the price, and the strategic placement is the money that will anchor it. The distinction matters because the strategic buyers have reasons to buy that the public does not share: they are buying a customer, a supplier relationship, and a claim on future compute demand, and the public is buying a share. A halal investor who reads the placement as an endorsement should read it as an endorsement by the suppliers, not by the market, and should remember that the suppliers' endorsement is worth exactly what their $10 billion buys them, which is a stake in a customer they already depend on. The placement is a fact about the deal, and it is not a screen result, and the two should not be confused, because the confusion is how an IPO becomes a must-own before a single share has traded.

The business behind the book: $33.2 billion of revenue, and still loss-making

The prospectus gives the business behind the book, and it is a business that is growing faster than almost any company that has ever gone public, and still losing money. Revenue was $33.2 billion in the nine months to 30 September 2026, up 1,206 percent from a year earlier, and the revenue run-rate was about $81 billion at the end of October, up $10 billion in a single month. Operating losses narrowed to $1.3 billion from $5.2 billion over the same nine-month period, which is a real improvement, and it is also a loss: the company is spending more than it earns, and the gap is being closed by growth, not by profit. The growth is the product of a single fact, which is that Claude, the company's chatbot and the model behind its application programming interface, is being bought by businesses and consumers at a pace that the company's own infrastructure can barely keep up with, and the run-rate rising $10 billion in a month is the measure of that pace. A run-rate is not revenue, and it is not a forecast. It is the revenue the company would earn in a year if the last month's pace held for twelve months, and the pace has held for a year, which is why the run-rate is the number the bankers are selling, and why the $33.2 billion of actual nine-month revenue is the number that should be read beside it.

The shape of the business is the same shape as the rest of the AI trade, and it is the shape that CoreWeave's print on Tuesday night made visible. CoreWeave reported revenue of $3.66 billion, up 168 percent, and a GAAP net loss that widened to $709 million, with capital spending of $14.5 billion in the quarter and debt of almost $47 billion, because it is building the data centres that companies like Anthropic rent. Anthropic is the customer at the other end of that trade: it rents the computing power, trains and runs the models, and sells the output, and its costs are the computing bills that CoreWeave's revenue is made of. The two companies are the two halves of the same transaction, and both are growing revenue faster than they are growing profit, and both are funded by the belief that the growth will eventually cover the cost. The difference is that CoreWeave's costs are visible as debt on a balance sheet, while Anthropic's costs are visible as computing commitments, including the roughly $121.3 billion minimum Google Cloud commitment and the TeraWulf leases that Google is reported to be backstopping, and the commitments are not debt in the accounting sense, but they are obligations in the economic sense, and the screen will have to decide how to treat them when the first post-listing filings arrive. That decision has not been made, and it cannot be made from the prospectus summary alone, which is why the screen section below refuses to call it.

There is a detail in the business that deserves attention because it explains why the losses are narrowing even as the spending grows. The operating loss fell from $5.2 billion to $1.3 billion while revenue rose 1,206 percent, which means the company is earning more per dollar of revenue than it did a year ago, and the improvement is the scale of the model business: once a model is trained, serving the next customer costs far less than serving the first, and the margin on the next dollar of revenue is higher than the margin on the last one. That is the same logic that Cloudflare's 39 percent revenue beat and AMD and Arista's AI hardware beats showed in their own quarters, and it is the reason the market is willing to pay $1.4 trillion for a company that is still loss-making. The logic is sound, and it is not a guarantee, because the margin depends on the cost of computing staying where it is, and the cost of computing depends on the price of the chips and the power, and both are rising, and the company's own commitments are the evidence. A business that grows revenue 1,206 percent and narrows its loss by three quarters is a business that is working, and a business that is still losing $1.3 billion in nine months is a business that has not yet proved it can pay for itself, and both facts are true at the same time, and the IPO is asking the public to pay for the first fact while accepting the second.

The warning in the fine print: oversubscribed does not mean up

An oversubscribed book is the most persuasive headline in the IPO market, and it is also the least reliable predictor of what happens next, and the record is worth stating plainly because the Anthropic book is being reported in exactly the terms that have misled investors before. Klarna and StubHub both had heavily oversubscribed initial public offerings, and both traded below their offering prices, which is the outcome the oversubscription was supposed to make impossible. The mechanism is not mysterious. An indication of interest is not a binding order, and an order is not a holding, and a book that is covered several times over on the first day of the roadshow is a book that can be repriced, reallocated, or withdrawn before the evening of 18 November, and the shares that are allocated on the 18th can be sold on the 19th, which is the first trade, not the last. The oversubscription measures demand at a price that has not been set yet, and the demand at $78 is not the same as the demand at $83, which is the price the bankers are said to be weighing, and the demand at $83 is not the same as the demand at the first trade, which is the price the market sets. A book that is oversubscribed at $78 and priced at $83 has already spent some of its demand, and a book that is priced at $83 and opens at $78 has spent all of it, and the difference between the two is the first trade.

The warning is also in the wording, and the wording differs between accounts for a reason. Some reports describe the book as oversubscribed several times over, others describe it only as covered, and the difference is the difference between a multiple and a threshold: covered means the orders equal the shares on offer, and oversubscribed several times over means the orders are a multiple of the shares on offer, and the two are not the same claim. The sources are unnamed in both cases, and the allocations have not been made, and the final price has not been set, so the honest reading is that the book is strong, the exact multiple is not public, and the strength is a fact about the roadshow, not about the first trade. The market's own history makes the same point from the other side: the worst outcome for the AI trade, as Jack Selby of Copper Sky and the Thiel family office put it this week, would be an underperforming Anthropic debut, and the warning is not that the debut will underperform, but that the debut is the test, and the test has not happened yet. A halal investor who buys the headline that the book is oversubscribed has bought a fact about Tuesday, and the fact about Thursday is the price, and the two are different facts.

There is a useful comparison inside this site's own run of articles, and it is the pattern of the beat that falls. Apple and Palantir both beat earnings on 2 November and both fell, Qualcomm beat and fell 10 percent on a handset warning, and Thursday 5 November wiped more than $2 trillion off the market before Friday's bounce took back about a fifth of it. The market of the last two weeks has repeatedly rewarded the headline and then sold the stock, and the IPO market is the same market with a longer timetable: the roadshow is the headline, the pricing is the earnings call, and the first trade is the reaction, and the reaction has been negative as often as it has been positive. The oversubscription is a reason to watch the pricing, and it is not a reason to buy the first trade, and the framework section below says so in the form of a checklist, because a checklist is harder to ignore than a warning.

The bubble question: a weak debut could mark the top

The Anthropic offering is not just a company going public. It is the test of the AI trade itself, and the market knows it, which is why the warnings about it are phrased as warnings about the whole market rather than about a single stock. Jack Selby, who runs the venture capital firm Copper Sky and is a managing partner in Peter Thiel's family office, warned this week that an underperforming Anthropic IPO could mark the top of the AI bubble, and the warning has been repeated across the market because it names the mechanism: the AI trade is a chain of companies that buy from each other, and Anthropic is the customer at the end of the chain, and if the customer cannot command a price the market believes in, then the suppliers' prices have to be rethought too. The chain is visible in the placement itself, with Google and Nvidia each putting in about $10 billion, and in the commitments, with Google Cloud at about $121.3 billion and the TeraWulf leases at about $19 billion, and in the earnings, with CoreWeave's $132.5 billion backlog and Nvidia's report on 17 November still to come. A weak debut would not break the chain, but it would reprice it, and the repricing would start with the companies whose prices depend most on AI demand, which is all of them, and the market's memory of Thursday 5 November, when the Nasdaq fell 5.52 percent and the semiconductor index fell 12.25 percent, is the measure of how fast a repricing can happen.

The bubble question has two sides, and both are visible in Wednesday's tape. On one side, the AI trade rebounded hard on the CoreWeave print, with memory stocks leading: Micron rose about 13.8 percent to about $824.85 by late morning, SanDisk rose about 17 percent to about $1,006.55, and Western Digital rose about 9.5 percent, while the semiconductor index rose with them and Nvidia bounced about 1.75 percent to $171.75 after falling 5.5 percent on Tuesday, even as Baird cut its rating to Neutral and slashed its price target from $500 to $200, citing uncertainty around government action and congressional investigations, and Morgan Stanley removed Nvidia from its top picks list. On the other side, the same tape is pricing a December rate hike as near-certain after Tuesday's hot CPI print, with headline prices up 0.4 percent against 0.3 percent expected and core up 0.3 percent for a third straight month, and the 10-year Treasury yield at about 5 percent, which is the cost of money for every company in the chain. The two sides are the same argument the market has been having for a week: whether AI demand is worth the price being paid to serve it, and whether the price of money will rise faster than the revenue. The Anthropic debut on 19 November is the next data point in that argument, and it is the biggest one, because it is the first time the public market has been asked to put a price on a frontier AI company itself, rather than on its suppliers, and the price it puts on Anthropic will be the price it puts on the whole trade.

For a halal investor the bubble question is a question about discipline, not about prediction, and the discipline is to separate the price from the screen. The screen does not measure whether a market is a bubble, and it does not measure whether a price is too high, and it does not measure whether a debut will go up or down. It measures the business, the debt and the interest income, and a company can be fairly priced and fail the screen, or wildly overpriced and pass it, and the two judgments are independent. The bubble question matters to the screen in exactly one way, and it is the same way the valuation matters: the market capitalisation is the denominator of the debt ratio, so a market that pays more for AI growth makes the debt ratio smaller, and a market that pays less makes it bigger, and a weak Anthropic debut would lower the denominator for every company in the trade at once. That is a fact about the arithmetic, and it is not a prediction, and the framework section below treats it as arithmetic, because arithmetic is the part of the market that does not change its mind.

The screen: no verdict yet — and why that is the honest answer

Anthropic has no screen result on this site, and it will not have one until after the shares trade, and the reason is the method, not a gap in the research. The screen has three gates, applied in order, and all three are measured on filed figures. Gate one is business activity: the company must not earn its living from alcohol, tobacco, gambling, weapons, adult entertainment, pork, or conventional finance. Gate two is debt: total debt must be below about 33 percent of market capitalisation, with cash and cash-like securities below about 33 percent as well. Gate three is non-compliant income: interest income and other non-compliant income must be below 5 percent of total revenue. The full method is set out in the screening methodology, and the results for the 939 names this site has screened are on the screener hub. None of the three gates can be run on Anthropic today, because the company has never traded, and the figures the gates need do not exist yet in the form the screen requires. The prospectus gives revenue and losses, but it does not give the post-listing balance sheet that gate two needs, and it does not give the interest income line that gate three needs, and it does not give the revenue mix in the form that gate one needs. A screen run on the prospectus summary would be a guess, and a guess is not a screen result, and this site does not publish guesses.

What can be said, honestly, is what each gate will ask when the filings arrive, and the questions are specific enough to be useful. Gate one will ask what Anthropic actually sells, and in what proportions: the Claude chatbot and the application programming interface that lets businesses build on the model, and whether any of the revenue comes from a prohibited activity, including any government or defence work disclosed in the filings. The business as described is artificial intelligence software, which carries no obvious prohibited revenue line, but the description is not the filing, and the filing is what gate one reads. Gate two will ask how much debt the company carries against its market capitalisation, and the answer will depend on the price on 19 November as much as on the balance sheet, because the market capitalisation is the denominator. The company has been loss-making and heavily reliant on computing commitments, including the roughly $121.3 billion Google Cloud commitment, and the screen will have to decide how those commitments are treated, because they are obligations in the economic sense even if they are not debt in the accounting sense, and that decision has not been made. Gate three will ask how much interest income the company earns on the cash it raises, and the answer will depend on how much of the $90.4 billion it keeps as cash and what it earns on it, and a company that raises $90.4 billion and holds a large part of it in cash can trip the income gate on the interest alone, which is exactly what happened to Cloudflare at 6.05 percent and to Airbnb at 5.76 percent, both of which fail on interest income that grew faster than revenue. The first post-listing filings will answer all three questions, and the first screen will be run on those filings, and the result will be published when it exists, not before.

There is no third-party rating to report either, which is worth stating plainly, because the absence is itself a fact. No Zoya rating, no Musaffa rating and no ShariaPortfolio rating for Anthropic could be verified from public sources, because the company has never been listed and the screeners have had nothing to screen. There is no BDS listing for Anthropic and no moral-screen finding, so the boycott and military-ties sections that appear on other screener pages do not apply here, and the absence is reported as an absence, not as a pass. The contrast with the names in this run of articles is what makes the point. CoreWeave fails at the debt gate at 74.7 percent, Cloudflare fails at the income gate at 6.05 percent, Palantir fails at the income gate at 5.12 percent, and Nvidia passes with debt of 0.15 percent and non-compliant income of 1.07 percent, and every one of those results is a fact about a filed balance sheet and a filed income statement. Anthropic has a prospectus, not a balance sheet in the form the screen needs, and the difference between the two is the difference between a company that can be screened and a company that cannot, and the honest answer is that Anthropic cannot, yet. Providers disagree on the debt ceiling itself, 30 percent versus about 33 percent, and on whether the income ceiling should be measured on total revenue or on a different base, and the disagreement is reported, not resolved, because it decides how close to the line a name really is, and the disagreement will matter the moment the first Anthropic screen is run, because the first screen will be run on a company whose cash pile is the largest variable in the whole calculation.

How a halal investor should approach the Anthropic IPO

An initial public offering that is oversubscribed several times over triggers a specific set of instincts, and they are worth naming before acting on any of them. The first is to buy the book, on the theory that demand this strong must mean the price will rise. The second is to buy the brand, on the theory that Claude is the future and the company behind it must win. The third is to buy the buyers, on the theory that Google and Nvidia putting in $10 billion each must mean the deal is safe. The fourth is to buy the fear of missing out, on the theory that a deal this large will never be this cheap again. None of the four is a framework. A framework is a set of questions, answered in order, before any order is placed, and for Anthropic the questions are these.

First, screen before you react, and accept that the screen cannot be run yet. There is no Anthropic screen result, because there is no post-listing balance sheet, and a halal investor who buys the shares on 19 November because the book was oversubscribed has let the roadshow do the screening, and the roadshow is not an input to the screen. The screen will be run on the first filings after listing, and the result will be published when it exists, and the discipline is to wait for it. Second, separate the demand from the business. The book is oversubscribed, the revenue is up 1,206 percent, the run-rate is about $81 billion, and the operating loss is $1.3 billion, and the demand and the business are the same story told from opposite sides, and the screen measures the business side. Third, do not trade a headline. The pricing is on 18 November, the first trade is on 19 November, the first filings will arrive in the months after that, and the Federal Reserve will meet on 8 and 9 December with more data than it has today. Fourth, remember that you do not use margin. A halal portfolio carries no margin debt, which means no lender can liquidate it on a bad morning, and a first trade that opens below the offering price is a price move, not a forced decision. Fifth, watch the right dates: the pricing on 18 November, the first trade on 19 November, the Friday livestream with Amodei and Dimon, Nvidia's report on 17 November, which is the last of the Magnificent Seven to report and the one whose chips fill the data centres Anthropic rents, and the first 10-K, which is where the debt and interest income figures will actually appear. Sixth, recheck at the refresh. If the ratios move, the next screen refresh will show it in the numbers. The framework is the position: screen first, separate the demand from the business, do not trade a headline, no margin, watch the right dates, and wait for the refresh.

There is one more question the framework asks, and it is the one that matters most for an IPO, because it is the question the screen cannot answer. The screen measures the debt, the business, and the interest income. It does not measure whether the price is fair. A company can be worth $1.4 trillion fully diluted and $1.02 trillion on shares outstanding, and both numbers can be right, and neither one says whether $78 a share is a good price, because the price is a judgment about the future and the screen is a measurement of the present, and the two judgments are independent. A halal investor who waits for the screen is not waiting for a buy signal. The screen is a filter, not a recommendation, and a pass does not mean buy and a fail does not mean sell, and the business analysis is the second filter, and the price is the third. The framework is to hold all three at once, and to wait for the filings, and to remember that the largest offering in history is also the offering with the least public history, and that the least public history is the reason to wait the longest. For readers who want to work through the method while they wait, the screening methodology sets out the three gates in full, the screener comparison sets out how the third-party providers differ, and the purification guide sets out what happens to the income that does not pass. The IPO is an evening. The screen is the house. Read the evening, and keep the house in order.

What to watch next

The next number that matters for Anthropic is not Anthropic's. It is Nvidia, which reports on Tuesday 17 November, the last of the Magnificent Seven to report, a day before Anthropic prices, and the one whose chips fill the data centres that Anthropic rents. Nvidia fell 5.5 percent on Tuesday to $168.89 on the board probe into chief executive Jensen Huang, and Baird cut its rating to Neutral and slashed its price target from $500 to $200 this week, and Morgan Stanley removed the stock from its top picks list, and the report will say whether the customers who are signing Anthropic's contracts are still buying Nvidia's chips at the same pace. If Nvidia reports the same pattern as the rest of the quarter, revenue up sharply and guidance up with it, then the demand story is an industry story, and the Anthropic book is the industry's book. If Nvidia reports something different, then the demand story is a company story, and the two stories have different implications for the pricing on 18 November.

Closer to the deal, watch the pricing itself. The shares are expected to price on Wednesday 18 November, and the bankers are weighing pricing above the $70 to $78 range, which would lift the fully diluted valuation to $1.5 trillion or more and push total fundraising past $100 billion, and the filing has not yet been amended to reflect a new range, so the first sign of the final price will be the amendment itself. Watch the first trade on Thursday 19 November, and watch it as a price, not as a verdict: a first trade above the offering price is a market that wants more, a first trade below it is a market that wants less, and neither one is a screen result. Watch the Friday livestream with Dario Amodei and Jamie Dimon, which is the first time the chief executive will speak to investors as a public company, and watch the retail allocation, targeted at about 20 percent, because the allocation is the measure of how much of the deal the public actually gets. Closer to the screen, watch the dates that decide the cost of money. The Federal Reserve meets on 8 and 9 December, and the Bank of Canada decides on 9 December as well, the same week, and the December hike is now near-certain after Tuesday's hot CPI print, with a 15.3 percent chance of a half-point move, and the January pricing has a cumulative 50-basis-point hike at 48.8 percent. A steeper rate path raises the yield on the cash Anthropic will hold from a $90.4 billion raise, which is the numerator of the income gate, and it raises the cost of the computing the company must buy, which is the denominator of the business. The first 10-K, due in early 2027, is the document that will decide whether the company passes or fails, and the next screen refresh is where the result will show up in the numbers. And watch the Nvidia board probe, because the probe is the governance event that the financial gates do not measure, and it is the reason Baird cut its target by 60 percent, and it is the thread that connects the week's biggest loser to the week's biggest deal. For readers who hold the names in this article, the practical next step is the same as it was after Friday's jobs report: note the screen result as it stands, note the margin, and recheck both at the refresh. Run it through the screening methodology, check the company on the screener hub, and build the rest of the decision from there. A pricing is an evening. The screen is the house. Read the pricing, and keep the house in order.

FAQ

What is Anthropic offering in its IPO?

Anthropic is selling 825 million Class A shares at $70 to $78 each, which raises up to about $64.4 billion at the top of the range, with an over-allotment option for 123.75 million more shares worth about $9.7 billion at $78. A concurrent $26 billion private placement to seven strategic investors lifts the total potential raise to about $90.4 billion. The shares are expected to price on Wednesday 18 November and to begin trading on the Nasdaq on Thursday 19 November under the ticker ANT, with Goldman Sachs and Morgan Stanley leading the deal. Bloomberg reported the offering has drawn orders for multiple times the shares on offer, with the book covered on the first day of the roadshow.

How much is Anthropic worth — $1.4 trillion or $1.02 trillion?

Both numbers are right, on different counts. At $78, the top of the filed range, Anthropic is worth about $1.02 trillion based on the shares that will actually be outstanding after the offering and the private placement. On a fully diluted basis, counting every share including stock options, restricted stock units and shares reserved for employee plans, the same $78 a share comes to about $1.4 trillion. The bankers are weighing pricing above the range, which would lift the fully diluted valuation to $1.5 trillion or more, or about $83 a share. Morningstar put the estimated fair equity value at $1.7 trillion, and a pre-IPO contract on the prediction market Hyperliquid implied about $1.8 trillion.

Who is buying the $26 billion private placement?

Seven strategic investors will buy $26 billion of non-voting stock at the IPO price, including Alphabet, Nvidia, Microsoft and TSMC, with Google and Nvidia each committing about $10 billion. The placement is concurrent with the offering, at the same price, which means the strategic buyers pay the same price as the public, and the money goes to the company. Google is also Anthropic’s largest cloud provider, with a minimum Google Cloud commitment of about $121.3 billion, and was reported to be backstopping about $19 billion of Anthropic’s TeraWulf data-centre leases. Alphabet, Nvidia and Microsoft all pass this site’s October 2026 screen, but a screen is a property of a company, not of its shareholders, so their results do not transfer to Anthropic.

Is Anthropic profitable?

Not yet. Revenue was $33.2 billion in the nine months to 30 September 2026, up 1,206 percent from a year earlier, and the revenue run-rate was about $81 billion at the end of October, up $10 billion in a single month. Operating losses narrowed to $1.3 billion from $5.2 billion over the same nine-month period, which is a real improvement and still a loss. The company is spending more than it earns, and the gap is being closed by growth, not by profit, with the cost of computing as the largest expense. A run-rate is not revenue and not a forecast: it is the revenue the company would earn in a year if the last month’s pace held for twelve months.

Does Anthropic pass the halal screen?

There is no screen result yet, because the company has never traded and there is no post-listing balance sheet to measure debt and interest income against. The screen has three gates, applied in order: business activity, debt below about 33 percent of market capitalisation, and non-compliant income below 5 percent of total revenue. Gate one will ask about the revenue mix in the filings, gate two will ask about debt against the market capitalisation set on 19 November and how the roughly $121.3 billion Google Cloud commitment is treated, and gate three will ask about interest income on the cash raised. No Zoya, Musaffa or ShariaPortfolio rating could be verified, there is no BDS listing and no moral-screen finding. The first screen will be run on the first post-listing filings, and the result will be published when it exists, not before.

What should a halal investor watch next on Anthropic?

Five things. First, the pricing on Wednesday 18 November, and whether the bankers lift it above the $70 to $78 range. Second, the first trade on Thursday 19 November, read as a price, not as a verdict. Third, Nvidia’s report on Tuesday 17 November, the last of the Magnificent Seven to report and the one whose chips fill the data centres Anthropic rents. Fourth, the Federal Reserve meeting on 8 and 9 December, because a steeper rate path raises the yield on the cash Anthropic will hold from a $90.4 billion raise, which is the numerator of the income gate. Fifth, the first 10-K in early 2027, which is the document that will decide whether the company passes or fails the screen.

Affiliate disclosure. This page contains no affiliate links, and the site currently earns no affiliate revenue. If we join affiliate programs in the future, this disclosure will be updated and links will be clearly labeled. Commissions never influence our scores or rankings — this review is based on official documentation and public data, and every product is Shariah-screened before review. Nothing on this site is financial advice.