Anthropic is reportedly pressing ahead with its IPO despite pressure to accept a lower valuation, while OpenAI has reportedly closed an initial $11 billion of a smaller private round. Both stories expose the same tension: expanding AI businesses still have to finance infrastructure on terms investors will accept.
Anthropic is still working towards a public listing before Thanksgiving, according to the latest reporting, even as prospective investors challenge the price. OpenAI, meanwhile, has reportedly received roughly $11 billion in a private financing that originally sought about $30 billion. The two frontier AI labs are testing how much money investors will provide—and how much ownership they want in return.
But neither transaction is the simple verdict on AI that its headline numbers might suggest. Anthropic has not announced a final IPO valuation or completed an offering. OpenAI's $11 billion is reportedly an initial close, not the final total. And its fundraising retreat was reported before the November 5 market rout, making that sell-off an intensification of an existing problem rather than its sole cause.
Anthropic is pursuing the IPO, not promising a price
The clearest confirmed development is Anthropic's October 26 announcement that it had filed for an initial public offering and intended to list on Nasdaq under the symbol ANT. Its November 2 amended prospectus still leaves the offer price and share count blank. Filing establishes an intention to sell shares; it does not establish that buyers have accepted particular terms.
The Wall Street Journal, in reporting republished by Mint, said Anthropic was aiming to list the week before Thanksgiving and had been expected to raise up to $100 billion at around a $1.5 trillion valuation. Three major prospective investors said the valuation would probably have to fall; one suggested closer to $1 trillion. The report also described a smaller offering or a delay as possibilities, with executives and advisers expected to discuss the plans.
That pressure has independent support. Reuters reported on November 5, citing four people familiar with investor discussions, that buyers wanted a more attractive valuation and questioned whether the planned roadshow could proceed on time. Bloomberg's reporting, republished by The Edge, put prospective buyers' discussions at $1.2 trillion to $1.3 trillion, with some considering as little as $1.1 trillion. It explicitly said no final decisions had been made.
The latest account pushes back against assuming a postponement. The Financial Times reported that Anthropic was pressing ahead and was willing to accept a lower price, according to Seeking Alpha's summary carried by TradingView. That summary describes plans to raise $80 billion to $100 billion at a valuation of $1.25 trillion to $1.75 trillion, while some investors expected around $1.3 trillion. CNBC's November 7 coverage also described a likely mid-November debut, with investors seeing a $1 trillion to $1.5 trillion valuation range.
These are overlapping but different accounts of expectations, not successive official price cuts. The defensible conclusion is that Anthropic is still pursuing a near-term listing under valuation pressure—not that it has accepted $1 trillion, $1.1 trillion or any other final price. Its willingness to proceed at a discount is reported through investors, not a new company announcement guaranteeing completion.
OpenAI's $11 billion is real financing in the reporting—not a finished $30 billion round
Bloomberg reported on November 6 that OpenAI had closed roughly $11 billion in new funding at a $1.25 trillion valuation. Its sources said the money had been wired as of Friday. The remainder could take the round to as much as $20 billion in the coming weeks. TechCrunch's account accurately describes the wider round as unfinished.
The chronology matters. TechCrunch reported on October 31 that the FT had already described a reduced $20 billion target, while a separate Journal report put secured funding at $11 billion and described a tepid response from some large investors. The earlier ambition was about $30 billion at a $1.4 trillion valuation. Bloomberg's later account adds an important distinction: reported commitments had progressed to money received.
On those reported figures, $11 billion is about 37% of the original $30 billion goal. Reaching $20 billion would still leave the completed round one-third smaller than that original target—but would also mean another $9 billion arriving after the initial close. It is reasonable to describe a fundraising shortfall so far. It is premature to call $11 billion the final ceiling, or to say OpenAI has failed to close any financing.
OpenAI's delayed IPO adds another difference. TechCrunch reports expectations of a listing next year; CNBC likewise describes executives signalling a 2027 debut. That is not an announced trading date. Nor does the evidence establish that the current fundraising shortfall alone caused the IPO delay. The Verge's account notes that chief executive Sam Altman had also cited confidence in model safety when discussing the decision to hold off.
Three numbers that should never be collapsed into one
A valuation is the price assigned to the equity of the business. Capital raised is money provided in the transaction. A fundraising target is an ambition, not a bank balance. Lowering one does not mechanically lower all three by the same proportion.
For Anthropic, moving from a hypothetical $1.5 trillion valuation to $1 trillion would be a one-third reduction in the expected price of the company. It would not mean $500 billion had disappeared from Anthropic's accounts. The original $1.5 trillion was itself an expectation, not an IPO price at which the entire company had changed hands.
A lower valuation can nevertheless make raising the same cash more expensive for existing owners. As an illustration—not a forecast of Anthropic's terms—$100 billion of entirely new shares at a $1.5 trillion post-money valuation would represent about 6.7% of the resulting equity; at $1 trillion, it would represent 10%. Existing holders would retain a smaller percentage. If those valuation figures were pre-money instead, the calculations would change.
The prospectus adds another complication: both Anthropic and existing shareholders plan to sell shares, and Anthropic receives no proceeds from shareholders' sales. Therefore an $80 billion or $100 billion total offering cannot automatically be treated as the same amount of fresh cash for the company. The eventual primary-versus-secondary split, fees and other uses of proceeds matter.
For OpenAI, $1.25 trillion is below the reported $1.4 trillion ambition, but that does not establish a down round relative to its last completed financing. In March, OpenAI announced a closed round with $122 billion in committed capital at an $852 billion post-money valuation. The new headline valuation is higher. Bloomberg's Anthropic coverage describes OpenAI's current target as pre-money; the March announcement explicitly says post-money. Those bases, security terms and share counts must be reconciled before calculating a precise investment return or dilution.
Underneath the repricing: revenue now, infrastructure obligations for years
Anthropic's public filing provides a stronger foundation for analysing the pressure than investor mood alone. It reports $33.242 billion of revenue for the first nine months of 2026 and $17 billion in the third quarter. CNBC's examination of the filing puts third-quarter growth at about 1,127% year on year, while noting that sequential growth slowed to about 47% from much higher rates earlier in the year. Those figures describe substantial customer spending, not a business with no demand.
Yet the same filing discloses approximately $895.4 billion of purchase and lease commitments entered into as of and after September 30, substantially all related to compute and infrastructure. Crucially, that is not a bill due immediately: approximately $69.8 billion is due in the 12 months after September 30, about $447.5 billion is expected before the end of 2030, and substantially all the remainder between 2031 and 2045.
The balance sheet also deserves precision. Anthropic reported approximately $95.6 billion of cash and cash equivalents plus $36.3 billion of short-term investments as of September 30—about $131.8 billion combined. That is not the same thing as the $146.6 billion described as cash on hand in the Journal account. The filing's components should take precedence when assessing immediately held financial assets; undrawn borrowing capacity is not cash.
Neither comparing all multi-year commitments with today's cash nor dividing cash by the next year's commitments produces a reliable insolvency forecast. Future customer receipts, operating costs, financing access and the detailed terms and timing of those obligations all matter. The revealing issue is the mismatch between infrastructure commitments made years ahead and revenues that still have to be earned.
There is an equally important accounting trap. Anthropic's nine-month net loss was about $104 billion, but approximately $94.3 billion came from fair-value changes in convertible notes, derivative liabilities and similar instruments. Its operating loss was about $1.26 billion. The enormous net-loss headline is not a measure of cash burned on running Claude. Operating loss, cash flow and financing-accounting charges answer different questions.
OpenAI's current operating finances are less visible in the material supporting this round. Its March announcement argued that greater computing capacity would support better products, adoption and revenue. That is the company's investment thesis, not proof that every additional dollar of compute earns an adequate return. An $11 billion initial close also cannot be used to infer OpenAI's remaining cash or financial runway without current cash-flow and balance-sheet information.
Why public-market selling changes a private lab's price
The mechanism does not require investors to decide that AI has stopped working. When listed technology shares become cheaper, investors can demand a discount from an unlisted business competing for the same capital. Buyers who have suffered portfolio losses may also have less appetite for a very large new position. These are financing channels, not direct measurements of ChatGPT or Claude usage.
There is an operating channel too. Anthropic's filing identifies constraints on compute inputs, regulatory uncertainty and reliance on a limited number of suppliers as risks. If power connections, permits or construction slip, computing capacity may arrive later than expected. That can postpone revenue while making the financing of expansion harder. CNBC's prospectus coverage also describes investor concerns that higher borrowing costs and pressure on cloud providers' cash flows could slow delivery.
The post-election data-centre backlash makes those risks more salient, but it is not evidence of a nationwide construction ban or cancelled spending of any specified size. A change in the risk premium—the extra return buyers demand for uncertainty—can reduce a valuation even while revenue continues growing.
The market-loss headlines measure different things as well. The Journal reported roughly $2.5 trillion erased from companies in the Nasdaq Composite on November 5. Bloomberg's Anthropic coverage used about $2.3 trillion for the narrower Nasdaq 100. Neither is the FT's reported $3.2 trillion S&P 1500 decline, explained in VibeCodAI's examination of the sell-off. These overlapping index losses are not amounts to add together, and they are not cash withdrawn from either AI lab.
What investors are testing next
For Anthropic, the next meaningful evidence is a formal offer range, share count, primary-versus-secondary split and eventual pricing—not another anonymous estimate. Continuing with a smaller or cheaper offering could secure a large capital buffer and public-market access while conceding price. Postponing could avoid selling into turbulence but leave uncertainty over when, and on what terms, that access returns. Neither option is costless.
For OpenAI, the test is whether the remaining financing closes, how much arrives, and on what terms. A $20 billion completion would not erase the retreat from the original target; stopping at $11 billion would be a different outcome from the one now anticipated in reporting. Its future IPO timetable remains a separate question.
Together, the reports show investors becoming less willing to fund frontier AI at whatever price its backers once hoped to achieve. They do not show that demand for AI services has vanished. The deeper question is whether rapidly growing sales can fund an infrastructure programme whose commitments extend far beyond the next model release—and whether new investors are being offered enough of the eventual returns to carry that risk.