
Optimism around the AI trade lifted the Nasdaq Composite more than 2% after Anthropic’s stock market debut on the morning of November 19. Priced at $83, above the upper end of its marketed range, the IPO raised approximately $81.3 billion, establishing a new global IPO record. The shares began trading at $111.50, a 34% gain, before briefly touching $115.10. Markets read the subsequent Nasdaq advance as a strong indication that the “AI trade remains healthy.” For Jake Behan, head of capital markets at Direxion, the sequence summed up investor sentiment: Nvidia had first delivered better-than-expected results, and traders then showed a willingness to buy Anthropic above its offer price. “The market isn’t ready to give up on this trade.”

That early confidence in AI did not survive the trading session. From around $115, Anthropic slipped steadily to close at $97.17, leaving it with a 17% gain. At almost the same time, the wider market changed direction. Having risen more than 2%, the Nasdaq ended down 2.25%, while the S&P 500 lost approximately $1.3 trillion in market value that day, a sum the article describes as closely matching the proceeds of Anthropic’s IPO. Declines extended across all 11 S&P 500 sectors, with more than 90% of constituents finishing lower. The VIX briefly crossed 29, reaching its highest intraday reading since March. How did a successful listing end up helping to trigger a sell-off across the market?
How Did the IPO Drain Market Liquidity?
Total fundraising exceeded $100 billion once approximately $26 billion in private placements to strategic investors was added to the $81.3 billion public offering. According to reports from trading desks, investors financed their subscriptions by selling gold, copper, cryptocurrencies and assets used as proxies for AI, including Nvidia. Price action supported that explanation: gold dropped below $3,800, copper slumped approximately 5%, and losses spread across cryptocurrencies.
The immediate issue was not so much whether Anthropic was worth buying as how investors would find the cash. Assembling more than $100 billion for one asset on a single day requires sales of holdings that can readily be converted into cash, even when investors have not become bearish on those assets. AI shares offered precisely that combination: ample liquidity and sizeable accumulated gains that could be realised.
Another setback arrived that day when short seller Hunterbrook published a report on AMD, alleging that the company had supplied chip firmware updates to blacklisted companies. A fall of more than 9% in AMD pulled the wider semiconductor sector lower. With the IPO already draining liquidity, this separate adverse development added to selling in AI stocks. It was not sufficient on its own, however, to account for the broader market swinging from a 2% gain to a 2% loss. The acceleration came from the market’s technical structure.
Why Did Negative Gamma and Options Expiry Deepen the Decline?
Ahead of the reversal, Nomura strategist Charlie McElligott had warned that Nvidia’s earnings, Anthropic’s listing and monthly options expiry had together taken market makers’ negative gamma exposure to its highest level since 2019. Negative gamma requires market makers to sell into falling prices to hedge their positions. Those sales then push prices down further, setting up a self-reinforcing loop. Substantial flows into 0DTE, or zero-days-to-expiration, options that afternoon put additional pressure on support from longer-dated positions.
Once Anthropic pulled back from its high and set off the initial selling, hedging by market makers transmitted that pressure and magnified it. Positioning and technical hedging effects consequently accounted for a significant part of the wider market’s decline through mechanical selling, rather than fresh fundamental news.
A warning from Goldman Sachs analyst Rich Privorotsky that morning highlighted a different aspect of positioning. At approximately 18%, institutions’ net exposure to technology shares was substantially below the sector’s 29% index weight and near the most underweight level in five years. Such positioning could, in theory, have fuelled a squeeze higher as investors restored exposure, so long as the market did not fall sharply. Instead, selling mechanically driven by negative gamma and options expiry proved stronger than that potential source of buying.
Why Did More Than Half of Institutions Miss Out as Retail Took a Larger Allocation?
Beyond the more apparent liquidity drain and negative gamma effects, Anthropic’s share allocation attracted less attention. Reports from The Wall Street Journal and other media outlets indicated that approximately 1,200 institutional investors placed orders, yet more than half were allotted no shares. Retail investors received around 15% of the offering. Allocations also favoured the United States heavily, leaving many European orders unfilled. ZeroHedge put forward a more consequential reading: the underwriters either had enough confidence in demand to limit institutional allocations deliberately, or institutions would not pay $83, leaving the shares to be placed with retail investors who were less sensitive to price. Markets favoured the second explanation.
The resulting shareholder mix influenced trading after the debut. With no allocation, institutions had little reason to support the share price following the opening or to buy on declines. Retail investors, whose decisions were more vulnerable to sentiment and short-term price changes, controlled a relatively large proportion of the tradable shares. Profit-taking followed the initial surge, and without institutional buying to provide support, the price quickly weakened. At $97.17, Anthropic’s closing price stood approximately 13% below its $111.50 opening level and around 16% below the intraday peak of $115.10. The scale of the pullback was itself evidence of how fragile the ownership structure was.
What Did the Fed’s Same-Day Warning Mean for AI Debt Scrutiny?
Released on November 19, the minutes of the Federal Reserve’s October meeting revealed that almost all officials still considered further interest rate increases necessary. More importantly, an official Fed document for the first time explicitly flagged opaque AI-related debt, off-balance-sheet lease commitments and circular financing arrangements as risks to financial stability. Staff cautioned that disappointing returns on AI investment could result in losses being magnified by off-balance-sheet lease commitments and circular financing arrangements between AI companies. Most participants noted the rapid expansion of AI-related debt, much of it funded through complicated, opaque structures.
This went well beyond a general increase in expectations for further rate rises. Until then, assessments of AI valuations had centred on the strength of demand, the volume of orders and rates of growth. The minutes added a financing test: was the funding underpinning demand transparent, could off-balance-sheet commitments deepen losses if returns fell short, and was circular financing generating artificial revenue?
The concerns found a telling illustration in Anthropic’s prospectus. It reported $33.2 billion in revenue for the first nine months, an operating loss of $1.26 billion and a net loss of $104 billion, primarily attributable to fair value changes in convertible notes and derivative liabilities. Computing capacity purchase commitments totalled $900.9 billion. Now that the Fed had formally identified opaque AI debt as a financial stability concern, investors confronted a systemic issue broader than Anthropic’s ability to turn profitable: could the financing model underpinning the whole AI infrastructure sector withstand formal scrutiny while interest rates remained higher for longer?