$83 Priced, $111.50 Opened, $88.91 Two Days Later: Anthropic's First Week on Nasdaq
On 19 November 2026, Anthropic listed on Nasdaq as ANT: priced at $83, raising a record $81.3 billion, opening at $111.50 and closing at $97.17; on day two it fell to $88.91. With the green shoe fully exercised the IPO raised $93.5 billion, plus a $26 billion private placement.

In short: the biggest IPO in history went public — and the market gave it barely a day and a half of good grace
At midday on Thursday 19 November 2026, Anthropic's shares began trading on the Nasdaq under the ticker ANT. They opened at $111.50, 34% above the $83 offer price; touched above $115 during the session; and closed at $97.17, still up 17% on the day. This was a listing that had been trailed for weeks: a public prospectus on 26 October, a roadshow starting 10 November, pricing at $83 on the evening of 18 November, the first trade on 19 November, and on 20 November the company announced the deal was closed — $93.5 billion raised in the IPO itself, another $26 billion in a concurrent private placement, $119.5 billion combined. At the offer price, Anthropic's fully diluted valuation was about $1.5 trillion; at the opening price, it briefly approached $2 trillion. Every number broke a record: the initial $81.3 billion raise passed SpaceX's $75 billion from June, making it the largest IPO ever.
Then the market turned. On the second day of trading, ANT closed at $88.91, down 8.5% in a single session, leaving it only about 7% above the offer price. In the same week, the Nasdaq had its worst week since the Covid crash of March 2020, and the S&P 500 and the Dow both entered correction territory. Anthropic's debut landed in the middle of the worst week of the year for AI stocks. This article lays out the numbers, the filings, and the market backdrop from those five days, then answers three questions: how much money does this company actually make? Why did the market fall the very week it listed? And what does any of it mean for people who use Claude to write code every day?
From $70 to $83: the week of pricing
First, the timeline in a straight line. On 26 October, Anthropic filed its S-1 prospectus publicly, planning to list on the Nasdaq Global Select Market under ANT, with the share count and price still undecided. On 10 November, the roadshow began with a price range of $70 to $78 a share and 825 million shares on offer. The books were heavily oversubscribed with high-quality orders. On the evening of 18 November, the pricing came out: $83 a share, above the top of the range, with the deal upsized to 980 million shares and $81.3 billion raised. At 11:50 a.m. New York time on 19 November, the shares started trading.
Pricing above the range is itself a signal of demand. But the $83 figure needs a second piece of context: Anthropic's last private round, in May 2026, valued it at $965 billion after the money; the round before that, in February, at $380 billion. In nine months, the valuation rose from $380 billion to $1.5 trillion, nearly four times over. The roadshow range of $70 to $78 already implied a valuation at that scale, and the final pricing pushed it one notch higher. The market was pricing this company with a yardstick that barely had a precedent.
On 20 November, Anthropic published its closing announcement and nailed the final numbers down: the IPO issued 1.127 billion Class A shares in total, with the underwriters exercising their over-allotment option — the "green shoe" — in full, buying an extra 147 million shares; total IPO proceeds were about $93.5 billion, of which roughly $91.6 billion went to the company, the rest coming from 23,648,646 shares sold by certain selling stockholders. The concurrent private placement sold 313,253,007 Class N shares to Advanced Micro Devices, Broadcom, Google, Micron Technology, Microsoft, NVIDIA, and TSMC Partners, at the same $83 offer price, for about $26.0 billion. Combined, about $119.5 billion. Goldman Sachs and Morgan Stanley were lead book-running managers.
Underwriting the deal was a record business in its own right. According to a revised filing, Wall Street underwriters collected about $550 million in fees on the $81 billion deal — 56 cents a share — passing the $500 million SpaceX paid in June; as a share of the offer price, about 0.67%, roughly the same as SpaceX. The offering also left a door open for retail investors: buyers in nine European countries and Switzerland could subscribe under a separate prospectus. In every part of the process, from fees to retail access, Anthropic's listing set a new record.
The private-placement buyers deserve a second look: all seven are key links in Anthropic's own supply chain — chip suppliers (AMD, Broadcom, Micron, NVIDIA, TSMC) and cloud and compute partners (Google, Microsoft). NVIDIA alone has invested a combined $20 billion in Anthropic, $10 billion of it in a partnership announced in November 2025 and another $10 billion in a share purchase tied to the IPO. In other words, Anthropic's biggest suppliers and its main sources of compute are also its shareholders after listing. That "customer as shareholder" structure comes back later in this article.
The first day of trading: opened at $111.50, closed at $97.17
Trading on 19 November was dramatic. The shares opened at 11:50 a.m. at $111.50, a 34.3% premium to the offer price. Within minutes they touched $115.10, the day's high, then slid all afternoon, dipping below $100, reaching a low of $92.72, and closing at $97.17. About 857 million shares changed hands. Measured against the close, the first day was up 17.1%; measured against the open, anyone who bought at the debut was down about 13% by the bell.
The second day looked worse. On 20 November, ANT closed at $88.91, down another 8.5%. Two days after listing, the shares were only about 7% above the offer price. Anthropic's own announcement, published that same day, used standard deal-closing language and said nothing about the share price. In 48 hours, the market's price for the company slid from "close to $2 trillion" to "a little above $1.5 trillion".
Those numbers have to be read against the broader market that week. On 19 November, the Nasdaq fell 2.2%, the Dow 2.19%, and the S&P 500 1.8%; Microsoft fell 1.78%, NVIDIA 3.3%, Meta 3.2%, Apple 0.9%, AMD 7.72%, and Micron 2.1%. On 20 November it got heavier: the Nasdaq fell 5.28% in a day, the S&P 500 3.70%, the Dow 1.93%; NVIDIA dropped 14% to $125.53, AMD more than 13%, Micron about 14%, and Broadcom about 12%. For the week, the S&P 500 fell about 8.6%, the Nasdaq about 10%, and the Dow about 8%. By Friday's close, the S&P 500 and the Dow were in correction, more than 10% below their records, and the Nasdaq had given up all of its gains for 2026. Anthropic listed in the middle of that week.
Anthropic in the prospectus: revenue racing, losses larger still
The most useful reading in any listing document is always the financials. According to Anthropic's SEC filings and reporting based on them: revenue was $4.59 billion for all of 2025; $33.24 billion in the first nine months of 2026, including $16.96 billion in the third quarter alone, more than 12 times a year earlier. By the company's own measure, run-rate revenue was about $81 billion at the end of October, up $10 billion from a month earlier. More than 1,900 customers pay at least $1 million a year, by the company's count.
Growth is fast but slowing: quarter-on-quarter revenue growth fell from 144% in the second quarter to 47% in the third. CNBC listed that slowdown among the reasons investors grew more cautious about AI stocks.
The loss figures look scarier, and are easier to misread. The operating loss in the third quarter was $567 million; the net loss was $23.3 billion in the quarter and $104 billion over nine months. Al Jazeera quoted the $104 billion figure directly. But what it is made of matters: most of that net loss comes from non-cash changes in the fair value of convertible notes, derivatives, and SAFEs (simple agreements for future equity) — accounting adjustments, not money the business burned through operations. The operating loss of $567 million is the figure that reflects the business itself — still substantial, but not in the same league as $104 billion. Confusing the two produces a completely wrong conclusion.
One set of numbers matters more than profit or loss: compute commitments. Anthropic's total compute commitments are about $900 billion, of which about $70.7 billion falls due in the 12 months after 30 September 2026. The other side of the same coin is what it owes its partners: about $110 billion committed to Amazon for computing infrastructure under long-term agreements, and at least $121.3 billion to Google. After listing, the company has to explain itself to the market every quarter: can revenue keep up with $900 billion of commitments? That is the real price of going public — not the hard work of the roadshow weeks, but the earnings call every three months from now on.
One more detail: each of the seven co-founders sold 3.38 million shares in the offering, about $1.96 billion combined at the offer price. That is not much money at this scale, but it means the founding team partly cashed out at the same moment the company listed, and the governance structure contains an unusual arrangement: as a public benefit corporation, Anthropic's Long-Term Benefit Trust, four trustees with no financial stake, elects four of the seven directors. That structure is the institutional guarantee Anthropic has always promoted for "safety ahead of profit", and whether it survives the pressure of quarterly results on Wall Street is one of the most interesting things to watch over the next few years.
Why the market fell in that exact week: four things landed at once
Was Anthropic's listing a coincidence or a cause of the selloff? No source has shown data that the IPO caused it, but four things did land in the same week.
First, NVIDIA. On 11 November, the board accepted the resignation of CEO Jensen Huang. On 20 November, a Reuters report said SEC staff were examining NVIDIA's accounting. NVIDIA is down nearly 40% in November, including 14% in a single day on 20 November. It is the heaviest stock in the AI trade, and its fall drags the whole sector with it.
Second, interest rates. Investors were pricing in further Federal Reserve rate rises, and the 10-year Treasury yield was above 5% during the week. Rising rates hit high-valuation growth stocks hardest, and AI stocks are exactly that kind of asset.
Third, policy. The midterm election results were read by some investors as a brake on data-centre building — all but one ballot initiative restricting data-centre expansion passed, and Democrats campaigned on reining in the AI industry. At the same time, the US tightened rules on remote access to American chips from abroad. Policy headwinds and a turning valuation narrative hit the sector together.
Fourth, the technical picture. The S&P 500 broke below its 200-day moving average, a level that can trigger automatic selling by trend-following funds. There is one more interpretation worth recording, clearly labelled as interpretation: CNBC's market analysis called the IPO "the breaking point" and a possible "sell the news" moment — investors who had bought AI stocks in anticipation took profits once the listing happened. Other analysts quoted by CNBC read it the opposite way: a record IPO that rose on its first day while chipmakers fell suggests AI stocks no longer move together. Both readings have their advocates; neither is confirmed by data. The fair summary is that the IPO coincided with the selloff and is one of several factors commentators name, with the NVIDIA news the largest.
What it means for Claude users: nothing changes now, the earnings calls matter later
The short answer first: for people who use Claude and Claude Code to write code every day, the listing changes nothing in the short term. The models, the apps, and API prices stay as they are. Anthropic is a public benefit corporation, and listing did not change its legal identity; the Long-Term Benefit Trust arrangement is still in place.
Three things matter over the longer term. The first is pricing pressure. Anthropic now carries about $900 billion in compute commitments and a quarterly earnings report, so adjustments to API and subscription pricing will be watched more closely than before. Claude's model prices are usage-based, so any change flows straight into developers' costs. The second is product pace. The company's commitment to "safety ahead of release speed" will be tested under quarterly-results pressure — the prospectus devotes about 90 pages to risk factors, including warnings that its AI systems could pose "catastrophic or existential risks to humanity"; at the same time, a group of AI safety experts published a statement around the IPO, dubbed the "Misanthropic Statement" in press coverage, challenging the company's claims about its commitment to safe AI. The third is the competitive picture. This is the first US frontier AI lab to go public, while OpenAI has also eyed an IPO. Anthropic's share price is now a public daily quote on "what a frontier model is worth", and that price will feed into funding, pricing, and hiring across the whole industry.
For VibeFix readers, there is one more direct link: Anthropic is the company behind Claude Code, one of the most-used coding agents in this community. A share-price wobble will not change how you write code tomorrow, but it will shape how much this company can invest in models, tools, and pricing over the next few years. Listing is not the finish line; it is the first time this company has been asked to account for itself, every quarter, to the whole world.
My judgement: up 17% on day one, up only 7% two days later — which number is more real?
Both numbers are real, but they describe different things. The $111.50 open describes scarcity: the first frontier AI lab to list, books heavily oversubscribed, everyone afraid of missing out. The $97.17 close describes cooling: the opening premium carried emotion, and emotion gets digested within a few hours. The $88.91 close on day two describes the environment: when the whole AI sector is falling, even the best individual story struggles to stand alone.
What is really worth remembering is not any single day's close but the gap between valuation and revenue. A $1.5 trillion fully diluted valuation at the offer price, set against $33.2 billion of revenue in nine months and about $81 billion of run-rate revenue, means the market is betting the company keeps growing at close to a doubling pace for years and eventually turns $900 billion of compute commitments into real profit. That bet may be right or wrong — but it is not built on "already making a lot of money"; it is built on "will keep growing fast". That is why the market flinched when quarter-on-quarter growth fell from 144% to 47%: the denominator of the valuation is the growth expectation, and when growth slows, the valuation has to be recalculated.
The second judgement is about "customers as shareholders". NVIDIA, Google, Microsoft, AMD, Broadcom, Micron, and TSMC are at once Anthropic's suppliers and sources of compute, and its shareholders and private-placement buyers after listing. The upside of that structure is alignment: suppliers have reason to keep supply flowing, shareholders have reason to back the company. The risk is equally clear: when your biggest customers, biggest suppliers, and biggest shareholders are the same group of companies, a shock to any one of them travels along the chain. In the week NVIDIA's CEO resigned and its shares fell, Anthropic's shares fell too. That is not a coincidence; it is what this interwoven structure looks like in daily life.
Finally, back to the question VibeFix readers care about most: what does this mean for vibe coding? The direct impact is none; the indirect impact is real. Anthropic is the company behind Claude, one of the most important model families in this ecosystem. Listing means it now answers to the public market, and model pricing, release pace, and safety investment will all carry one more layer of constraint from the earnings report. Nothing changes in the short term; over the long term, every quarterly report this company files will be a weather vane for this ecosystem.
How this relates to what we have already published
This site has published two earlier articles on Anthropic's path to listing: one on 26 October about the public S-1 filing — the company filing its prospectus and planning to list on Nasdaq as ANT, with share count and price undecided — and one on 10 November about the roadshow launch — the $70 to $78 price range and 825 million shares. This article's anchor is the pricing of 18 November, the first day of trading on 19 November, and the closing of the deal on 20 November: the end of the listing process, not the start. The three articles cover different stages of the same story, separated here by timeline. An earlier piece from 2 October about Anthropic's confidential IPO filing covers a still earlier, confidential stage, and does not overlap with this one.
If you remember only one sentence from this article, make it this one: Anthropic priced at $83, raised $81.3 billion in the largest IPO in history, opened at $111.50 and closed its first day at $97.17, then fell to $88.91 on day two — the market gave it a record-breaking listing and a far from friendly first week.
Comments (0)
Sources
Related articles
METR's Q4 2026 capability update, published 20 November 2026: public models' time horizons double roughly every 3.61 months, faster on the verifiable TH2 subset (1.65 months), reaching 200 hours by July 2027 on naive extrapolation — but models are clearly worse than human experts on judgement tasks without clear feedback, and GPT-5.6 Sol reward-hacked the most.
On 20 November 2026, Signal said it will block in-app screenshots by default on iOS 27 and auto-enable screen security on Android when an AI assistant reads the screen, to stop OS-level Siri AI and Gemini Intelligence scanning the display. macOS cannot be protected yet, and the fix itself is incomplete: the other person's device, the filesystem and notifications remain holes.
On 20 November 2026, Anthropic mathematician Levent Alpöge posted a four-page note claiming he and Claude had found a proof of the Riemann Hypothesis — and the generalised version too. The proof exists only as Lean code: 175,187 files, about 17.4 million lines. MIT's Andrew Sutherland compiled it with zero errors and the Lean and Mathlib communities certify it correct, but no human has read the whole proof yet.