Key Points
ใปAnthropic, the US company behind the AI assistant Claude, listed on Nasdaq under the ticker ANT on November 19, 2026, during a broad sell-off in US stocks.
ใปThe IPO was about cash: the company expects to receive about $117 billion to secure years of computing capacity, and that money stays with the company even if the share price falls.
ใปShareholders are now spread around the world, but the founders keep a fixed majority of the general vote and a trust still picks most of the board. The filing does not show that Claude’s prices, quality or ease of use will change soon because of the listing.
โผ Watch the Japanese-language video explainer
This video has Japanese audio and slides. To read English captions, open Settings โ Subtitles/CC โ Auto-translate โ English, where available. Automatic translations may contain errors. If the option is unavailable in the embedded player, watch on YouTube. The accompanying English article provides the written explanation.
Anthropic Lists on Nasdaq as AI Stocks Fall
Anthropic, the US AI company that develops the AI assistant Claude, began trading on the Nasdaq Global Select Market under the ticker “ANT” on November 19, 2026. Claude is used to draft emails, summarize long documents and write software.
The shares were priced at $83 the day before, opened at $111.50 and closed the first day at $97.17, according to Anthropic and CNBC. On November 20 they closed at about $88, down roughly 9% on the day but still above the offering price.
| Price per share | |
|---|---|
| IPO price (set November 18) | $83.00 |
| First trade (November 19) | $111.50 |
| Close, November 19 | $97.17 |
| Close, November 20 | about $88 |
On November 20, Anthropic announced the closing of the offering. According to that announcement and its prospectus, new shares issued by the company raised about $91.6 billion, while existing shareholders sold about $2 billion. Separately, the company completed a $26 billion private placement of generally non-voting shares to seven companies: AMD, Broadcom, Google, Micron, Microsoft, NVIDIA and TSMC.
The wider market fell on both days. The Nasdaq Composite dropped 2.25% on November 19, and on November 20 the Nasdaq Composite lost 5.3% and the S&P 500 3.7%, according to the Associated Press. In Tokyo, the Nikkei Stock Average fell 5.06% on November 20. Reports cited continuing doubts over whether the huge sums invested in AI will pay off.
Index provider MSCI said in a November 2026 announcement that it would add Anthropic to its major global stock indexes, which funds around the world use as benchmarks, as of the close on November 30.
At a White House press conference on November 20, US President Donald Trump called Anthropic’s launch “very successful” and said he had “a lot of respect for Dario,” referring to CEO Dario Amodei. At the same press conference he said the US would put “a tremendous tax” on American companies that build data centers in other countries, but he did not give a rate, a timeline or a legal mechanism, and said he had not yet told his own staff.
Related Articles
Reading an IPO in a Falling Market
Why Anthropic needed so much money now
Compute refers to the specialized chips and data centers that train and run AI models. Every answer an AI gives takes a large amount of computation, and as more people use it, and as companies train larger models, they need more of these facilities.
According to Anthropic’s prospectus (Form 424B4, November 2026), the company spent about $18.8 billion on compute and infrastructure in the first nine months of 2026, more than four times the same period a year earlier. Capacity is usually reserved years in advance by contract, and the company’s future payments under contracts it has already signed total about $900 billion. That is the total of commitments spread over many years, which is different from debt it owes today.
The same prospectus says compute rented at short notice to meet sudden demand cost significantly more than capacity under multi-year contracts. In other words, raising money first and locking in capacity on long-term contracts is a cheaper way to support growth.
The prospectus describes the use of proceeds broadly, as general corporate purposes that include investment in compute and infrastructure and hiring.
Anthropic had already raised money before the listing through investments from Amazon and Google and through bond sales, so an IPO was one funding route among several. Still, for raising more than $100 billion at once and spreading ownership among investors worldwide, a listing was the largest channel available.
The cash raised and the share price move separately
The money a company raises and its share price move on separate tracks.
When a company sells new shares in an IPO, the proceeds belong to the company once the sale closes. After that, when shares change hands on the market, the money passes only between sellers and buyers. A falling share price does not take back the cash the company has already received. Combining the public offering and the private placement, Anthropic expects to receive about $117 billion after fees and expenses, according to its prospectus. The roughly $2 billion from existing shareholders went to those shareholders, not to the company.
The share price, by contrast, reflects expectations about future profits. The offering price was set above the range the company had indicated earlier, according to CNBC.
At the second-day close of about $88, investors who bought at the offering price still had a paper gain. Those who bought at the opening trade of $111.50 had a paper loss of about 20% within two days. The same stock can mean a gain or a loss depending on when it was bought, and neither is locked in until the shares are sold.
As a way to raise money, the IPO delivered a large sum of cash even in a weak market. But if the share price stays low, raising money again by issuing new shares later would require giving up more shares for the same amount. Whether the IPO succeeds will be decided by whether the business can live up to the expectations in the share price.
A two-day sell-off around the listing
In order of time, US stocks fell on November 19, the day of the listing. Japanese stocks fell sharply on November 20, and US stocks fell further when New York trading opened that evening, Japan time. Tokyo’s November 20 session ended before US trading on November 20 began.
Reports pointed to doubts over whether the enormous sums put into AI will be recovered. It cannot be said that the listing, or Trump’s comments on taxing overseas data centers, caused the indexes to fall. The more natural reading is that the IPO took place in a market where hopes and fears about AI were both running high.
The founders hold the vote, and a trust picks most of the board
The listing spread Anthropic’s shareholders around the world, but the people who set the company’s direction are largely the same.
Anthropic is a public benefit corporation (PBC), a for-profit company organized under Delaware law whose directors must balance shareholders’ financial interests with the company’s effect on society and its stated public benefit. Its charter states that purpose as the responsible development of AI for the long-term benefit of humanity. A PBC is not a nonprofit.
Decision-making power is split in two. 50.1% of the vote on general matters at shareholder meetings is fixed in a company held by the seven co-founders. The power to elect a majority of the board belongs to the Long-Term Benefit Trust, a body of five trustees with no financial stake in the company. The trust does not run the business day to day; that remains the job of the board and management.
According to its prospectus, Anthropic is not aware of a direct precedent among public companies for this combination of a founders’ company and a trust.
Amazon and Google, the two largest shareholders, own about 15% and 13% of all shares respectively, according to the November 2026 prospectus. Their shares, however, generally carry no vote. What they hold are economic rights such as dividends.
The share classes and their roles are as follows.
| Share class | Main holders | What they can decide |
|---|---|---|
| Class A | Public investors | One vote per share |
| Class N | Business partners such as Amazon and Google | Generally no vote |
| Class F | A company owned by the seven co-founders | 50.1% of the vote on general matters |
| Class T | Long-Term Benefit Trust | Elects and removes a majority of the board |
AI Moves Ahead With Money and Control in Different Hands
AI spending first becomes revenue for chipmakers and cloud providers
The contracts Anthropic has signed first support the revenue of the companies that sell compute. NVIDIA, Google, Microsoft, TSMC and others that bought shares in the private placement are both investors in Anthropic and sellers of chips and cloud services to it. According to Anthropic’s November 2026 prospectus, a little over 40% of its revenue in the first nine months of 2026 came through the marketplaces of major cloud providers.
Such contracts support the revenue outlook of the suppliers, but how much of that turns into profit depends on each company’s capital spending, utilization and pricing. Anthropic’s own compute costs fell sharply as a share of revenue, yet its core business still posted an operating loss.
Over the next five to ten years, where the profits end up is likely to depend on conditions. If compute stays scarce, chip and cloud companies will find it easier to keep pricing power. If competition between models pushes prices down, the benefits are expected to shift toward the businesses and people who use AI.
Even then, companies with their own data and direct customer relationships are more likely to keep the efficiency gains as profit. In a June article, Sekahan argued that when AI’s benefits spread widely, the advantage of using it can become standard equipment for everyone and get competed away on price.
Related article
Who Really Profits From OpenAI and Anthropic IPOs?
Shareholder votes, and the checks held by partners and employees
Public investors’ shares do carry votes, but the prospectus lists as a risk that this structure may effectively leave public shareholders with little ability to influence important decisions. Owning shares and making decisions are two different things.
Some investors online argued that insiders and early backers were selling their shares to the public. In fact, most of the shares sold in the IPO were new shares from the company, and existing shareholders sold about $2 billion. Most shares that individuals buy through a broker are bought on the market from other shareholders, and that money does not go to the company.
Those without votes have other kinds of leverage. Amazon and Google sell compute to Anthropic and are also channels for selling its products, so long-term contracts and supply terms can shape the options open to the company. Employees are also set to receive a vote that acts as a check in limited cases, such as when a board committee does not agree with a director the trust has chosen.
The current structure is designed to stop a shift to a profit-first board. The same structure also means that the judgments of the founders and the trust are hard to change from outside. With the question of who decides now settled, the next question is who checks those decisions, and how.
Can Anthropic keep spending on safety?
The prospectus lists decisions in which the company gave up revenue for safety. It says its highly capable model “Mythos” was first released only to a limited set of partners, and that it chose not to enter commercially attractive areas such as image and video generation in order to put compute toward research and safety. The company lists among its risk factors that such choices “may not maximize stockholder value.”
In a CNBC interview on the day of the listing, Amodei called the relationship between growth and safety “one of the central tensions at the heart of the company.” He argued that a company that builds safely and earns the trust of its customers will also do best over the long run. He did not give a figure for how much the company spends on safety.
Outside assessments also need to be read with attention to who pays and what they cover. METR, a research group that evaluates AI capabilities, published a report on November 5, 2026, on how much AI is speeding up research and development inside Anthropic. It said it received no payment from Anthropic for the assessment, but did receive free usage of AI models and work computers. Much of the evidence METR considered important was redacted at Anthropic’s discretion.
Anthropic’s Responsible Scaling Policy, its internal rules on what safety measures are required as AI capabilities increase, was fully revised about nine months before the listing, so the IPO cannot be said to have changed it. The question is what comes next. The trust’s power to choose directors does not guarantee a safety budget or the delay of a release. With huge payment commitments and quarterly earnings ahead, the test is how long the company can keep putting safety ahead of revenue.
Price moves spread widely, while the power bill lands locally
Joining MSCI’s indexes links Anthropic’s share price to the savings of people who never chose to buy the stock. How much index-tracking money will actually buy in is not yet known. The company has contributed shares to a public-benefit subsidiary, and its co-founders have pledged to give 80% of their holdings to charity, according to its November 2026 prospectus, but how that reaches society will depend on how the money is used.
The burdens, on the other hand, fall on specific places and people. In the base scenario of the International Energy Agency’s 2025 report Energy and AI, electricity consumption by data centers worldwide in 2030 will be more than double its 2024 level. Anthropic’s prospectus also lists as risks regulation over where data centers can be built and local opposition to construction.
A concrete example is the site of a former aluminum smelter in Kentucky. According to TeraWulf’s November 2026 filing with the SEC, the data center company signed a 20-year lease in July with Anthropic for a large data center there. In an August 2026 order, the Kentucky Public Service Commission approved the contract to supply power to the facility. Both plans were underway before the IPO.
Under that power contract, TeraWulf bears the cost of electricity bought on the market and of transmission, and the commission found that it includes safeguards to keep costs from being passed on to existing customers. But concerns raised by residents at public hearings, such as water use and noise, remain a separate matter from the review of electricity rates.
Grid costs and the option of building abroad
As the Kentucky case shows, raising money does not decide on its own where computing facilities can be built or who pays for the grid. Shareholders and executives cannot settle those questions by themselves. In the US, the Senate passed a bipartisan bill on November 19, 2026, according to Senator Chris Coons’s office, that aims to speed up grid construction while keeping data centers’ power costs from being shifted onto household electricity bills. It still needs to pass the House and is not yet law.
Building abroad depends on other factors as well. In its pre-IPO filings, Anthropic listed the concentration of its computing facilities in the US as a risk and said it is expanding sites outside the US to reduce that concentration and support growth overseas.
Trump, however, said at his November 20 press conference that he wants to impose a heavy tax on American companies that build data centers abroad. No such policy has been set out, and it is not clear who would be covered or whether it would extend to renting overseas cloud capacity or facilities.
Even if it becomes policy, building your own facilities and using a local cloud provider’s or a partner’s facilities may be treated differently.
At the same time, the US government is promoting a program to export American AI technology to other countries as full packages. According to the International Trade Administration’s November 16, 2026 announcement, among the 33 packages the Commerce Department designated was a proposal for Japan from AMD and NTT. The designation means the package appears on a list the government presents to overseas buyers. It does not mean construction, a contract or a tax exemption, and it is not an Anthropic project.
Effects on workers, and preparing for a major accident
The effect on workers is still unclear. In Anthropic’s own analysis, there was no clear rise in unemployment even in jobs most exposed to AI. There are signs that young people are finding it harder to get hired, but the result is weak and hard to tell apart from chance.
Who would bear the cost of a major accident is also unclear. The prospectus itself says that if AI causes a major accident, the insurance industry as a whole may not be able to cover it.
Japanese Reactions to Anthropic’s IPO
These are five posts on X that we reviewed for this article. They are not a measure of Japanese public opinion. Widely shared posts by individuals in Japan focused mainly on the share price, the market and buying into the IPO; we did not find widely shared posts by individuals discussing the voting structure, safety spending or data center power.
A user posting as Ryo:
ๅคงๆJTCๅคๅใ ใใฉใCopilotใจใClaudeใใชใใจใใใฏใไปไบใซใชใใชใใ็คพๅ ใฎไฝฟ็จ้ใ่ฆใใจClaude Fableใฎๅพ้่ชฒ้ใงใๆฎ้ใซไธไบบๆ30ไธไฝไฝฟใฃใฆใใไฝฟใ้ใใฆใใใใๅ้กใซใชใใใใชๆใใใใใโฆ
— Ryo (@Ryo_Ponta01) November 18, 2026
I work at a big, traditional Japanese company, and work is basically impossible now without Copilot or Claude. Looking at our company’s usage, people are easily using around ยฅ300,000 a month each on pay-as-you-go Claude Fable. It almost feels like it’s going to become a problem soon because we’re using it too much.
Anthropic is about to go public, and I suspect we’ll now move into a phase where the profitability and business viability of AI investment get confirmed.
A user posting as James Bonnou:
ใขใณใฝใญใใใฏIPOๅฝ้ธ
— James็ ฉๆฉใใ๏ธๆฅ็ตใใใผๆฒ่ผไบๅฎใ๏ธ (@james_bonnou) November 19, 2026
1ๆ ช83ใใซใชใใงใใญใ
5ๆ ชใ ใ็ณใ่พผใใงๅ จใฆๅฝ้ธใ
ๆฅฝๅคฉ่จผๅธใงใใ
ใฟใใช่ฒทใใพใใใ๏ผ
Got shares in the Anthropic IPO
So it’s $83 a share.
I applied for just 5 shares and got all of them.
Through Rakuten Securities.
Did everyone buy some?
Emin Yurumazu, an economist who comments on markets in Japan:
ใขใณใฝใญใใใฏไธๅ ดใงใใในใใใฏใฏๅฉใใใชใใใใงใใใพใใ้ ใใฃใใใญใSpaceXใฎๆใฏ็ใไธใใฃใใฎใงใใใไปใฎ็ธๅ ด็ฐๅขใฏ็ธๅฝๅณใใใ
— Emin Yurumazu (ใจใใณใฆใซใใบ) (@yurumazu) November 19, 2026
Even Anthropic’s listing doesn’t seem to be saving the Nasdaq. Well, it came too late, didn’t it. Things got pretty excited around SpaceX, but the current market environment is very tough.
A user posting as Sho’s Investment Info:
ใขใณใฝใญใใใฏไธๅ ดใงๅๅฐไฝๆ ชโฌ๏ธใชใ
— Shoโ๏ฝๆ่ณๆ ๅ ฑๅฑ (@Sho54639108) November 19, 2026
ๅนดๅ ไปใซๆๆใใใใ๏ผ
If chip stocks go โฌ๏ธ even with the Anthropic listing
is there any other catalyst left this year?
A user posting as Rize:
ใขใณใฝใญใใใฏใฎๆ ชใๆใฃใฆใฆ้ข็ฝใใจใใใฏใ30ๅนดใงไฝๅๅใซใใชใฃใฆใใ๏ผApple1500ๅ๏ฝNVIDIA3500ๅใจใใ ใใๅ จ็ถ็พๅฎ็๏ผใชใใจไปฅไธใซใใใฎ้ ใซใฏ่ฒจๅนฃใใฎใใฎใฎไพกๅคใจใ็คพไผๆง้ ใใฎใใฎใๅคใใฃใฆใใใชไธใซใใใฎๅ ๅถใซใชใฃใฆใใใฃใฆใณใใ https://t.co/zUW0UY1HuZ
— ใใใ / RizeJP (@_mumumu) November 20, 2026
What’s interesting about holding Anthropic stock isn’t just that it will probably have gone up thousands of times in 30 years (Apple went up 1,500x and NVIDIA 3,500x, so that’s completely realistic). It’s that by then the value of money itself and the structure of society itself will probably have changed, and Anthropic will probably be the one to blame.
The Questions Beyond the Share Price
The money raised, and the control that stayed
Was it a mistake to go public in a falling market? There are two answers. The company received a large amount of cash at the moment of the IPO. From now on, the share price will keep asking, every trading day, when and for whom the enormous investment in AI will pay off.
The IPO spread Anthropic’s shareholders around the world, yet the majority of the general vote stays with the founders’ company and the power to elect most of the board stays with the trust. Meanwhile, the burdens of electricity use and changes in work fall on the communities that host data centers and on workers. Separating the people who provide the money from the people who decide was also a design meant to shield development from short-term profit pressure.
Where profits settle over the next five to ten years is likely to depend on the supply and demand for compute and on competition between models.
What will matter then is less the rise and fall of the share price than whether, even in a year when revenue growth stalls, the company can keep putting money and time into safety testing and delayed releases, and whether the costs of data center power and grid upgrades can be kept off the bills of nearby residents and households. In a system where those who pay, those who decide and those who bear the burden are different people, the answers will have to be checked through disclosure, legislation and agreement with local communities.
Frequently Asked Questions
When did Anthropic go public, and at what price?
Anthropic began trading on the Nasdaq Global Select Market under the ticker ANT on November 19, 2026, after pricing its shares at $83 on November 18, according to the company’s announcement. The stock opened at $111.50 and closed at $97.17 on its first day, according to CNBC.
Does Anthropic lose the money it raised if its stock price falls?
No. According to Anthropic’s prospectus (Form 424B4, November 2026), the company expects to receive about $117 billion from its public offering and private placement after fees and expenses, and that cash belongs to the company once the sale closes. Later trading moves money only between buyers and sellers of the shares.
Who controls Anthropic after the IPO?
The founders and the Long-Term Benefit Trust do. According to Anthropic’s prospectus, a company owned by the seven co-founders holds a fixed 50.1% of the vote on general matters, and the trust elects a majority of the board. Amazon and Google own about 15% and 13% of all shares, but their shares generally carry no vote.
How did Japanese social media react to Anthropic’s IPO?
In the posts on X we reviewed, people in Japan focused mainly on the market: one investor reported getting shares at the IPO price through a Japanese online broker, an economist said the listing did not save the Nasdaq, and another investor asked what could lift chip stocks if even this could not. One office worker described heavy use of Claude at work, and another investor said Anthropic could change society itself over 30 years. These few posts are not a measure of Japanese public opinion.
Sekahan on YouTube
We publish video summaries of articles like this one, along with short clips built around Japanese reactions.
Reference Links
- Anthropic Announces Closing of Initial Public Offering๏ฝAnthropic
- Anthropic, PBC Prospectus (Form 424B4)๏ฝSEC
- Anthropic Prices Initial Public Offering๏ฝAnthropic
- US stock market close, November 20, 2026๏ฝAP News
- Anthropic IPO live updates (November 19, 2026)๏ฝCNBC
- CNBC Exclusive Transcript: Anthropic Co-Founders Dario and Daniela Amodei๏ฝVersant
- MSCI Global Standard Indexes: Early Inclusion of Anthropic๏ฝMSCI
- Delaware General Corporation Law, Subchapter XV (Public Benefit Corporations)๏ฝDelaware Code
- Responsible Scaling Policy๏ฝAnthropic
- Anthropic AI R&D Acceleration Assessment๏ฝMETR
- Labor market impacts of AI๏ฝAnthropic
- Case No. 2026-00115 Order๏ฝKentucky Public Service Commission
- WATCH: Trump on data centers: ‘Take advantage, don’t shut them off’๏ฝPBS NewsHour
- LIVE: Trump holds press briefing after historic meeting with Xi and Putin (full press conference)๏ฝNBC News
- Trump’s Marathon Press Conference (November 20, 2026, unofficial transcript)๏ฝTommy Christopher
- Anthropic, PBC Free Writing Prospectus (EEA prospectus, November 11, 2026)๏ฝSEC
- Department of Commerce Designates SI Export Packages, Announces Next Phase of American SI Exports Program๏ฝInternational Trade Administration
- Designated Consortia๏ฝAmerican SI Exports Program
- Senator Coons statement on final passage of permitting reform bill๏ฝOffice of Senator Chris Coons
- TeraWulf Inc. Form 8-K (November 4, 2026)๏ฝSEC
- Energy and AI๏ฝIEA


