Anthropic Lists on Nasdaq Under ANT, Record-Breaking IPO Sparks Controversy in the AI Industry
Anthropic listed on Nasdaq under the ticker “ANT,” offering 980 million shares at $83 each for a public offering totaling $81.34 billion. The article says several technology companies separately subscribed for $26 billion in Class N shares, while Anthropic disclosed $900.9 billion in computing and infrastructure commitments. On the day of the listing, nearly 100 AI researchers, company founders, and scientists jointly signed the “Statement of Disgust with Anthropic,” questioning the company’s safety narrative, expansion of power, and governance structure.
Why read this
SelectedAI score 94
Drawing on the prospectus and an open letter, the article highlights the tensions among Anthropic’s public offering and fundraising, its control structure, and its safety mission.
Topics
Body · AI translation
Just now, Anthropic officially debuted on Nasdaq under the ticker symbol “ANT,” setting a record for the largest IPO in global history.
The AI giant behind Claude ultimately issued 980 million shares at $83 each, raising $81.34 billion in the public offering and surpassing the roughly $75 billion record set by Elon Musk’s SpaceX in June this year.
In addition to the public offering, seven technology giants—Nvidia, Microsoft, Google, TSMC, AMD, Broadcom, and Micron—also agreed to purchase an additional $26 billion of Class N shares at the offering price. Excluding approximately $1.96 billion in sales of existing shares, Anthropic itself expects to receive about $105.4 billion upon completion of the two transactions.
However, on the very day Anthropic celebrated its historic listing, nearly 100 leading AI researchers, company founders, and scientists jointly published an open letter titled “The MisAnthropic Statement,” publicly opposing the company.
The signatories include effective accelerationists who advocate advancing AI at full speed, including Guillaume Verdon and prominent hacker George Hotz, as well as safety advocates who have long warned that AI could destroy humanity.
Hugging Face co-founder Thomas Wolf, xAI co-founder Igor Babuschkin, Turing Award winner Judea Pearl, a Nobel laureate, and several Fields Medalists also appear among the signatories.
These people, who ordinarily clash over virtually every AI-related issue, have now come together because of Anthropic’s listing and reached a shared conclusion: Anthropic is not a safe AI company, nor should it continue expanding its control over AI’s future in the name of protecting humanity.
What exactly does Anthropic’s listing mean? Why has it drawn opposition from so many people? And what details are buried in its more than 300 pages of IPO filings? As ANT begins trading on the U.S. stock market tonight, APPSO sorts it all out.
A 5,300-Person Company Is Signing More Than $1 Trillion in Orders
To understand why this backlash against Anthropic has erupted so intensely in recent weeks, we first need to examine how its model releases have changed over the past year.
Over the past year, Anthropic has almost completely shattered the stereotype that it is a “conservative follower,” repeatedly pulling ahead of its competitors in both the pace of model iteration and product dominance.
In June, Anthropic broke with its existing Opus/Sonnet naming system and launched the new flagship series Claude Fable 5, along with Mythos 5, a guardrail-free version available only to vetted cybersecurity and biosecurity researchers.
It released Opus 5.5 in September, then moved first again on November 9 with Claude Fable 5.2 and Mythos 5.2, maintaining its lead in long-horizon agentic coding and complex knowledge-work benchmarks.
It even disrupted OpenAI’s model-release cadence. OpenAI hurriedly responded a week later with GPT-6.2 Astra and Sol, only to face relentless criticism on social media and a decline in public sentiment.
With Claude Code—a product that exploded in popularity this year and was subsequently imitated by numerous software products—programmers can use natural language to direct an Agent to complete all kinds of coding tasks. Anthropic has achieved what is arguably the fastest adoption of enterprise technology in human history.
Revenue also climbed from $41 million in 2023 to $4.6 billion in 2025, reaching $33.2 billion in the first three quarters of 2026 alone.
Revenue for the third quarter of 2026 alone reached $17 billion, keeping Anthropic ahead of OpenAI, while more than half of the Fortune 500 have become paying customers with contracts worth at least six figures.
To sustain this race toward superintelligence, Anthropic disclosed in its prospectus as much as $900.9 billion in commitments to purchase and lease computing capacity and infrastructure, of which $451.6 billion will be spent by the end of 2030.
The central narrative underpinning this enormous spending commitment is what Anthropic calls the “closed loop”: each generation of Claude models directly assists in building the next, continually shortening R&D cycles through recursive self-improvement and driving parallel gains in model capabilities and revenue.
In the prospectus summary, Anthropic makes no attempt to hide that it regards this recursive dynamic as “one of the company’s most fundamental structural competitive advantages,” and explicitly states that the primary use of the IPO proceeds will be to purchase more computing capacity so it can achieve this breakthrough before its competitors.
During the final push to complete the prospectus, METR, a prominent independent AI evaluation organization, dispatched a team equivalent to 2.5 full-time researchers to Anthropic and produced quantifiable data showing how far “AI accelerating AI R&D” has already progressed.
Over the past year, Anthropic has completed the equivalent of roughly two years of its previous technological progress.
After excluding gains attributable solely to increased training compute, METR estimated that around 80% of the current acceleration in R&D was due to Anthropic’s extensive use of its own AI models in internal research and development. In an internal Slack poll, employees self-reported personal productivity gains of as much as 12-fold.
Although AI currently serves primarily as an assistant to human researchers, METR has already observed early signs of autonomous R&D and believes that no algorithmic, computing, or data bottleneck will be able to halt this accelerating trend over the next 12 months.
Anthropic also included this evidence of a “closed loop in which AI accelerates AI R&D” in its IPO filing, presenting it as a structural competitive advantage as it seeks to raise tens of billions of dollars from public markets.
Yet in the “Risk Factors” section of the very same prospectus, Anthropic clearly set out its own doomsday warnings about the technology.
Frontier models deployed at scale could “highly concentrate power and wealth within and among nations, erode trust, harm society, and undermine geopolitical stability,” while misaligned models could “corrupt institutional decision-making or cause large-scale harm through autonomous actions that users and operators can neither detect nor understand.”
This is precisely the hypocrisy denounced by the “Statement of Disgust with Anthropic”: in public, Anthropic presents itself as Silicon Valley’s only safety-minded company and warns that RSI could escape human control; in the capital markets, however, it touts RSI as a core selling point in order to go public and raise more money.
The Company Is Complicated, and Its People Are Even More Extraordinary
But judging Anthropic solely from its IPO filing as a calculating commercial behemoth may underestimate the complexity within the company.
Over the past two years, New York Times columnist Jasmine Sun conducted in-depth interviews with dozens of Anthropic employees, who call themselves “Ants.” To some extent, the article paints a portrait of Silicon Valley imbued with an almost religious sensibility and a strong sense of fatalism.
In the company’s tightly secured, sparsely decorated San Francisco offices, the bookshelves hold The Making of the Atomic Bomb, CEO Dario Amodei’s Machines of Loving Grace, and Claude’s Constitution, drafted by the company’s in-house philosopher Amanda Askell.
Many young researchers there could become multimillionaires or even billionaires overnight through the IPO. Yet they have no interest in luxury cars or first-class travel, continue to crowd into shared apartments, and voluntarily maintain a “996” work schedule...
To the “Ants,” money is secondary. What is truly pressing is time.
With several recent incidents involving rogue AI agents carrying out cyberattacks, coupled with RSI advancing far faster than expected, Anthropic co-founder Jack Clark and the independent forecasting organization AI Futures Project have both moved their timelines for the arrival of “superhuman AI researchers” sharply forward to 2027.
Image | Over the past several years, some AI researchers have repeatedly brought forward their predictions for the arrival of “superintelligent AI”—from initially believing it would not arrive until 2060 to now believing it could emerge as early as 2027–2028. The horizontal axis represents when the researchers made their predictions. The vertical axis represents the year in which they predicted at the time that such powerful AI would appear. The differently colored lines show how the forecasts of five individuals changed. https://blog.aifutures.org/p/q35-timelines-update-looks-like-the
Since they understand full well that models can easily slip beyond human control once they learn to iterate autonomously, why have these self-proclaimed morally upright idealists not hit the brakes, but instead floored the accelerator?
In 2022, Dario Amodei publicly declared that Anthropic “should not race ahead” and should act as a “fast follower” in safety research. That same year, Anthropic delayed Claude’s release over safety concerns, allowing OpenAI to launch ChatGPT first.
Company lore enshrined this sacrifice as a model of “principles over profits,” but Amodei himself described it as a “one-time-only” decision.
Afterward, he proposed a new theory: the “race to the top.” It rests on three assumptions: competition is inevitable; compared with its “reckless competitors,” Anthropic is humanity’s most responsible guardian; and only by becoming the undisputed commercial and technological winner can Anthropic gain the leverage to shape industry rules.
A former employee used an allegory from Tolkien’s The Lord of the Rings to describe the true mindset of the company’s leadership: whoever creates superintelligence first will don Sauron’s One Ring. That person will command the world’s most dangerous biological and cyber weapons, the secrets to curing every disease, and an army of tireless digital geniuses.
Dario Amodei’s logic is this: since the Ring is destined to be forged, the “good guys” must be the first to put it on.
Another former employee, speaking anonymously, revealed that Dario Amodei once said in an internal discussion that Anthropic might one day become more powerful than governments and would therefore need to establish its own independent relationships with sovereign states.
To prove to the outside world that it would not be corrupted by power, Anthropic deliberately sent a series of costly “moral signals”: it rejected the Pentagon’s request to use Claude for mass surveillance within the United States; barred Chinese entities from purchasing its products; placed 5% of its equity into a public-benefit trust; and devised an exceptionally elaborate five-class common-stock structure in its prospectus.
The Class T shares held by the Long-Term Benefit Trust (LTBT) have the power to appoint and remove a majority of the board, while the Class F shares held by the seven co-founders lock in 50.1% of the total voting power. Neither the Class A shares purchased by public investors nor the Class N shares subscribed to by chip giants can challenge that control.
Yet as the author of the New York Times article points out, the Ring had to be destroyed in Middle-earth precisely because even the purest soul could not wield its power safely.
More importantly, the field of AI safety faces a series of seemingly insoluble scientific black boxes. Humanity still has no technical means of verifying whether the obedience displayed by a superintelligent AI during testing reflects genuine “value alignment” or merely “feigned compliance” under intense competitive pressure.
Likewise, we cannot verify whether a CEO is an altruistic believer in the mission or a performative power seeker. From the dual histories of runaway AI and runaway executives, we can observe only one thing: competitive pressure drives unethical behavior.
The Nasdaq opening bell has already rung for Anthropic. The $90.09 billion compute gamble in its prospectus, the cross-faction denunciations in the joint statement, and the apocalyptic anxiety of young researchers in its offices will all become questions traced in candlestick charts and interrogated by earnings reports every 90 days.
The more strongly Anthropic believes it is charged with saving humanity, the more reason it has to convince itself that it must run faster than every competitor.
Ordinary people naturally want AI to become smarter and cheaper and to do more work for us. But when a company’s sense of mission ultimately becomes its justification for continually expanding its power, that should not mean users must accept all of its judgments unconditionally.
This article originally appeared on the WeChat public account “APPSO.” Author: APPSO, Discovering Tomorrow’s Products. Republished by 36Kr with permission.
AI Trends