After Anthropic Goes Public, Can Its Values Withstand the Pressures of Capital and National Security Interests?
Anthropic debuted on Nasdaq under the ticker ANT at an offering price of $83, valuing the company at approximately $1.5 trillion. The article focuses on the governance structure through which its founders retain 50.1% of the voting power via Class F shares, as well as the company’s disagreements with the U.S. defense establishment over military uses of Claude. With its computing power and infrastructure commitments approaching $900 billion, whether Anthropic can uphold its AI safety principles amid commercial expansion has become the central question.
Why read this
SelectedAI score 91
Examining equity governance, reliance on capital, and controversies over military applications, the article highlights the structural tension between Anthropic’s safety values and its commercial expansion.
Topics
Body · AI translation
From leaving OpenAI to a $1.5 trillion IPO; from AI safety idealists to a commercial giant swept along by capital, computing power, and the U.S. national security establishment. What Anthropic truly needs to answer may not be how much more powerful its models can become, but whether it can still uphold what it once believed in.
On November 19, 2026, the artificial intelligence industry reached a landmark moment.
Anthropic officially debuted on Nasdaq under the ticker symbol ANT. The developer of the Claude large language model had previously priced its shares at $83, above the original range of $70 to $78. Based on its post-offering share count, the company was valued at approximately $1.5 trillion, while the IPO also set a record for the largest fundraising in the history of U.S. capital markets.
The market’s enthusiasm was unmistakable. Institutional orders had at one point reached roughly six times the size of the offering, with some reports putting the figure at nearly seven times.
Judging solely by the figures assigned by the capital markets, Anthropic is undoubtedly one of the most dazzling technology companies of this era. It not only represents a high point in the commercialization of generative AI, but also signals that artificial intelligence companies are beginning to redefine the limits of capital-market valuations at a pace unimaginable even to internet giants and traditional energy behemoths.
Yet behind this grand listing ceremony lies a question that investors should seriously consider:
This is not an abstract philosophical question. Anthropic’s development over the past several years has already provided a highly complex answer.
01 From Baidu to OpenAI: What Dario Amodei Truly Believes In Is the Scaling Law
To understand Anthropic, one must first understand its founder, Dario Amodei.
Many people interpret Anthropic’s story as one of a group of AI safety idealists leaving OpenAI to strike out on their own. But tracing the timeline further back reveals that the company’s technical DNA actually comes from a highly pragmatic engineering logic.
In 2014, Amodei joined Baidu’s U.S. research team and participated in research related to deep learning. In the technical environment shaped by Andrew Ng and others, he gradually came to recognize a principle that would later transform the entire AI industry: as data, computing capacity, and model scale continue to expand, the capabilities of artificial intelligence systems may leap beyond conventional expectations.
This is what later became widely discussed as the scaling law.
Pile on data, parameters, and computing power, and model capabilities improve as scale increases.
This logic later became an important technical foundation for the development of the GPT series. Amodei joined OpenAI in 2016, becoming one of its core researchers. He participated in the development of models including GPT-2 and GPT-3 and also helped advance reinforcement learning from human feedback.
Judging from his technical résumé, he is not someone who talks only about safety without understanding business and engineering. Quite the opposite: he understands the technical path of large models profoundly and knows what resources are required to prevail in this race.
The problem was that he and OpenAI’s management gradually developed disagreements over who should control this technology, how quickly it should be commercialized, and how much decision-making power safety principles should have.
In December 2020, Dario, his sister Daniela Amodei, and five other co-founders left OpenAI and subsequently founded Anthropic.
These seven people formed Anthropic’s original founding team.
Their departure, of course, cannot simply be explained as a personal power struggle. In a later interview with Stanford Graduate School of Business, Daniela explained that the team wanted to build a company that genuinely placed AI safety and responsibility at the core of the organization.
But another reality cannot be ignored: OpenAI was already moving steadily toward commercialization at the time, the strategic value of GPT technology was rising, and conflicts among research priorities, organizational power, and commercial interests were becoming increasingly acute.
From the day it was founded, Anthropic carried a distinctive identity: it was both a major inheritor of OpenAI’s technical approach and a rebellion against OpenAI’s model of governance.
And in the era of large models, survival means enormous amounts of funding, computing resources, chip supplies, and cloud-computing contracts.
Several stories from Anthropic’s early fundraising are worth noting.
First was FTX.
In 2021, Sam Bankman-Fried’s Alameda Research invested approximately $500 million in Anthropic, when the company was valued at only around $4 billion. The investment later became one of the most dramatic cases in the history of technology investing.
Following Anthropic’s listing, had the stake been retained, its theoretical value might have reached tens of billions of dollars. In reality, however, after FTX went bankrupt, the stake was sold in 2024 for approximately $1.3 billion to repay creditors.
There was another noteworthy connection behind the investment: Bankman-Fried had long been involved in the effective altruism movement, and Anthropic’s emphasis on AI safety and humanity’s long-term interests overlapped with that intellectual framework at the time.
Next was Eric Schmidt.
The former Google CEO later became an important early investor in Anthropic. According to media reports, Schmidt’s connection with Amodei began with an introduction by Cami Clark, who later became Amodei’s wife.
Schmidt invested during Anthropic’s early funding stage in 2021. As the company’s valuation soared, the paper return on this early investment also became astonishing.
These stories illustrate that Silicon Valley’s capital networks are not merely cold, impersonal institutional investments. Technical talent, personal relationships, trust among investors, and a shared judgment about the future direction of technology can all determine a startup’s early fate.
But normal relationship networks must be distinguished from corporate governance problems.
Dario and Daniela are siblings, and Daniela serves as Anthropic’s president and chair of its board, a publicly disclosed management structure. The fact that they are related does not, by itself, establish that the company engages in improper transfers of benefits. As for online claims that their brother-in-law serves as the head of safety, there is currently insufficient reliable public evidence to support them, and they should not be treated as fact.
What truly merits discussion is how Anthropic integrates capital, strategic partners, and the founding team’s power into a single governance system.
Giants such as Amazon, Google, and Microsoft are not only investors but also have deep cloud computing and infrastructure partnerships with Anthropic.
This is a mutually dependent business relationship.
But it also raises a question: when investors are simultaneously suppliers, cloud computing platforms, and strategic partners, how much independence does Anthropic actually have?
03 50.1% of the voting power: Who oversees the guardians of its values?
If the early funding stories are still merely common features of Silicon Valley’s startup ecosystem, Anthropic’s IPO governance structure is the aspect of this listing that most deserves close examination.
According to the prospectus filed with the U.S. Securities and Exchange Commission, Anthropic has adopted a multilayered shareholding and governance arrangement.
The most striking element is its Class F stock.
This is a special class of founder shares held by Founder LLC, an entity established by the founders. Dario, Daniela, and five other co-founders are its initial members.
These shares carry no economic rights in the ordinary sense, yet they hold voting power equal to 50.1% of the company’s total voting power.
More importantly, the prospectus explicitly stipulates that Class F’s voting power will remain at 50.1% of the total. Even if the company issues additional Class A or Class N shares in the future, this voting power will not be diluted.
At the same time, Anthropic has established the Long-Term Benefit Trust. Through Class T shares, the trust appoints special directors who will constitute a majority of the board.
In other words, Anthropic is not a publicly traded company in the traditional sense, where public shareholders determine corporate control in proportion to their capital holdings.
It has effectively created two interwoven systems of power: the founding team retains an absolute voting majority at the shareholder level through Class F, while the Long-Term Benefit Trust exercises significant power at the board level.
There is a rationale for this design.
The development of AI technology may create extraordinarily complex long-term risks. If the company must answer to the capital markets for short-term profits every quarter, management may be forced to sacrifice long-term investment in safety. A special governance structure can prevent short-term capital pressures from exerting excessive influence over the company’s technological direction.
But this is precisely where the problem lies.
If the founders possess control that cannot be diluted over the long term, who will oversee them?
If public shareholders cannot change the company’s strategy through ordinary shareholder votes, are they buying an economic interest in the company or placing their trust in the founding team’s values?
Anthropic’s Long-Term Benefit Trust attempts to address this issue, but the trust itself must also withstand tests of its independence, accountability mechanisms, and actual decision-making capacity.
An idealistic governance structure does not mean that governance problems automatically disappear.
This is especially true when a company is valued at $1.5 trillion, as every major decision by its founding team could affect trillions of dollars in capital interests.
04 The Pentagon incident: Upholding principles or losing the market?
If corporate governance represents Anthropic’s internal tensions, its conflict with the U.S. Department of Defense brought those tensions into the real world.
In 2026, Anthropic and the U.S. Department of Defense had serious disagreements over the military use of Claude.
The Pentagon sought authorization for a broader range of lawful uses. Anthropic, however, insisted on retaining these two red lines.
From the company’s perspective, this is not difficult to understand.
Anthropic has long emphasized that advanced AI should not be permitted in every scenario merely because its use is technically possible. With autonomous lethal weapons in particular, once a system makes life-or-death decisions without effective human control, accountability and risk management become major concerns.
But the U.S. government’s reasoning is equally pragmatic.
The military argues that if an operation is lawful, the government should not allow a private technology company to determine, through product restrictions, which tools the country may use.
The two sides ultimately failed to reach an agreement.
According to a ruling issued by the U.S. Court of Appeals for the District of Columbia Circuit on September 25, 2026, the court upheld the Defense Department’s decision to exclude Anthropic from the relevant supply chains. The court found that Anthropic’s restrictions on AI uses could pose a supply chain security risk and rejected the company’s related challenges.
The most dramatic aspect of the matter is this:
Anthropic originally broke away from OpenAI precisely because it was unwilling to let commercial interests override AI safety. Now, having actually applied that principle to U.S. government military procurement contracts, it finds itself confronting another, far more powerful system.
It can reject certain commercial contracts, but it cannot demand that the national security establishment operate according to the values of a private company.
When artificial intelligence becomes national security infrastructure, who should decide the ethical boundaries of technology companies?
05 The Real Test Is Not Valuation
Anthropic’s public listing means that it has grown from an AI laboratory founded by researchers into one of the world’s most important technology businesses.
Its commercial achievements are indeed remarkable.
Claude has established a competitive advantage in programming, enterprise applications, and complex knowledge work. The company’s revenue is growing rapidly, and the market has placed extremely high expectations on its future commercialization potential.
But behind the capital market figures lies another set of numbers.
According to disclosures in its prospectus, Anthropic’s revenue is expanding rapidly, but its commitments to computing power, cloud computing, and infrastructure have also reached astonishing levels. Updated offering documents show that the company’s future computing and infrastructure-related commitments are approaching $900 billion.
This means it can no longer operate on idealism alone.
It must continually expand its customer base, secure more computing power, and release more powerful models, while also proving to investors that these investments can ultimately be converted into profits.
In this process, will safety principles become a cost of commercial competition?
When competitors are willing to loosen restrictions on model usage, can Anthropic still maintain its boundaries?
When major cloud service providers demand larger-scale commercial partnerships, can the company still exercise independent judgment?
When the U.S. government makes new demands, will the founding team be willing to bear the cost of saying no?
These are the questions Anthropic truly needs to answer.
There is another detail that cannot be ignored: amid media coverage concerning Nvidia and its founder Jensen Huang, there has been speculation in the market that Anthropic supplied information to the media and sought to influence public opinion. To date, however, there is insufficient reliable public evidence to substantiate claims that Anthropic systematically manipulated the media or specifically conducted a public-opinion campaign against Jensen Huang. Such claims should not be conflated with facts supported by court documents and the prospectus. For a company that espouses transparency and responsibility as core values, its media relations should likewise be subject to public scrutiny.
Anthropic is not a perfect company, and neither is OpenAI.
In this sense, Anthropic’s public listing is not only a capital-markets event involving an AI company, but also a business case study in how technological ideals confront real-world interests.
A company can declare its values at its founding, enshrine safety in its charter, design a complex trust structure, and reject contracts that appear highly profitable.
But genuine values have never been merely a sentence written on a company’s website.
They are what a company remains willing to pay a price to uphold when conflicts of interest arise.
Anthropic has already demonstrated through its technology that it is capable of becoming a major player in the global AI industry.
Next, it must prove something else to the market:
When it has a valuation of $1.5 trillion, vast commercial interests, and increasingly significant political influence, can it still remain true to what it believed when it left OpenAI?
That may be the real variable determining just how far Anthropic can go.
This article is from the WeChat public account “Jinghe Artificial Intelligence,” by Jinghe, and is published by 36Kr with authorization.
AI Trends