AI Trends

Anthropic Goes Public in the Largest-Ever IPO, Stirring Controversy Over Special Equity and Computing Power Procurement Arrangements

Phoenix TechnologyUpdated Nov 20 14:03

AI summary

Anthropic officially went public on November 19, closing with a market capitalization of approximately $1.77 trillion. A special class of F shares gives the founding team 50.1% of the total voting power. The company allowed existing shareholders and employees to sell up to 315 million shares on the first day of trading, while disclosing $900.9 billion in long-term computing power and chip procurement commitments. On the day of the listing, nearly 100 AI researchers and entrepreneurs published an open letter titled “The MisAnthropic Statement,” sparking debate over the company’s safety stance and commercial expansion.

Why read this

SelectedAI score 84

The article examines special voting rights, employee share sales, and circular financing arrangements to reveal the control and supply-chain strategies behind Anthropic’s public listing.

Topics

Tags

Body · AI translation

Beating OpenAI to the Biggest IPO in History, Anthropic Has Angered Half of Silicon Valley TechFox November 20, 2026, 13:11:36 Updated November 20, 2026, 13:14:39

On November 19, Anthropic, the AI giant founded just five years ago, officially went public.

In one stroke, it shattered the previous record set by SpaceX, becoming the largest IPO in the history of human commerce.

Even more astonishingly, its market capitalization briefly surged past $2 trillion after trading opened, before closing at about $1.77 trillion.

In a surreal twist, on the very day the company rang the opening bell, nearly 100 of the world’s leading AI researchers and entrepreneurs jointly published a combative open letter titled “The MisAnthropic Statement.”

The letter was sharply worded and quickly sparked fierce clashes on social media, even escalating into a public war of words.

The bitter divide in public opinion, combined with first-day selling pressure, sent Anthropic’s stock on a roller-coaster ride.

It surged by more than 37% at one point during the session, then steadily surrendered its gains, ultimately closing up about 17%.

But a look through Anthropic’s prospectus reveals that something even more interesting than its share price is the institutional structure of the listing itself.

From its ownership structure to arrangements allowing employees to cash out, Anthropic did not play by Wall Street’s traditional rules at all.

The Amodei siblings and five other co-founders hold a single special Class F share through Founder LLC, and that one share carries 50.1% of the company’s total voting power.

Meanwhile, the seven technology giants—including Nvidia, Microsoft, and Google—that contributed $26 billion to the private placement received only non-voting Class N shares.

The Class A shares available to ordinary investors on the secondary market nominally carry voting rights, but cannot challenge the founding team’s voting advantage.

Simply put: you provide the money and the computing power, but the founders alone decide what the models are trained toward and where the company is headed.

Why was control locked down so tightly?

Because the “boardroom coup” at OpenAI that caused an uproar three years ago taught all of Silicon Valley a lesson.

In 2023, Sam Altman was briefly removed by the board, only to return after pressure from employees, investors, Microsoft, and others.

Anthropic clearly did not want history to repeat itself, so before going public, it embedded the founding team’s control directly into the ownership structure.

Anthropic also made special arrangements for employee liquidity in this listing.

When traditional technology companies go public, shares held by executives and employees are usually subject to a six-month lockup and cannot be sold until it expires. This time, however, Anthropic created an unusually generous exception for employees.

It allowed existing shareholders and employees to sell up to 315 million shares on the very first day of trading and scheduled 23 separate liquidity windows during the first year.

Bear in mind that Silicon Valley compensation packages for recruiting a top large-model researcher now routinely start in the tens of millions of dollars.

Anthropic made no empty promises: on the first day of trading, it turned paper wealth directly into hard cash. According to reports, its 5,300 employees instantly included more than 600 billionaires and 2,800 multimillionaires.

If Anthropic’s equity design demonstrates its shrewdness, its alignment of interests with Silicon Valley’s technology giants reveals an even greater ambition: securing in advance the scarcest infrastructure of the AI era.

This is also the most controversial part of the entire prospectus.

Many financial media outlets used the phrase “moving money from the left hand to the right”—what Wall Street calls “circular financing.”

Looking at the flow of funds, seven giants, including Nvidia, Google, AMD, and Microsoft, spent $26 billion subscribing to Anthropic shares.

Anthropic then turned around and disclosed long-term commitments of up to $900.9 billion for computing capacity and chip purchases in its prospectus, including more than $230 billion in cloud-service bills to Google and Amazon alone.

The money flowed out of the giants’ investment divisions, made a circuit, and then returned as orders on their revenue statements.

Viewed from another angle, however, this is not entirely a case of capital circulating in place.

After all, Anthropic has a very real annualized revenue run rate of $81 billion and more than 10,000 Fortune 500 corporate customers.

Fundamentally, its willingness to bring nearly every major Silicon Valley chip and cloud company onto its shareholder register at any cost represents a bid to secure its place in the supply chain now that the large-model arms race has entered deeper waters.

When competition among next-generation models comes down to chip allocations, wafer capacity, and electrical power, binding the supply-chain giants to the same ship through equity stakes means locking in the scarcest infrastructure resources ahead of time.

At this scale, no giant can afford to let Anthropic lose.

Anthropic is sprinting ahead under nearly $1 trillion in infrastructure commitments while simultaneously occupying the moral high ground and urging the entire industry to “slow down for safety.”

To the open-source community and its peers, this is tantamount to “booking the entire playing field and buying up every pair of cleats, then announcing that, for everyone’s safety, the game must henceforth be played by my rules.”

To understand this sense of dissonance, we need to turn the clock back to 2020.

That year, Dario Amodei, then OpenAI’s vice president of research, left the company together with his sister Daniela, who served as vice president of safety and policy, and more than a dozen core researchers. They founded Anthropic in 2021.

Amodei had become increasingly concerned about the company’s growing tilt toward commercialization, especially after Microsoft invested in OpenAI.

But five years later, the plot has taken a somewhat ironic turn.

OpenAI, once criticized for pursuing commercialization too aggressively, is still preparing to go public.

Anthropic, by contrast, despite holding the banner of safety aloft, has already beaten it to the Nasdaq opening bell, with the siblings’ combined net worth surpassing $52.5 billion on the first day of trading.

The seven co-founders have publicly pledged to donate 80% of the value of their personal equity holdings to public-interest and charitable causes, in an attempt to preserve the last vestige of their idealism.

The change behind this may reflect more than just the founders’ choices.

When training frontier models requires tens of billions of dollars, and computing power, talent, and commercial scale all become barriers to competition, only by first transforming itself into a trillion-dollar commercial behemoth can a company earn a seat at the table.

Anthropic has already proven that it possesses not only technological ideals, but also the ability to turn technology into revenue.

Anthropic’s prospectus reports a net loss of $104 billion for the first three quarters, an alarming figure at first glance.

But about $94.3 billion of that came from fair-value accounting adjustments to financing instruments rather than actual operating losses.

On the adjusted basis disclosed by the company, operating profit reached $4.08 billion, while its annualized revenue run rate hit $81 billion.

While rival OpenAI is still competing for consumer scale, Anthropic has gone all-in on enterprise engineering and AI agents.

With tools such as Claude Code directly taking over enterprise software-development pipelines, it has secured more than half of the Fortune 500, nearly 2,000 customers worth at least $1 million each, and an astonishing net revenue retention rate of 900%.

Anthropic’s bell-ringing may signal that the large-model industry has entered a new phase.

In the past, companies competed over whose models had more parameters, higher benchmark scores, and more dazzling demos.

Anthropic has already been the first to make this path work, so the pressure now naturally shifts to competitors such as OpenAI.

Looking back, the Amodei siblings left OpenAI five years ago because they feared the over-commercialization of AI.

Five years later, do you see Anthropic as “the dragon slayer who ultimately became the dragon,” or as making the inevitable choice required to survive in the era of large models?

Source

Phoenix Technology