Anthropic goes public

Platformer

Source published: Site published:

Original article · فارسی

What a protest outside the company's headquarters today tells us about its future — and everyone else's

Ella Markianos

— 7 min read
Anthropic's wordmark on mirrored black glass
(Shutterstock)

Editor's note: Because of my conflicts of interest here, I did not write this article. But I did edit it. — Casey

And now, the company that has promised it may soon replace us has carried out the biggest stock market launch in history.

Anthropic raised $81.3 billion on Thursday when it began trading on Nasdaq. This is the largest amount of money ever raised in an initial public offering, or IPO. By the end of the day, Anthropic was worth $1.76 trillion. This made the AI company the world's largest public benefit corporation and one of the world's largest publicly traded companies of any kind.

At first glance, this should look like a huge victory for the AI industry. One of the most popular companies in the AI boom has successfully gone public. Its success shows that investors are eager for very advanced AI, and it sends hundreds of billions of dollars back into the technology industry. People still rushed to invest in Anthropic, even though there are concerns about an AI bubble. These concerns include companies making deals that send money around between one another and the effects of Nvidia CEO Jensen Huang's dealings on international politics.

But Anthropic's goal is to build AI systems that are smarter than people. The company's founders have repeatedly warned that such systems could kill everyone. So how should we feel about this moment? Is it bittersweet? Does it make us feel confused or angry?

It depends on whom you ask.

In a speech at Nasdaq, Anthropic CEO Dario Amodei repeated a message he has given many times. He spoke about how AI could lead to scientific discoveries, while also warning that people could misuse it. “My company’s co-founders and I started Anthropic because we believe that the most powerful technology of our time should be built for the benefit of humankind,” he said.

Not everyone believes him. While the founders were celebrating on Wall Street with what we assume was champagne made without alcohol and sourced responsibly, more than 100 protesters gathered outside Anthropic's San Francisco headquarters. They held signs reading “Stop the AI race,” “Earth is in danger,” and “Please stop.” One of them was Jacob Coxon. A post he made online about quitting Anthropic had gone viral and brought new attention to the AI safety movement. He also spoke at the protest. Four people were arrested because they were accused of blocking the entrance to the company's office.

Elsewhere, people with very different views about AI signed a document called “The MisAnthropic Statement”. They included Gary Marcus, who has long been doubtful about AI, and Guillaume Verdon, also known as Beff Jezos, who strongly supports making AI progress faster.

The statement says:

“Anthropic warns that AI systems that keep improving themselves could put too much power in a few hands and escape human control. Yet the company is racing to build these systems first. It uses its safety message to attract investors, workers, and public support, even as it has quietly weakened its safety promises and pushed for rules that benefit itself. Anthropic is not a safe AI company, and we must reject its reckless effort to gain power in the name of doing good.”

People who study the future have long discussed the dangers of something called “recursive self-improvement.” This is when an advanced AI builds newer versions of itself, which then build even more advanced versions. The process could become impossible to control. As it continues, people may become less and less able to understand what the AI is doing or stop it when something goes wrong.

A recent report from AI safety group METR found that AI had roughly doubled the speed of Anthropic's AI research and development. This is important because doubling the speed of research would mean Anthropic had passed its own risk limit for “Automated R&D,” as described in its “Responsible Scaling Policy.”

Under this policy, Anthropic should now be adding new safety checks, stronger security rules, and more open reporting to keep its promises. But the policy is not legally binding. Anthropic supported California's SB 53 law, which asks companies to publish safety policies that they must follow by law. However, after the law passed, Anthropic submitted a weaker plan under SB 53. It changed some of its stronger safety promises into voluntary commitments that it did not have to follow by law.

During his Nasdaq speech, Amodei promised that Anthropic would “pace the training and deployment of AI systems, taking as long as is necessary” to make sure they are safe. He added, “We will not bend to either business nor political pressure.”

But Anthropic will face pressure from its business needs to keep building the most powerful AI systems. It needs to keep its customers and satisfy its investors. Robert Armstrong of the Financial Times estimated that Anthropic would need to earn more than $600 billion a year to give investors the same 20 percent return on their investment that similar companies earn. For comparison, Apple brings in only $478 billion in revenue. So there is a lot of pressure! In a conversation with CNBC, Amodei said Anthropic wants to reach “investment grade,” meaning it wants a high credit rating. He also said the company plans to borrow more money.

Some people think that raising billions of dollars and taking on even more debt to buy more computing power is a strange way to “pace the frontier”, or slow the development of the most advanced AI. Amodei has explained that the company's approach is needed to create a “race to the top.” He believes that if Anthropic succeeds financially while keeping its safety promises, its competitors will copy those practices to catch up.

Jasmine Sun, writing for the New York Times, reported hearing a striking explanation of this idea from a former employee who did not give their name:

“One former Anthropic employee referred to a story by J.R.R. Tolkien to explain how the company thinks. Whoever builds superintelligence first will get Sauron's ring. That person or company will control the world's most dangerous cyberweapons, the cures and recipes for every disease, and an army of tireless digital geniuses that will do whatever they are told. The most important question, then, is whether the person holding the ring is a good person.”

Many readers will remember that the One Ring in The Lord of the Rings slowly corrupts the minds of everyone who wears it, turning them toward evil.

Like Frodo Baggins, who eventually came to hate the Ring, one of Anthropic's earliest investors now seems to regret investing in the company, and in OpenAI before it. In a statement to the New York Times, Facebook co-founder Dustin Moskovitz wrote: “Cari and I invested in Anthropic to promote safety, not racing.” Moskovitz invested with his wife, Cari Tuna. He continued: “A.I. is moving too fast. Anthropic and everyone else on the frontier should slow down and stop racing.”

A spokesperson for Moskovitz and Tuna said that they gave their Anthropic shares to a nonprofit in 2025, so they could not personally make money from them.

No one knows for sure what will happen next. Amodei clearly believes strongly in his plan. And Anthropic deserves credit for making a smart choice to focus on scaling laws while its competitors spent time on robotics, world models, and other less important projects. Looking back, that choice seems to have been very wise.

At the same time, the IPO may let ordinary investors make money if AI succeeds, but it has also drawn attention to a risky bet that all leading AI companies are making. They believe they can keep building smarter and smarter systems with very little government oversight, even though they themselves believe this technology could put everyone at risk.

Many parts of Anthropic's story are hard to believe. The company began as a small group of former OpenAI employees who seemed to care little about business. Somehow, they went on to lead the biggest IPO in history. People will study this story for many years.

But for the protesters, the hardest part of Anthropic's story to believe right now is how little governments have done to stop it. Perhaps AI really will bring a wonderful new age. But opinion polls suggest that the public does not believe this yet. For now, Anthropic has not convinced the world that the benefits are worth the risks.