Anthropic's IPO: How do you price the future of AI?
Talk about how much Anthropic could be worth on the stock market has brought attention to one of the biggest risks in finance: trying to put a price on the future of artificial intelligence.
A few days before its initial public offering (IPO), the company behind the AI assistant Claude said it was aiming for a market value of up to $1.4 trillion.
This would make the young company one of the biggest companies ever to enter the stock market. However, the figure is lower than the $2 trillion that some estimates suggested just a few weeks earlier.
Since then, worries about AI development and the ability of AI companies to make money have spread through the market. OpenAI, a rival that plans to go public in 2027, is valued at $1.025 trillion on the private market, according to US media reports.
Michelle Lowry, a finance professor at Drexel University, said there are two main ways to work out how much a company is worth. The first is to estimate how much money the company will bring in over the coming years.
Money expected in the future is worth less than money available today, especially when there is uncertainty about whether it will arrive. So these future amounts are reduced to calculate what the company is worth now.
The second method is to compare the company with similar businesses that are already listed on the stock market. For example, if those businesses are worth five times their yearly revenue, the same ratio can be used to estimate the value of the company being studied.
Both methods are well known and widely used. But Lowry told AFP that both are quite difficult to use in Anthropic's case.
Like most AI companies, Anthropic is spending money faster than it earns it and has not yet found a way to make a profit. This makes it difficult to predict how much money it will earn in the future.
There are also very few listed companies that can be used for comparison. The closest example is SpaceX, which made a record-breaking stock market debut in June with a value of $1.77 trillion.
But the comparison is not perfect. The company led by Elon Musk combines a social network, space operations and an AI research lab. SpaceX also benefits from Musk's popularity with individual investors.
'Back and forth'
So how did Anthropic arrive at the $1.4 trillion figure? Minmo Gahng, an assistant professor of finance at Cornell University, said companies preparing to go public and large professional investors take part in a process of discussion and negotiation.
Investment banks help bring the two sides together, he said. The process can take a lot of time.
A company's market value is calculated by multiplying its share price by the total number of shares available. A company preparing to go public wants its shares to have the highest possible price because this affects how much money it can raise.
However, the company must also leave enough room to attract buyers. If demand is weak, the share price could fall.
Investors, meanwhile, want to be confident that they can sell their shares later and make a profit. They also hope the share price will jump sharply on the first day of trading. This is known as a "pop."
Speaking with JPMorgan Chase chief executive Jamie Dimon on Friday, Anthropic co-founder and CEO Dario Amodei mentioned one of the company's guiding ideas when discussing the IPO: "underpromise and overdeliver."
He also said the main reason for going public was that the company needs more money. Anthropic plans to raise a total of $90 billion through the offering. It will use this money to develop advanced AI.
'No guarantee'
Jay Ritter, an expert on IPOs at the University of Florida, said investors are willing to accept such high company values because a small number of businesses, including Amazon, Alphabet and Nvidia, have shown that companies can earn $100 billion or more in yearly profits without competitors taking those profits away.
But Ritter warned that there is no guarantee Anthropic will achieve the same success.
"It could be the case that 10 years from now, Anthropic will have been left in the dust where nobody is using frontier large language models because open-source models are good enough and cheap," he said.
Anthropic's growth has already shown signs of slowing down. Its IPO prospectus, a document explaining the company and its plans to investors, spends 90 pages describing risks to the company's future and even to humanity. These risks depend on how AI develops.
