Subscribe
Should you invest in Anthropic’s IPO?
The company behind Claude is preparing one of the biggest share market listings in history. But its extraordinary growth comes with unusual risks for everyday investors.
- Anthropic’s proposed IPO could raise as much as US$100 billion, although investors are questioning its valuation.
- Its founders will retain majority voting control, while the company must balance shareholder returns against broader obligations.
- Explosive revenue growth has attracted investors, but mounting costs, competition and an uncertain share price complicate the outlook.
Warnings in company prospectuses rarely come as starkly as those in Anthropic’s November filing with the US Securities and Exchange Commission.
The company behind the Claude AI chatbot warned that, without appropriate safeguards, its products could “pose catastrophic or existential risks to humanity, particularly if such systems were misused, replicated, weaponised, or allowed to operate with a high degree of autonomy.”
While Claude often steals the limelight, Anthropic also has Mythos 5.1, an agent so advanced and dangerous that, for now, it is available only to the US government and trusted partners.
Amid all that fanfare, Anthropic is trying to raise a record US$100 billion from members of the public through a so-called Initial Public Offering (IPO), which could value the entire company at more than US$1 trillion.
However, after a dive in US stock prices following last week’s midterm elections, investors have told the Wall Street Journal they now expect a lower price, enough to value the company at roughly $500 billion.
Still, it’s worth pointing out that roughly a year ago, Anthropic was valued at $US183 billion.
Australians looking to jump in on the hype can apply for shares through brokers including CommSec. They are due to trade on the US Nasdaq exchange from late November.
But they are unlike the shares Australians are used to. Most Australian shares carry voting rights, giving shareholders the right to overrule the board by a majority vote.
By comparison, the Anthropic IPO limits ordinary shareholders to 49.9% of the votes. The seven founders get a special type of share that gives them the other 50.1%. This will give them the ability to “determine the outcome of all matters submitted to a vote of our stockholders”.
It means they won’t need to prioritise shareholders’ financial interests. The IPO says so. The structure is designed to balance “the pecuniary interests of our stockholders” against other things including “the best interests of those materially affected by our conduct” and “the long-term benefit of humanity.”
For ordinary investors, that means having little say over major company decisions, even if those decisions affect the value of their shares. And it’s something worth considering.
Funding needs
In any case, Anthropic is hungry for money.
In September last year it raised about US$13 billion, then in February this year another US$30 billion, then in May another US$65 billion. And it will need plenty more.
Its latest prospectus reveals almost US$900 billion in future computing and infrastructure commitments, including nearly US$70 billion due within a year.
Revenue is growing at an extraordinary pace, too. Anthropic brought in almost US$17 billion in the September quarter, more than 12 times what it earned a year earlier.
Still, the cost of running its AI systems is eating into a growing share of that revenue. In the June quarter, Anthropic spent 48 cents on computing and infrastructure for every dollar it earned.
By September, that had climbed to 58 cents, according to its updated IPO prospectus. The increase was partly driven by more expensive short-term computing arrangements, so it doesn’t necessarily mean Anthropic’s technology has become less efficient.
Even so, it has left less revenue available to cover other expenses and eventually turn a profit.
The trouble is that rapid growth doesn’t necessarily make a company a good investment at any price.
At a valuation of hundreds of billions of dollars, investors are betting that Anthropic can keep growing and eventually turn that growth into substantial profits.
Still, the potential upsides are enormous. Anthropic says many businesses are just getting started with Claude. And it paints the broader upsides for humanity as enormous.
 “The possibilities include compressing decades of scientific progress into years; accelerating drug discovery, disease detection, and the delivery of personalised therapy,” according to the prospectus.
It adds that its technology could also provide “individually tailored education and expert financial and legal advice” while “advancing power grids and battery design for greater efficiency and reliability.”
An uncertain future
One downside is that its products might develop so quickly that we can’t understand or stop them.
As Anthropic puts it, “our increasing use of AI systems to accelerate our own research and development may cause the capabilities of our models to advance more rapidly than our ability to develop, validate, and implement corresponding safety, alignment, and security measures”.Â
The company said many of the legal risks it faces are impossible to insure against.
One major financial risk is that, as revolutionary as its products are, it will lose market share to cheaper, more nimble new entrants. Netscape was the first commercial web browser. AltaVista was one of the first search engines. Neither survived.
Legendary investor Warren Buffett has often said innovators need “economic moats,” or barriers that make it hard for new entrants to steal their customers.Â
Anthropic has competitors advancing about as fast as it is (even though they are not yet as advanced), meaning it might not have a moat. Unless that moat is the ability to use its own AI to accelerate the pace at which it outruns competitors.
Anthropic is the first pure AI company to attempt to list on a stock exchange. OpenAI plans to follow next year. If demand for its shares is high and the price is strong, the entire US tech industry will be relieved.
If it goes badly, as did the IPO for the Australian data centre operator Firmus early last month, when the share market listing was scrapped, it could raise uncomfortable questions about the extraordinary valuations being placed on AI companies.