Anthropic IPO: AI titan’s massive public float sparks questions for investors as cheap competitors loom

Cheap Chinese AI models are among the biggest risks to Anthropic’s growth prospects, as millions of Australian investors are invited to bid for shares in the US tech giant’s initial public offer over the next week.

Australian investors have until November 17 to apply for shares through popular online brokers including CommSec, Stake, and Bell Potter in a deal that offers the chance to own the world’s hottest AI company.

Minotaur Capital co-founder Thomas Rice said Anthropic’s prospectus is the first time investors had gained a look inside an AI lab’s audited financials and the results are impressive.

“The numbers in the S1 (prospectus) were much better than I and a lot of people expected,” he said. “They hadn’t burned too much capital and they’re at the tipping point of profitability, so it looks interesting.”

Anthropic’s sales surged 13-fold to $US33.2 billion from $US2.5 billion over the nine months to September 30, 2026, compared with the prior corresponding period.

The jaw-dropping growth is fuelled by demand from corporate customers for Anthropic’s AI tool Claude, which helps businesses automate workflows, code software, cut costs, and improve sales.

Anthropic counted more than 11,900 customers paying subscription fees of $US100,000 per year or more over the trailing twelve months to September 30. That’s a more than twelve-fold increase on its total number of customers - just over 950 - a year earlier. It also revealed 1,900 customers paying $US1 million per year, or more, and about 250 paying at least $10 million annually.

For potential investors, the key question is whether Anthropic can keep growing sales and corporate subscribers faster than its costs rise.

Its largest cost is compute and infrastructure for the chips and data centres that run its models. Those costs reached $US18.8 billion over the nine months to September 30, to make up more than half its total operating costs of $US34.5 billion.

Anthropic’s operating loss of $US2.4 billion for the nine-month period is the difference between the operating costs and sales of $US33.2 billion.

Rice says he’s more than lukewarm on the IPO and Anthropic’s potential to turn highly profitable.

But he warns Anthropic’s fight to maintain its dominant market share faces competition.

“The question mark I still have in my mind is that when you look at the workflows that can be automated with AI in theory you don’t really need a frontier model versus a second-tier one, or Chinese open weight models, where the products aren’t dramatically different,” he says.

In theory you don’t need a Ferrari to deliver Uber Eats.

Just as many companies won’t need best-in-class AI models to deliver their workflow objectives.

Minotaur Capital has previously warned that OpenAI, Google, xAI’s Grok, Meta’s Muse and fast-improving Chinese open models are all chasing Anthropic.

Eventually, corporate customers may decide that a model a few steps behind the frontier is good enough, especially if it is cheaper or free.

The IPO’s structure has also been criticised for allowing Anthropic’s seven founders to retain a 50.1 per cent voting share in the company.

This means retail investors will have no effective say over the company’s direction, although many will only be interested in whether the share price rises or falls.

Bull case

The argument in favour of buying Anthropic shares is that it’s a business growing lightning fast in a tech field that will expand strongly for at least the next decade.

The ‘lead husky’ company in its field tends to be the strongest bet in investing, and every business faces competition in one form or another.

“An obvious point is that AI demand is real and I don’t think it’s going anywhere,” says Rice. “It’s also a really important part of the US and global economy.”

The stock picker, regarded as one of Australia’s top AI investing experts, also points to the fact Anthropic’s updated prospectus shows its models are moving closer to improving themselves.

This scenario may unlock another wave of huge demand for its services over the next five years.

“I noticed they (Anthropic) did update the S1 (prospectus) to add autonomy to capabilities which is basically where they’re getting a dramatic acceleration in their AI Research & Development program,” says Rice. “That’s the idea that these models then have a self-reinforcing loop, so that disclosure is essentially saying they are getting there.”

Rice says investors need to ask themselves if corporate AI adoption is near the peak, or still at the beginning?

If it’s still early days then demand over next five years will accelerate, with market leader Anthropic positioned to benefit.

AI’s advances over the next five to 10 years may prove so dramatic that Anthropic’s $US1.4 trillion valuation in 2026 looks very cheap by 2030.

The valuation represents about 17 times Anthropic’s annualised sales rate of $US81 billion as of October 31. And projections already point to annualised sales reaching $US100 billion by year-end.

Anthropic’s supporters also argue that its computing costs could decline under new or renegotiated contracts over time, boosting profitability, if it at least maintains market share.

Minotaur’s investment team is still evaluating its final decision in whether to buy into the float, but sees the risks as finely balanced.

Why Nvidia fallout may aid Anthropic

Reports that Nvidia founder Jensen Huang had dated an alleged Chinese spy could also work in Anthropic’s favour if the controversy prompts a renewed push by the US to restrict China’s access to AI technology.

Rice says any widening split between the US and China will strengthen Anthropic’s competitive position and force most Western companies away from cheap Chinese competition.

“Anthropic has a strategic interest in not selling chips to China because the open-weight Chinese models are the biggest threat to their business,” he says. “So, if you think the Nvidia scandal leads to more scrutiny about the Chinese models that could be a positive.

“The other point is that aside from Huang’s problems, Nvidia is arguably the most important company in that ecosystem.”

On that basis, Rice said he expects Huang to survive the drama and argues Nvidia may be good value at just 11 times forecast profits over the next 12 months.

“The entire company revolves around Huang and it would be a big blow to Nvidia and the entire AI ecosystem for him to step down,” he warns. “So, I’d be shocked if he knowingly broke export laws and I think he probably stays. “

Nvidia shares rebounded 2 per cent in early trade on Wall Street on Wednesday, after two days of heavy losses linked to the crisis around AI talisman Huang.