SIAgentNews.com
← All news

Anthropic's record $81.3 billion listing raises the stakes in AI pricing

Anthropic's share sale is the largest ever, and analysts say the cash could fund a price war with cheaper Chinese AI labs. No price cut has been announced.

· First reported by South China Morning Post on MSN

$81.3 billionsize of Anthropic's IPO, the largest on record

The short version

  • Anthropic priced shares at $83, above the $70 to $78 range, for a market value of around $1.5 trillion.
  • The $81.3 billion offering beats SpaceX's $75 billion listing from five months ago.
  • A further $26 billion is being raised privately at the IPO price, with Google and Microsoft among the backers.
  • Revenue was $33.24 billion in nine months, but net losses widened to $104 billion.
  • Moonshot AI and DeepSeek face a tougher yardstick on profit and enterprise sales.

Why it matters

If you pay for AI tools, prices could shift as well-funded labs compete. No price change has been announced, so treat this as a risk to plan around, not a reason to act today. It also shows how investors will judge every AI lab: by profit and recurring business revenue, not just popularity.

A record sale, priced above range

Anthropic, the San Francisco company behind the Claude chatbot, sold shares to the public at $83 each. That was above the $70 to $78 range it had marketed to investors. At that price the company is worth around $1.5 trillion, and it begins trading on Nasdaq under the ticker ANT.

An IPO, or initial public offering, is the first time a company sells shares to the general market. The base offering was 980 million shares, made up of about 956.4 million new ones and 23.6 million sold by existing investors. As the South China Morning Post reported on MSN, that makes it the largest IPO in history, ahead of SpaceX's $75 billion listing five months ago.

On top of the public sale, Anthropic is raising another $26 billion through a private placement at the same price. Regulatory filings name Google and Microsoft among the backers.

Why analysts talk about a price war

The argument is simple. A company with a very large pile of fresh cash can afford to charge less than rivals for the same service, and keep doing so for longer. Analysts quoted by the paper say that is the option Anthropic now has.

The likely targets are Chinese labs such as DeepSeek, Moonshot AI and Z.ai. They have won global attention with cheaper open-weight models, meaning AI systems whose inner settings are published so customers can run and adapt them. Felix Wang of Hedgeye Risk Management said the price war in large language models is still going and that Anthropic's capital lets it compete more aggressively.

He also noted the Chinese models keep an appeal for businesses that want control over how the software is deployed, customised and used with their own data, even if the models trail the best US systems in capability.

This is the analysts' reading. Nothing in the report says Anthropic has cut or plans to cut a price.

A tougher yardstick for Chinese listings

Moonshot AI is eyeing a listing in Hong Kong, and DeepSeek one on Shanghai's Star Market. Poe Zhao, a Beijing-based tech analyst, says Anthropic's debut sets a benchmark for them on profitability and enterprise revenue. Once they file, he expects investors to ask how far gross profit covers research spending, and how much revenue comes from recurring business customers instead of low-price traffic.

That matters because many Chinese developers built their user base with free consumer apps and heavily discounted APIs, the connections that let other companies plug an AI model into their own products. Zhao said heavy usage shows up as a computing bill before it shows up as revenue. He argues these labs need to move faster toward usage caps, tiered pricing and paid enterprise contracts.

Ivan Su of Morningstar doubts the good mood around Anthropic will spill over to Chinese companies. He points to Anthropic's higher margins and its access to more advanced chips.

The numbers behind the excitement

Anthropic was founded in 2021 by former OpenAI researchers, led by CEO Dario Amodei. Its prospectus says it earned $33.24 billion in revenue in the first nine months of 2026, roughly 13 times the same period a year earlier. The report credits booming business demand for its coding assistant, Claude Code.

Losses grew too. Net losses reached $104 billion over the period, up from $25.1 billion a year before. The report says this was driven largely by accounting adjustments tied to convertible notes (loans that can turn into shares) and derivative liabilities, rather than by day-to-day operating costs alone.

Counterpoint Research's Yang Wang called the IPO the "clearest public market test yet of frontier model economics". He pointed to an annualised revenue run rate of around $81 billion, meaning current earnings projected over a full year, set against about $450 billion of computing commitments through 2030. The question, he said, is whether revenue, cash generation and efficiency in running models can keep pace as those bills come due.

Confidence, caution and what to watch

Dan Ives of Yorkville Ives & Co called the offering a near-term win for the tech sector. He expects between $3 trillion and $4 trillion to go into AI-related spending on infrastructure over the next few years. Morningstar's Malik Ahmed Khan put fair value at $92 per share, above the $83 offer price, citing Anthropic's strength in enterprise coding and cybersecurity.

Others were more careful. Khan warned about the enormous cost of building and running the models. Hedgeye's Wang said investors may be projecting early growth too far ahead while underestimating how models become interchangeable over time, and the effect of AI safety rules.

For a typical user, the practical message is modest. Prices may move, but nothing has changed in your plan or bill today. Keep your workflow portable and avoid locking into a long single-vendor deal at current prices.

What to watch

  • Whether Anthropic actually changes its prices after the listing, since none has been announced.
  • How Moonshot AI and DeepSeek present gross profit and recurring enterprise revenue if and when they file.
  • Whether revenue growth can keep up with about $450 billion of computing commitments through 2030.

Try it

Pick five or six tasks you really do with AI, run them on two different models, and compare accepted results against what you spent. Ask vendors for month-to-month terms.

Sources and links

First reported by South China Morning Post on MSN. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

More AI news