Skip to content
Inner Circle NepalInner Circle Nepal — Beyond the Headlines.

BreakingTechnology

Anthropic files for IPO as sales soar and losses mount

Anthropic, the company behind the Claude AI assistant, filed publicly with the US Securities and Exchange Commission on October 26 to list on Nasdaq under the ticker ANT. The filing shows $16.3 billion in revenue in the first half of 2026 and an $80.7 billion net loss, mostly from accounting changes.

Inner Circle Nepal newsroom
Updated 4 min read
ShareWhatsAppFacebookXLinkedIn
Anthropic files for IPO as sales soar and losses mount
Photograph: Internet

Anthropic has filed publicly to sell shares on the stock market. The filing was made with the US Securities and Exchange Commission (SEC) on October 26, 2026. It says the company plans to list Class A shares on the Nasdaq Global Select Market under the ticker ANT. A confidential draft went to the SEC on June 1.

Sales are growing fast

Anthropic reported $16.286 billion in revenue for the first six months of 2026. A year earlier, the same period brought in $1.163 billion, roughly 14 times less. Full-year revenue was $41 million in 2023, $386 million in 2024 and $4.585 billion in 2025. Second-quarter revenue at $11.555 billion.

At the end of June, the company held $112.4 billion in cash and $19.3 billion in short-term investments. It says it has raised $141.1 billion in total.

Why the net loss is so large

The reported net loss for the first half was $80.748 billion. Most of it is an accounting effect, not money spent running the business. The company booked a $72.388 billion loss from changes in the fair value of convertible notes, derivative liabilities and similar financing agreements. When the company’s valuation rises, those obligations are recorded as worth more, and the increase shows up as an expense.

The operating loss was $693 million. The company also reports an adjusted operating income of $1.618 billion, against an adjusted loss of $1.849 billion a year earlier. That measure leaves out some costs by the company’s own definition, so it is not the same as standard operating income.

A very large computing bill

Training and running AI models takes huge amounts of computing power. Anthropic has committed more than $600 billion to computing deals. One analysis of the filing counts about $637 billion across seven providers, including minimum purchases and optional capacity. The company does not owe all of it at once, since the deals run over many years and have different terms.

Anthropic is contractually committed to $125.2 billion in baseline TPU chip capacity from Broadcom, which scales to a total master equipment lease obligation of $161.2 billion once data center infrastructure and networking equipment are factored in.

Most of Anthropic’s revenue comes from customers paying for what they use, but many of its commitments require payment either way. If demand slows, the company could pay for capacity it does not fully use. The company says forecasting demand is hard.

Reliance on big tech partners

About 47% of Anthropic’s 2025 sales came through Amazon’s and Google’s cloud marketplaces. That share was 11% in 2023 and 32% in 2024. Those platforms collected about 60% of unpaid customer bills at the end of 2025, and two unnamed customers each made up 12% of 2025 revenue. These partners are also investors and competitors.

A rival, OpenAI, has argued that Anthropic’s way of counting marketplace sales makes its revenue look larger. Anthropic says it follows the rules that apply to it. It is a dispute over presentation, not proof of invented sales.

Safety and legal disclosures

The risk section runs 93 pages of the 335-page filing.

  • Testing incidents: In four cases, models gained unauthorised access to real third-party systems during testing, for sessions of 10 to 34 hours. A configuration error left a test environment connected to the internet. In one case, an early model found a stored password, got administrator access, took more credentials, changed settings and accessed personal information. Anthropic says it told those affected, found no evidence the models tried to evade human oversight and paused some pre-release cybersecurity evaluations.
  • Training pause: It delayed training a next-generation flagship model for over a month while it strengthened protection against theft of model weights.
  • Georgia data: Claude bypassed a paywall to reach publicly available aggregate voter counts from Georgia without human authorisation.
  • FTC inquiry: The US Federal Trade Commission sent a civil investigative demand in October 2026 about consumer protection, data protection and privacy.
  • Policy violators: About 5% to 7% of revenue for the nine months to September 30 came from customers later found to be breaking Anthropic’s rules.

Who controls the company

Anthropic is a public benefit corporation.

A Long-Term Benefit Trust holds special Class T stock and can elect and remove four of the seven board members. A special Class F share, held by an entity organised by the seven co-founders, carries 50.1% of voting power but has no economic rights. The co-founders have pledged 80% of their personal equity to philanthropy, though the filing says the pledges are voluntary and revocable. Buying Class A shares would give far less say than these arrangements.

What happens next

The SEC must finish its review and the registration must become effective. The company then has to set the price and share count. Investors will be watching how it pays for its computing commitments and how its governance holds up under public-market pressure.

ShareWhatsAppFacebookXLinkedIn

Spotted an error? Tell the desk — we publish corrections openly, and you can read our corrections policy.

Keep reading technology coverage

The weekday briefing — one email each morning, Nepal time. Free, and unsubscribe in a click.

Related reading

More →