Connect with us

BUSINESS

Anthropic Reserves the Right to Compete With Startups

Anthropic’s IPO filing says it is a model platform first, then reserves the right to compete with application companies in select categories it already occupies.

Published

on

Anthropic reserved the right to compete with application companies in its October 26, 2026 IPO filing. The Claude maker told the SEC its primary focus is the model platform, then added that first-party apps may compete in select categories from time to time.

European application-layer founders have worried for years that a model lab would ship the product they were building. The sentence they wanted as comfort is also the sentence that keeps that door open. Anthropic, PBC, founded in 2021, registration statement on Form S-1 names a Nasdaq listing under the ticker ANT and leaves the share count and price blank.

A Legal Hedge in the Competition Section

In the competition discussion, Anthropic splits the field into model labs, open-weight models, big-tech bundles, inertia inside companies, and its own apps. OpenAI and Google are the named frontier rivals. Then comes the line application founders will read twice.

As a platform company, our primary focus is on building the foundation upon which other companies and software developers can create products and services driven by Powerful AI. However, as we continue to develop our own first-party applications, we may from time to time compete with enterprise software and application companies in select categories.

Anthropic, PBC, Form S-1, October 26, 2026

That is not a product launch. It is a legal reservation, written so public-market buyers cannot later claim surprise if Claude’s own apps take work from firms that currently pay for the model. The filing still says the company is first and foremost a research lab selling access to Claude. The hedge sits beside that claim, not instead of it.

A May funding round valued Anthropic at $965 billion, a figure that already assumed the company could keep growing far past the model API. The same document now carries its $2 trillion IPO case into public view without a price range. The application-layer sentence is how the lab tells investors it can capture more of each customer’s workflow if the model fee alone is not enough.

Claude Code and Cowork Already Occupy Those Categories

The “select categories” are not hypothetical. The filing describes Claude as models, a platform, and applications, and it names the apps.

THREE CLAUDE SURFACES IN THE FILING

  • Claude Platform: The agentic development environment that gives customers the same primitives Anthropic uses to build its own apps, so other companies can ship products on the model.
  • Claude Code: The coding agent, offered as a research preview in February 2025, that plans, writes, tests, debugs, and commits across production codebases.
  • Claude Cowork: The knowledge-work agent, offered as a research preview in January 2026, that reads files, opens other software, and runs multi-step office tasks with the code hidden.

In September 2026, Anthropic began merging Cowork into Claude.ai so a chat question and a long agent task live in one product. The filing says Cowork’s early adoption outpaced Claude Code at the same stage after Code’s own preview. It also says Code “shifted the market’s understanding of what an AI system could do,” and it cites NASA engineers using it to plan and write navigation commands for the Mars Perseverance rover.

Those are not platform footnotes. They are end-user products in coding and office work, the two markets where independent application startups have spent the past two years wrapping Claude and rival models. The agent that sits in the terminal, or in the folder, is the product the customer touches. The model behind it is easier to swap.

The Same Startups Sit on the Revenue Line

The filing is careful on concentration. “No single customer represented more than 10% of our revenue for the nine months ended September 30, 2026,” it said. The top five customers together were about 25% of that nine-month total.

It also warned that a significant portion of the customer base consists of startups and technology-oriented enterprises that may be more sensitive to a downturn and may cut or delay spend. That is the awkward pairing. The companies most exposed if Anthropic ships a competing app are, by the lab’s own risk language, a material slice of who pays the bills.

Revenue is still mostly metered. About 80% of the first half of 2026 was consumption, priced in tokens; about 20% was subscriptions. Cloud marketplaces run by Amazon, Google, and Microsoft handled 48% of first-half sales, with Anthropic booking the full contract and paying the platform a fee. International customers, by billing address, were about 28% of the half.

None of that names a European application startup. It does not have to. Any firm whose product is “Claude, plus a workflow, plus a sales team” now has to underwrite a listed competitor that has reserved the right to sell the workflow too.

Why Heavy Claude Code Users Spend 20 Times More

The commercial reason for the hedge is in the usage gap. In the three months ended September 30, 2026, Claude Enterprise users who did not use Claude Code spent less than $75 a month on average. The top 5% of Claude Enterprise users who did use Code spent about 20 times that amount.

That gap is the business case for first-party apps. An account that treats Claude as a chatbot is a small check. An account that lets the agent plan, edit, and ship work is a large one. Cowork is the attempt to copy that pattern for people who do not write code. If office teams spend even a fraction of what heavy Code users spend, revenue per seat rises without a new logo on the customer list.

ENTERPRISE ACCOUNTS IN THE S-1

Measure Figure As of
Customers with $100,000+ trailing-twelve-month revenue More than 11,900 September 30, 2026
Same tier a year earlier More than 950 September 30, 2025
Customers with $1 million+ TTM revenue More than 1,900 September 30, 2026
Customers with $10 million+ TTM revenue About 250 September 30, 2026
Net revenue retention About 900% September 30, 2026
$100,000+ customers using platform and a first-party app Over 50% Q3 2026
Same dual-use share About 30% Q4 2025

Net revenue retention near 900% means customers active a year earlier spent about nine times as much in the latest quarter. The filing excludes Claude Pro and Max subscribers from that figure. Fortune 500 companies in the $100,000-plus group still spent, on average, less than 0.1% of their total expenses with Anthropic in the September quarter, which the company presents as unused room inside accounts it already has.

More Than Half Now Use Both Surfaces

For the three months ended September 30, 2026, over 50% of customers with $100,000 or more of trailing-twelve-month revenue used Claude Platform and at least one first-party application, up from about 30% in the three months ended December 31, 2025. That is a nine-month jump in how many large accounts live on both sides of the hedge.

THE HALF-YEAR BOOKS

  • H1 2026 sales: first-half revenue of $16.29 billion, up 1,300% from $1.16 billion in the first half of 2025.
  • Operating line: A $693 million operating loss for the half, after a $974 million operating profit in the June quarter, the first the filing reports.
  • Headline loss: An $80.75 billion net loss, of which $72.39 billion was a fair-value charge on convertible notes, derivatives, and SAFEs as the private valuation rose.
  • People: About 5,300 employees at September 30, 2026, with Claude writing over 90% of contributions to Anthropic’s own codebase that month.

The $80.75 billion figure is not cash leaving the building. The first-half 2025 net loss was $12.97 billion, also swollen by the same kind of notes charge. Second-quarter revenue was $11.56 billion. The operating story is a company whose compute bill has fallen as a share of sales and whose first-party apps are how it wants that sales line to keep compounding.

Once an agent owns the daily habit, the model underneath is a contract that can be repriced or swapped. Independent app companies have been living that shift already, as buyers shop harnesses and defaults rather than raw token lists. Anthropic writing its own harness is how it keeps the habit on its own books.

Amazon, Google, and Microsoft Play Every Role

The application hedge is only one conflict in the filing. Amazon, Google, and Microsoft are investors, customers, distribution partners, compute suppliers, and competitors at once. The S-1 says those overlapping roles could hurt Anthropic’s access to compute, its cost, or the priority of its workloads. It also lists Broadcom among its compute suppliers.

That is the same pattern as the startup line, pointed at much larger firms. The companies that sell Claude also sell rival models and, in Microsoft’s case, coding tools of their own. Anthropic still needs them to reach customers. It still needs the application startups that wrap Claude, until it does not.

The filing’s own words on the platform are the giveaway. Claude Platform, it said, gives developers “direct access to the primitives to build, deploy, and scale their own applications,” and “these primitives are the same ones we use to build our own applications.” Partners get the toolkit. Anthropic keeps a copy.

The Warning European Founders Already Knew

The European worry was never that Anthropic would abandon the model business. It was that a lab with a $965 billion private mark, compete with enterprise software companies language in an S-1, and two first-party agents already in market would not stay a quiet supplier.

Nothing in the document schedules a raid on a named startup. Nothing prices the IPO. What it does is put the lab’s own apps on the same page as the promise that other people should build on Claude. Code is in the IDE and the terminal. Cowork is merging into the chat product. More than half of the large accounts already use both the platform and an Anthropic app.

Application companies that remain customers will keep paying for tokens, and the biggest Code seats will pay a lot. Application companies that sit in those select categories now have the reservation in writing, in a prospectus aimed at public investors rather than at a conference audience in Europe.

Disclaimer: This article is news reporting and analysis of Anthropic’s Form S-1 and related public statements. It is informational only and is not a prospectus, an offer to sell securities, or a solicitation of an offer to buy them. It does not constitute investment, legal, or tax advice, and it should not be used as the basis for any buy, sell, or hold decision in Anthropic or any other company. Readers should consult a licensed financial adviser, broker, or securities lawyer before acting on IPO or secondary-market information. Figures and status details reflect the cited filings and statements as of the dates given in the article and may change as amendments are filed or the offering is priced.

Harry is the editor of IAQABA, an independent publication he owns and runs. A decade in journalism, beginning as a reporter and now as the editor of his own titles, has left him with a clear test for what deserves a story: it has to change what a reader knows or decides, and it has to rest on something he can point to. That rules out recycled press releases, forecasts with no data behind them and rumours that no document supports. It leaves room for a great deal, and the site covers news, business, science and technology alongside sports, entertainment and lifestyle, with travel, auto and gaming given the same standard rather than lighter treatment. Sources are primary wherever possible: the regulator's filing, the company's own statement, the transcript, the dataset, or the product on Harry's desk. Figures are checked before they are published and rechecked if a reader questions them. Mistakes are corrected under a published policy. Readers across the world can reach him directly at support@iaqaba.com.

Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending