Anthropic's IPO prospectus says Anthropic "may increasingly compete" with some of its own customers, and that the risk is "particularly acute with software-as-a-service and other software companies." It now has about $91.6 billion of fresh money from the offering, so for a SaaS company built on Claude, the supplier under the product is also a very well-funded possible competitor.
If you run a Shopify brand, you're a customer of those SaaS companies, so this shows up at your renewals too. None of this is investment advice.
How Much Did Anthropic Raise, and Where Is It Going?
Anthropic priced at $83.00 a share on November 18, 2026 (pricing announcement), started trading on Nasdaq as ANT on November 19, and closed on November 20. The closing announcement says the underwriters took all 147 million of their extra shares, and that AMD, Broadcom, Google, Micron, Microsoft, NVIDIA, and TSMC Partners bought another $26.0 billion of a separate class of stock at the same price.
Example: the $119.5 billion, from the share counts (gross, before underwriting discounts and expenses)
| Who sold the shares | Shares | Price | Gross proceeds |
|---|---|---|---|
| Anthropic (public offering) | 1,103,351,354 | $83.00 | $91.58 billion |
| Existing stockholders (public offering) | 23,648,646 | $83.00 | $1.96 billion |
| Public offering total | 1,127,000,000 | $93.54 billion | |
| Private placement (Class N stock) | 313,253,007 | $83.00 | $26.00 billion |
| All shares sold | 1,440,253,007 | $119.54 billion |
The $1.96 billion went to the stockholders who sold, not to Anthropic. The final prospectus says the money goes to "general corporate purposes," and names compute, infrastructure, and hiring among them.
For scale, revenue was $33.2 billion for the nine months ended September 30, 2026, up from $2.5 billion a year earlier. Customers paying $100,000 or more over the trailing twelve months went from more than 950 to more than 11,900.
Why Does Anthropic's Prospectus Single Out SaaS Companies?
Because a SaaS product with AI features may run on what Anthropic sells and do what Anthropic's own products do. Here's the passage, from a risk factor about deriving "substantially all of our revenue from a limited number of products and applications":
"In addition, we currently compete, and may increasingly compete, with some of our own customers. As we expand our product portfolio or introduce new capabilities, our offerings may overlap with or directly compete with the models, platforms, and applications that our customers provide. This risk is particularly acute with software-as-a-service and other software companies, whose offerings may incorporate AI functionality similar to, or dependent on, the capabilities we provide."
The next sentence is the blunt one. Those customers "may view our product expansion as a competitive threat" and may buy less, "seek alternative providers," or build their own. Anthropic is telling investors it could lose customers because of where it's going.
You can see the overlap in how the prospectus describes revenue. What it calls pay-for-intelligence revenue "is primarily generated from customers building applications incorporating Claude or utilizing our first-party applications." Its own products include Claude Code and Claude Cowork, which it calls "our knowledge work agent." So the same company sells the model a SaaS product runs on and its own agent for knowledge work.
The dependence runs both ways, though. No single customer was more than 10% of Anthropic's revenue for those nine months, but the top five were about 25%, roughly $8.3 billion. And in a risk factor about compute, Anthropic says "Customers, partners, and software developers may shift to competitors whose models offer superior capabilities". It treats its own customers switching models as a real risk. The SaaS companies built on it should treat switching as a real option.
What Should You Ask Your Software Vendors Now?
Ask whether they can switch models, because a vendor that can't has tied its product, and your workflow, to one supplier's roadmap. Before the next renewal on any app with AI features, I'd ask:
- Which model does this feature run on?
- If that model's maker changes its price, its terms, or its own product plans, can you move to another model, and how long would that take?
- Does my data go to the model's maker, and is any of it used for training?
- When the AI changes a record in my books or my store, who approves it, and where's the log?
The second question is the one this prospectus makes urgent. The supplier has now said, in an SEC filing, that it may compete with the companies buying from it.
Why Aren't We Worried at Eleven Elves?
Two facts, and no forecast.
I already build my own elves. An elf is an AI agent that does one back-office job. On November 19, 2026, I built one that reconciles Daily Grace's sales tax payments in QuickBooks: the 45 monthly reports used to take me about 2 hours every month, and the elf did it in about 2 minutes. Somebody still has to approve what an elf changes and own the books.
And Eleven Elves, which isn't open yet, is built with its own native AI and built to let customers plug in the AI of their choice. That's my answer to question 2: the customer can pick the AI.
The sales tax elf gets its own post soon. If you want practical strategy for running a Shopify brand in the meantime, sign up for our strategy newsletter.
JJ
This article is general information, not tax or accounting advice. Check with your accountant about how it applies to your business.