AI vendor dependence is no longer a theoretical question for small businesses that build their operations on Claude, ChatGPT, or Gemini. On October 26, 2026, Anthropic publicly filed its Form S-1 with the US regulator (the SEC), the official document that precedes an initial public offering. This filing, more than 90 pages of risk factors, discloses for the first time precise figures on the real financial health of the company behind Claude, its staggering financial commitments, and the political tensions weighing on it. For an SME that has built processes or agents around the Anthropic API, this information is not trivial: it sheds light on a vendor risk that is too often ignored.
In brief
- Anthropic filed its S-1 on October 26, 2026 to list on Nasdaq under the ticker "ANT," via Goldman Sachs and Morgan Stanley (source: Anthropic, CNBC).
- First-half 2026 revenue reached $16.29 billion, up from $1.16 billion a year earlier, a 1,304% increase.
- The first-half 2026 net loss stood at $80.7 billion, of which roughly $72 billion is a non-cash accounting charge tied to convertible financial instruments, not money spent day to day.
- Future compute commitments exceed $417 billion, spread across Amazon, Google, Microsoft, and SpaceX.
- The filing reveals an open conflict with the Trump administration and the Pentagon, which labeled Anthropic a "supply chain risk" to national security.
- For an SME, the lesson isn't alarmism: it's the need to check the stability and continuity of service of any AI vendor before relying on it too heavily.
What Anthropic's filing actually reveals
An S-1 is the document a US company files with the SEC before going public. It details its finances, risks, and governance, and becomes public at the time of filing. Anthropic had already submitted a confidential draft in June 2026 (partially revealed by Reuters on September 28), but the October 26 filing is the public, official version, with figures updated through mid-2026.
Revenue growth is real and fast: according to the S-1, implied gross margin rose from about 12% to 85% between the first half of 2025 and the first half of 2026, a sign of a business model improving with scale. The filing also notes that more than 11,900 customers spend at least $100,000 a year with Anthropic, including roughly 1,900 above $1 million. Notably for the question of dependence: no single customer accounts for more than 10% of revenue, and the top five customers combined represent about a quarter of revenue. Anthropic, it turns out, depends far less on any one customer than its customers depend on it.
Extraordinary financial commitments with cloud giants
The most striking point for understanding the structural fragility of the generative AI sector is the scale of commitments Anthropic has made to its compute providers. These numbers give a sense of the investment level needed simply to stay in the race.
These figures (source: Anthropic S-1, cited by CNBC, October 26, 2026) come on top of future lease obligations of $321.4 billion, up from $50.2 billion at the end of 2025. Amazon has invested $13 billion in Anthropic and can inject up to $20 billion more, with its stake capped at 12% of the company. This level of financial interdependence between AI labs and cloud providers explains why the ecosystem's stability remains fragile, even for the fastest-growing player.
What this means in practice
A vendor committing hundreds of billions of dollars over ten years is betting on continuous demand growth. If that growth slows, pressure on margins and prices could trickle down to customers, including SMEs.
A year of tension with Washington
Beyond the financial figures, the S-1 details a political dispute that has directly affected service to Anthropic's customers, potentially including those abroad.
February 2026
Presidential order
June 2026
Export control pause
Summer 2026
Conflicting court rulings
October 26, 2026
Public S-1 filing
For a European SME, this episode sends a clear signal: even a financially solid AI vendor remains exposed to political or regulatory decisions that can interrupt a service overnight, even if only for a few weeks. It's one more reason never to build a critical process on a single tool without a fallback plan.
What this means for an SME that depends on an AI vendor
This isn't a call for alarm: Anthropic's revenue growth (14x in one year) is a sign that demand for generative AI is real, and that "measured optimism" remains justified by the facts. But a business owner building tools or agents around any AI vendor should regularly check their degree of dependence.
Map the dependence
Read the continuity clauses
Keep your data portable
Test an alternative
Confidential (June/October) vs public filing: what changed
| Metric | Confidential draft (revealed 09/28/2026) | Public S-1 (10/26/2026) |
|---|---|---|
| Period covered | Full year 2025 | First half 2026 only |
| Revenue | $4.6B (full year 2025) | $16.29B (H1 2026 alone) |
| Net loss | $42B (full year 2025) | $80.7B (H1 2026 alone) |
| Implied gross margin | Not disclosed | ~85% (H1 2026), up from ~12% a year earlier |
| Cloud/compute commitments | $518B cited (aggregate figure) | $417B signed compute + ~$125-150B expected 2028-2031 |
| Political risk detail | Lightly covered | Dedicated section: Pentagon, Trump, FTC, litigation |
This table illustrates something simple: the closer an AI company gets to going public, the more precisely it must document its risks. That's good news for SME customers, who gain more reliable information to assess their vendor's stability, provided they take the time to read it.
FAQ
What is an S-1 form?
An S-1 is the official document a US company files with the SEC (the securities regulator) before an initial public offering. It details the company's financials, risks, and governance, and becomes public as soon as it's filed, unlike a confidential draft submitted beforehand.
Does Anthropic's $80.7 billion loss mean the company is in trouble?
Not directly. About $72 billion of the $80.7 billion loss for the first half of 2026 is a non-cash accounting charge tied to the valuation of convertible financial instruments, not money actually spent. The real operating loss is much smaller, and revenue is growing very fast (+1,304% year over year).
Should an SME worry about using the Anthropic API or Claude?
Not particularly more than with any other major AI vendor. The broader takeaway is that every AI vendor, even a fast-growing one, remains exposed to financial, political, or regulatory risk. Good practice is to diversify dependencies and keep a continuity plan, whichever provider you choose.
When will Anthropic actually go public?
The number of shares and price range had not yet been set at the time of the S-1 filing on October 26, 2026. The registration is not yet effective, and shares cannot be sold before final SEC approval.
Going further
This news connects to our October 4, 2026 article on Anthropic's confidential filing, which detailed the first leaked figures. To structure your thinking on choosing an AI vendor, also check our guide on choosing your AI model in 2026 and our customer success stories on integrating AI in small businesses.
