No moats at AI providers: What the Anthropic IPO means for your SME
In short
Even the largest AI IPO in history may lack a moat, according to Steve Eisman. For your business, the model layer is not a strategic advantage. Your real edge is in data, processes, and orchestration. Models are becoming commodities, while your workflows make the difference. AI outsourcing keeps you vendor-independent.
The Anthropic IPO is being discussed as potentially the largest in history. Steve Eisman, known from The Big Short, says there are no moats. For your business, that means the model layer is not your home turf. Your advantage comes from how you orchestrate processes and data. This is where AI outsourcing for SMEs comes in.
The Anthropic IPO: Numbers you should understand
Anthropic filed for its IPO on October 26. The valuation is estimated at around USD 1.5 trillion. Revenue in the first half of 2026 was USD 16.3 billion. Growth is slowing: Q3 revenue grew 45 percent, down from 144 percent in Q2.
USD 807 bn
Anthropic's compute commitments per IPO filings
There are no moats. You commit to spending 800 billion dollars while your whole company is based on price.
Why the model layer is becoming a commodity
If a company with these numbers has no durable competitive advantage, then choosing a specific language model is even less strategic for your SME. Prices for AI models are falling, open-source alternatives are improving, and differences between frontier models are shrinking. What matters is not which model you rent, but how you integrate it into your workflows.
Your real moat: Data, processes, orchestration
A lasting advantage for an SME does not come from the model, but from your own master data, documented processes, and the ability to orchestrate different AI components. This layer is vendor-independent. You can outsource it without locking yourself into a single provider.
- Strategic focus
- Building an internal AI team vs. using an external AI division
- Vendor binding
- Multi-year contract with one model provider vs. model-agnostic orchestration
- Migration costs
- Known cases: CHF 30,000 to 120,000 for SMEs, peaks up to USD 315,000 vs. low with clear exit clauses
- Speed
- Internal hiring takes months vs. AI outsourcing starts in weeks
- Risk
- Vendor lock-in and price increases vs. switching providers without rebuilding
How to avoid vendor lock-in as an SME
- Review exit clauses in every AI contract: data return, model switching, notice period.
- Use model-agnostic orchestration instead of proprietary single APIs.
- Keep your prompts, data pipelines, and process documentation in an open format.
- Commission AI outsourcing with a requirements document that defines vendor independence and exit costs.
Practical tip
Ask every AI service provider to confirm in writing how a switch works and what it costs. If they avoid the topic, they often have little interest in your independence.
AI outsourcing as a vendor-independent entry point
You don't need to build your own AI team to benefit from AI. A specialized partner can handle orchestration, swap models as needed, and protect you from dependencies. The Swiss AI podcast discusses exactly this: How DeepSeek, Apertus, and AI agents change your AI strategy. This keeps you agile without chaining yourself to a single vendor.
The Anthropic IPO is a reminder that AI models are not durable moats. Your moat is in your data and processes. AI outsourcing keeps you independent.
Frequently asked questions
- Why does Steve Eisman say AI providers have no moats?
- He argues that AI providers compete mainly on price and commit to huge compute spending without a durable competitive advantage.
- What does the Anthropic IPO mean for my SME?
- It shows that choosing a single AI model is not a strategic advantage. Your data, processes, and orchestration matter more.
- How can I avoid vendor lock-in with AI?
- Use clear exit clauses, model-agnostic orchestration, and open data formats. A requirements document defines switching costs and data return.
- Can I use AI without building my own team?
- Yes. AI outsourcing handles orchestration and provider switching for you, keeping you agile and independent.
- Are open-source models a good alternative?
- Yes, they improve and reduce dependence on single vendors. The key is orchestration, not the specific model.
Sources
Chris Jon Graf · AI Strategist & CEO
For over 15 years, Chris Jon Graf has built productive Swiss software platforms and advised executive teams on the strategic deployment of artificial intelligence. Would you like to evaluate this topic for your company?