Gurman: internal development at Apple “runs on Anthropic at this point” and Apple wanted to rebuild Siri around Claude but Anthropic “wanted a ton of money”
Bloomberg's @markgurman says that even though Apple partnered with Google Gemini for Siri, they actually run their business on Anthropic. “Apple runs on Anthropic at this point. Anthropic is powering a lot of the stuff Apple's doing internally in terms of product development and [video]
Context & Ripple Effects
Apple's reported use of Anthropic for internal product work extends an earlier effort to use Claude in Xcode for coding, editing and testing, while a later external-model review reportedly found Anthropic's technology the strongest candidate for Siri. Apple's planned Claude-based Xcode work and the internal evaluation of outside Siri models therefore point to a long-running dependence on external model capability.
The reported decision to proceed with Gemini for Siri puts a commercial boundary around that technical preference: Apple can use Anthropic internally while selecting Google for the consumer-facing assistant after Claude's terms reportedly proved too expensive.
First-order effects
- Apple reportedly splits its AI stack: Anthropic supports internal development workflows, while Google Gemini is the chosen partner for the Siri integration.
- Anthropic forgoes a potentially central Siri deployment despite its models reportedly performing well in Apple's evaluation; Google gains the prominent customer-facing role.
Second-order effects
- Model vendors gain leverage when their systems are embedded in a customer's engineering workflows, but a high-stakes consumer deployment still depends on price and commercial terms, not model quality alone.
- Apple's dual-vendor approach reduces reliance on a single external model supplier while increasing the integration burden of maintaining distinct internal and consumer AI paths.
Third-order effects
- If this pattern persists, major device platforms may become multi-model orchestrators: selecting providers by workload, economics and deployment requirements rather than committing their entire AI strategy to one lab.
- The bargaining power in foundation models may increasingly shift toward vendors that can prove both technical advantage and terms acceptable for large-scale consumer distribution.
The trend: This is one example of large technology platforms turning foundation models into a strategically sourced, multi-vendor layer across internal tools and consumer products.