According to 9to5Mac, Bloomberg’s Mark Gurman, in a new interview, stated that Apple currently “runs on Anthropic,” using custom versions of Claude on its own servers for internal product development tools. This revelation comes after Apple’s public announcement of an AI partnership with Google. Gurman notes Apple initially pursued a deal with Anthropic before settling with Google, but the talks collapsed because Anthropic wanted several billion dollars per year with fees that would double over time. In contrast, Apple’s deal with Google is reportedly costing the company just one billion dollars annually. The initial uncertainty around Apple and Google’s existing Safari search deal is what led Apple to prioritize talks with Anthropic and OpenAI first.
The Internal Reality
Here’s the thing: Gurman’s scoop makes perfect sense. Big tech companies often use one set of tools for their own engineers and a completely different set for what they ship to customers. It’s a classic “eat your own dog food” scenario, but with a third-party vendor. Apple building internal tools on Claude suggests its engineers find it genuinely useful for coding, prototyping, or analysis. That’s a powerful, quiet endorsement for Anthropic‘s tech. But it also highlights a weird split identity. Your iPhone might one day get AI features powered by Google Gemini, while the team that built your iPhone is using Claude. That’s… awkward.
The Billion-Dollar Standoff
Now, the money part is fascinating. Anthropic wanted *several* billion per year, with fees that would double? That’s an insane ask, even for Apple’s war chest. It tells you two things. First, Anthropic was (and probably still is) in massive demand, giving it serious pricing power. Second, Apple must have seen tremendous value to even entertain that conversation. The fact they walked away and got Google for a fraction of the price is a masterclass in supply chain negotiation. But it also raises a question: did Apple settle for a cheaper, potentially inferior partner for its consumer products because the premium one was just too rich? That’s a risky long-term bet if AI is the next platform war.
The Strategic Tightrope
So what’s Apple’s endgame here? They’re clearly hedging. They’re using Anthropic internally, partnering with Google for consumers, and we all know they’re working furiously on their own foundation models in the background. This multi-vendor approach avoids lock-in, but it’s also messy. It creates complexity and potential conflicts. And let’s be real—relying so heavily on a direct competitor’s infrastructure (Google) for a core future technology is a precarious position for a company that loves control. This whole saga feels less like a cohesive strategy and more like a series of reactive moves in a market that’s moving faster than Apple’s usual pace. They’re playing catch-up, and their checkbook is open.
