John Ternus becomes CEO of Apple on Tuesday, taking over a company whose defining strength has always been tight control of its own technology stack - hardware, software, and the chips connecting them - but whose current artificial intelligence strategy runs largely on outside partners. Tim Cook will become executive chairman the same day, managing Apple's relationships with governments and global policies while Ternus, a 24-year Apple veteran who most recently ran hardware engineering, takes the CEO title. It is Apple's first CEO transition since Cook succeeded Steve Jobs in 2011.
Ternus inherits a specific, high-stakes bet Cook already made. Apple agreed in January to pay Google roughly $1 billion a year to license a custom version of Gemini, Google's AI model, to power a rebuilt Siri capable of cross-app actions and multi-turn conversation. Apple said at the time it was making no major changes to its existing ChatGPT integration, which lets Siri hand off complex queries to OpenAI's model. That arrangement now appears secondary, since Gemini handles the core Siri experience. Both partnerships mean Apple's most consequential AI features run on models Apple did not build itself.
First public test comes days into the job
Apple will hold its annual showcase on Sept. 9, titled "Surprise and Shine" this year. The event, which dates back to the Steve Jobs era, is expected to unveil Apple's first foldable iPhone alongside the rebuilt, Gemini-powered Siri. Gene Munster, managing partner at Deepwater Asset Management, said before the announcement that the new Siri needed to be a "10 out of 10" when it arrived. That bar has effectively become Ternus' opening review as CEO rather than a one-time product launch he can distance himself from.
The challenge is not simply catching up to OpenAI or Google on raw model capability. It is proving that Apple can build products people trust and prefer around someone else's AI model, the same way it has spent two decades building devices around chips, displays, and components it did not always manufacture itself, without losing the product experience, cost control, or differentiation that outside dependency usually threatens. A Siri built primarily on licensed technology only works commercially if it still feels distinctly like an Apple product once a user is talking to it, not simply a wrapper around Gemini with an Apple logo attached.
The better Siri works, the more it costs
That calculation gets harder if the new Siri succeeds. Apple has built its AI approach around a capex-light strategy, renting AI capability from partners like Google rather than building the data center infrastructure that rivals including Meta, Alphabet, and Microsoft are spending well over $100 billion apiece on this year. For executives weighing similar build-versus-buy decisions, the AI for Executives & Strategy resource library covers the tradeoffs involved in outsourcing AI capability versus developing it internally.
Cook addressed that tension directly on Apple's July 30 earnings call, his final one as CEO, with Ternus on the line. "We'll see what Siri AI does from the cost side of it," Cook said. "There's also the ability when people use it a lot for them to move up on an iCloud Plan as well, and so what the balance of that is, is a bit uncertain at the moment." That uncertainty is the exact decision Ternus inherits along with the CEO title: how Apple pays for Siri AI's compute costs once real usage starts coming in.
Ternus does not need to build the best AI model in the industry. He needs to prove Apple can keep building the best products in the industry around AI models it does not own, at a cost structure that does not erode the margins those products have always depended on. The AI Learning Path for CEOs addresses this type of strategic challenge directly: how senior leaders evaluate AI partnerships, licensing arrangements, and infrastructure investment without ceding control of the product experience.
Why this matters for executives and strategy
Apple's situation is a live case study in a decision many companies now face: whether to build, buy, or rent AI capability. The stakes are visible in the numbers - $1 billion annually to Google for Gemini, versus the $100 billion-plus infrastructure bets Meta, Alphabet, and Microsoft are making. Ternus' success or failure will test whether a product company can preserve differentiation and margins while depending on a competitor's core technology. For executives, the lesson is not that one approach is right. It is that the cost structure of AI partnerships changes as usage scales, and those second-order costs need to be priced into the decision before the contract is signed, not discovered on an earnings call afterward.
Your membership also unlocks:
