June 9, 2026 · 9 min read

Even Apple Is Renting Its Brain

Let me tell you what actually happened on that WWDC stage yesterday.

Apple — the most valuable, most vertically integrated, most “we build our own everything” company on the planet — stood up and announced that the new Siri runs on a 1.2-trillion-parameter Google Gemini model. Licensed. For about a billion dollars a year.

Then they went further. iOS 27 ships with an Extensions system that lets the user decide which brain answers: Gemini, ChatGPT, or Claude. Three rented engines. Each with its own voice so you know who's talking.

Apple designs its own silicon. Its own operating system. Its own retail stores. And it just decided NOT to build the one thing everyone assumed it would. Sit with that.

The Build-Versus-Rent Decision Just Reached the Summit

Every leader I talk to is wrestling with the same question, whether they say it out loud or not: do we build our own AI capability, or do we rent someone else's? Train a model or call an API. Hire a team or license a platform. Own the stack or own the outcome.

For two years, the prestige answer has been BUILD. Building sounds like control. Building sounds like moat. Building sounds like the kind of thing a serious company does.

And then the company with more cash than most governments, the company that builds its own chips, looked at the build option and said: not worth it. Not at the frontier. Not at this pace. We'll rent the brain and pour our energy into the experience wrapped around it.

That should end the debate in most of your conference rooms. If Apple can't justify building its own frontier model, your 400-person logistics company almost certainly can't either. That's not an insult. It's arithmetic.

Renting Is Not Surrender. But It Is a Decision.

Here's where leaders get sloppy. They hear “rent the model” and they exhale, like the strategy work is done. It isn't. It just changed shape.

When you rent the brain, your competitive advantage CANNOT be the brain. Everyone can rent the same one. Your advantage has to live somewhere the vendor can't sell to your competitor: your proprietary data, your workflow design, your judgment about which 5% of decisions matter, the experience you wrap around the model.

Look at what Apple actually kept for itself. Not the model. The integration. The personal context. The on-device privacy layer. The two-billion-device distribution. They rented the commodity and doubled down on the things only Apple has.

That is the move. Rent what is becoming a commodity. Build what is uniquely yours. I walk through exactly how to draw that line — what to own, what to rent, and how to tell the difference before you burn a year and a budget — in The AI-First Leader. The single most expensive mistake I see is companies building commodity capability while renting out the parts that should have been their moat.

The Trap Hiding Inside the Model Picker

Notice the second half of Apple's announcement. The user picks the model. Gemini, ChatGPT, or Claude, swappable. Apple didn't just rent a brain — it engineered itself the freedom to FIRE that brain.

That is the part most companies skip. They sign a single-vendor deal, wire it into everything, and eighteen months later discover they've built their entire operation on a foundation they can neither inspect nor leave. The price goes up. The model changes behavior overnight. A capability they depended on gets deprecated in a release note. And they have no exit.

Apple, with a billion dollars on the line, built in optionality from day one. You should too — long before you're writing billion-dollar checks. Abstraction layers. Swappable providers. A second model qualified and ready. Concentration risk on a single AI vendor is the supply-chain exposure nobody put on the risk register, and it is sitting in your stack RIGHT NOW.

Strategy Is What You Refuse to Do

The real lesson here isn't about models. It's about discipline. Apple had every reason to chase the shiny thing — ego, capability, a war chest, an army of engineers. And it declined. It defined precisely where it would compete and refused to spend energy everywhere else.

That is strategic execution in the age of autonomous systems: knowing the difference between the work that compounds your advantage and the work that just feels impressive in a board deck. Most AI strategies fail not because the company chose wrong between build and rent, but because it never honestly answered the prior question — what are we actually trying to win? I built an entire framework around that discipline in Geometric Precision, because strategy without execution is a wish, and execution without strategy is just expensive motion.

Apple just gave every leader a free case study. The most capable company in the world looked at the frontier, did the math, and rented the brain so it could build the things that matter.

The question isn't whether you're humble enough to rent. It's whether you're disciplined enough to know what to build instead.