The Elephant in Every Product Meeting
John Ternus is about to walk into one of the most awkward boardrooms in tech. Not because he's unqualified—the man shipped every iPad variant and helped orchestrate the M-series transition. But because Apple just promoted someone whose entire career was built on making silicon you can touch, while the entire industry is screaming about language models you can't.
Apple didn't mention AI once in their CEO announcement. Not once. I noticed this immediately because when I integrate Claude API or OpenAI's GPT into a product, the first conversation with stakeholders is always: "Where does AI live in this?" It's the mandatory question now. So seeing Apple's succession announcement dodge it entirely felt like watching someone walk into a meeting pretending the Slack channel doesn't exist.
Here's what I think is happening—though I'm genuinely uncertain about this. Either Apple knows something about enterprise AI adoption that we don't, or they're banking on hardware differentiation mattering more than anyone thinks. Maybe both. Maybe neither.
What Hardware Guys Don't Always Understand About AI Integration
I spent three months last year building a feature request system powered by embeddings. The technical part—chunking documents, calculating vector similarity, connecting it to Pinecone—took two weeks. The other seven weeks were debugging why the hardware the feature ran on wasn't powerful enough for inference, why latency killed the user experience, and why nobody told us about memory constraints earlier.
This is where Ternus's background becomes actual advantage. He knows:
- How thermal limits kill performance
- The gap between what's theoretically possible and what ships without draining a battery in ninety minutes—honestly, this might matter more than we talk about
- Cost per unit at scale, which directly impacts how much compute you can embed versus defer to the cloud
- That making something elegant requires understanding the physical constraints nobody discusses in framework blogs
But hardware expertise also means you might underestimate how fast the AI landscape is actually moving. I'm not sure this is the right move, but the velocity of model improvements—we went from GPT-3.5 to GPT-4 to Claude 3 in what felt like eighteen months—is something that requires someone obsessed with software iteration. Ternus built iPads. Beautiful, fast iPads. He didn't build systems that need to adapt weekly because a new frontier model dropped.
The Real Problem: Siri Still Exists
Let's be direct. Siri is bad. Not "needs improvement"—bad. I've watched developers build better voice interfaces with open-source models and Hugging Face than Apple has with Siri after fifteen years and unlimited resources. That's either a product strategy choice (which is fine) or it's a failure of integration that haunts every Apple ecosystem narrative.
Ternus will own this. The new CEO, whoever they are, has to make a decision: do we rip out the Siri foundation and rebuild with modern architectures like we're seeing from competitors, or do we accept that voice isn't our differentiator?
I honestly can't predict which direction he'll choose. A hardware person might think voice interfaces are solved problems that shouldn't need reinvention. A software person would know that everything changed with transformer architectures.
What This Means for Builders
If you're integrating AI tools into products—whether that's RAG pipelines, fine-tuned models, or basic API calls to something like Together AI—you're already living in a world where Apple's strategy matters less than you might think. You're building on Python, running inference on GPUs you rent by the hour, and shipping updates whenever you want.
Apple's move signals something though. It says: we're betting that the interface layer and the silicon layer matter more than the model layer. That's not wrong. It's just incomplete.
Ternus starts September 1st. By then, we'll probably have another three new frontier models and a hundred hot takes about whether he's the right person for this moment.
I'm genuinely unsure what happens next.
