Something unusual is happening at Apple next Tuesday, and it's not a keynote. On September 9, John Ternus officially takes over as CEO, and he's inheriting something Tim Cook spent years quietly setting up: the widest product refresh cycle Apple has attempted in over a decade. Livemint's rundown frames it as Apple building something while everyone else chased AI, and the framing mostly holds up, even if I'd quibble with the implication that Apple ignored AI entirely (the ongoing Siri work says otherwise).

Let's start with the hardware, because that's the part I actually understand at the silicon level. The iPhone 18 lineup is expected to move fully to 2-nanometer process nodes, reportedly following a path Apple's silicon team has already started down. When you shrink transistors from 3nm to 2nm, you're not getting a marginal bump, you're getting meaningfully better performance-per-watt, and in a phone that translates directly to battery life or sustained performance, pick your poison. Heat is the enemy of everything in a phone. Every watt you save is screen brightness or GPU headroom you get back.

And it's not just the iPhone. The wave reportedly includes the foldable iPhone that's been leaking in increasingly confident terms, a refreshed Vision line, a smart home hub, new AirPods, watch updates, and the Mac refresh train continuing on schedule. That's a lot of simultaneous product lines all hitting maturity at once.

Why do I think the timing matters more than the products individually? A few reasons.

Ternus is a hardware engineer. He ran product engineering, he's the guy who stood on stage and explained the M1 transition, and now he's CEO at the exact moment Apple's entire stack converges on custom silicon: modems, radio frequency chips, Wi-Fi and Bluetooth modules, the works. Apple spent years pulling component after component in-house while Google and Meta and Microsoft poured billions into foundation models. Both bets were expensive. But Apple's bet produced a pile of physical goods it can ship in a single fiscal year, and the other bet produced, well, an extremely impressive chatbot attached to a business model nobody has fully solved yet.

Not that Apple skipped AI. The whole industry is clearly drifting toward running more inference on-device rather than renting GPUs in someone else's data center, and Apple's custom-silicon-first approach is looking less old-fashioned every quarter. When the marginal cost of a token falls on someone else's data center, your margins bleed. When it falls on a chip you already sold the customer, the story changes entirely.

As a developer I keep thinking about the second-order effects. A 2nm iPhone 18 with a serious neural engine and a foldable form factor changes what apps can assume. If Apple opens up the kind of on-device model access that local AI models are already enabling on consumer hardware, you could see genuinely local, private AI agents running in your pocket within a couple of years, and the App Store economics around that would be enormous. That's speculative. But the hardware runway for it is being poured right now.

There are risks, obviously. A foldable iPhone enters a market where Samsung has been iterating on hinge and crease engineering for the better part of a decade. Apple being late has historically meant Apple being better, but Samsung isn't standing still either. And every CEO transition has a stumble in it somewhere; Ternus gets no honeymoon period, he gets a product wave where anything that slips becomes The Story.

The thing I'll actually be watching on September 9 isn't the ceremony. It's whether the first keynote under Ternus leads with silicon and hardware integration the way his presentations always have, or whether he forces himself to lead with the AI story anyway because the market expects it. That choice tells you what kind of CEO he's going to be.

Either way, the next twelve months at Apple look busier than any stretch since maybe the original Watch launch. I've got my calendar marked. My wallet is already nervous...