Apple's Chip Lead Lasted Two Weeks. Its Real Moat Compounds Forever.

On September 12, 2018, Apple held up a fingernail-sized slab of silicon with 6.9 billion transistors etched into it and called the A12 Bionic 'the industry's first 7nm chip.'¹ It was a thrilling claim, delivered with the usual confidence. It was also already false. Two weeks earlier, on August 31, Huawei had stood on a stage in Berlin and announced the Kirin 980 — also 7nm, also fabricated by TSMC, also 6.9 billion transistors.³ Same node, same fab, same transistor count, announced first by someone else. The honest version of Apple's claim is narrower: the A12 was the first 7nm chip to actually ship in a product you could buy.² First to the shelf, not first to the future.

The official story is that Apple's silicon is its moat — the magic chip nobody can match. That story dissolves the moment you notice a rival matched the headline spec in roughly fourteen days. A lead measured in weeks is not a moat. It is a sprint. So if the chip isn't what protects Apple, what does?

The node was shared. The advantage wasn't.

Here is the part the spec sheet hides. The 7nm process — the thing both companies fought to claim first — came from TSMC, and TSMC sold it to both of them. A manufacturing node is rented, not owned. What separated the two chips wasn't the node; it was what Apple put on it. Apple designed its own CPU cores, named Vortex and Tempest, and its own GPU. Huawei reached for ARM's reference parts off the shelf: Cortex-A76 cores, a Mali-G76 GPU.⁴ The gap showed up where it counts. In Geekbench single-core, the A12-powered iPhone XS scored about 4,821; the Kirin 980-powered Mate 20 Pro managed about 3,333.⁴ Same factory, same nanometers — and a roughly 45% performance chasm, opened entirely by the architecture stacked on top.

Custom design is a more durable edge than a node — but it is still an edge in the hardware itself, and hardware edges erode. Huawei matched the spec in a fortnight; competitors close architecture gaps over a few cycles. Even the A12's most futuristic feature points past the chip rather than at it: the Neural Engine jumped from the A11's 2 cores at 600 billion operations per second to 8 cores at 5 trillion⁵ — an eightfold leap whose entire purpose is to run software faster. The silicon, in other words, exists to serve something else. That something else is where the real moat lives.

The flywheel that runs on every device it sells

The thesis is simple, and it has nothing to do with nanometers. Apple's durable moat is the Services flywheel: each device it sells deepens the customer's entanglement with iCloud, iMessage, the App Store, and a dozen other recurring streams, and that entanglement converts a one-time hardware sale into a high-margin annuity. In fiscal 2024, Services threw off $96.2 billion in revenue — up nearly 13% year over year, and roughly a quarter of Apple's ~$391 billion total — at margins materially higher than hardware.⁷ The phone is the toll booth you buy once. The Services are the road you keep paying to drive.

Why is this a moat and the chip wasn't? Because it compounds, and it compounds in a way a rival can't shortcut. The mechanism runs in a loop. Each new device added to a household raises the cost of leaving — your messages, your photos, your purchased apps, your subscriptions all live inside the same walls. That stickiness keeps the installed base growing, which Apple reported hit an all-time high across every product category and region, with management pointing to 'very high levels of customer satisfaction and loyalty.'⁸ A larger, stickier base means more Services revenue, which funds more silicon and software, which makes the next device better, which adds more devices. The chip is one spoke. The flywheel is the whole wheel — and it turns faster the more it spins.

Isn't this just lock-in by another name — and isn't it fragile?

The fair objection is that 'flywheel' is a flattering word for being trapped, and that lock-in built on switching friction is exactly the kind of moat regulators love to drain — break open the App Store, mandate interoperable messaging, and the walls come down. That pressure is real, and Apple's own 10-K concedes its markets are 'highly competitive' with 'aggressive price competition and resulting downward pressure on gross margins.'⁶ But notice what the objection actually grants: it concedes the moat is the ecosystem, not the chip. And the reason the ecosystem is harder to pry open than it looks is that it isn't a single wall — it is silicon, OS, services, and 164,000 employees⁶ all tuned to each other. A competitor can rent TSMC's node, as Huawei did. It cannot rent the full silicon-to-software stack, because no one sells the integration. Regulation can lower the walls. It can't hand a rival the loop that took two decades to spin up.

Apple won the 7nm race by a fortnight and lost the framing entirely, because the chip was never the thing protecting it. The transistor count was matched, the node was shared, the headline was wrong. What couldn't be matched was the quiet machine the chip was built to serve: a base of devices too entangled to leave, throwing off ninety-six billion dollars of high-margin Services a year and growing. The moat was never the slab of silicon held up on stage. It was the road every one of those devices keeps paying to drive — and Apple owns the road.