Apple's Chip Strategy Keeps Paying Dividends -- Barrons.com

By Adam Levine

When Apple's iPhone turned into an immediate hit upon its 2007 release, CEO Steve Jobs was quick to realize the value of designing and owning as many of the components in its devices as possible. By 2008, it was working on a smartphone chip and in 2010 the iPhone 4 shipped with the first Apple chip, the A4. Since then, designing the main chips for all its devices has been a priority for Apple, and a potent differentiator, enabling deeper integration between software, services, and hardware than any other consumer technology company.

This week's release of iPhone 18 Pro and the A20 Pro chip inside of it, makes the advantages of Apple's chip strategy as clear as day. In single-core computing-the majority of what people do on their devices-the A20 Pro is faster than the best PC chips from Intel, Advanced Micro Devices, or Qualcomm, according to GeekBench benchmarks, and remains better than the high-end Android smartphone chips from Qualcomm and MediaTek.

Apple's smartphone chips are so good that one from 2024-the A18 Pro-can power the relatively inexpensive MacBook Neo that has been a big seller for Apple. Because it can seamlessly marry hardware and software, the company is able to deliver better performance, with less power consumption.

Conventional benchmarks measure the main brain of the chips: the CPU. But that now takes up a minority of the space on what has become known as a "system-on-a-chip," or SoC. The rest of the SoC demonstrates Apple's priorities for its devices. For example, Apple often markets battery life in terms of how many hours of video playback a device can deliver-up to 36 hours for the iPhone 18 Pro. The reason it can pull that off is because the A20 Pro SoC has a special block for decoding video files for playback without drawing as much power as the CPU or graphics unit would.

The graphics processing unit has hardware raytracing for better gaming quality. There is an image signal processor to improve photos. The Secure Enclave, meanwhile, is a walled-off portion of the SoC that holds all users' deepest secrets like encryption keys and FaceID data, and has special cryptography accelerators helping it out.

Apple's current artificial-intelligence priorities are also evident in its 2026 chip designs. The company favors on-device AI processing rather than cloud-based, because it is much more private and secure. Where more computing power is required, Apple relies on its custom-built Private Cloud Compute servers and software which hide what's happening from everyone but the user, including Apple. The company wants to be the pre-eminent provider of safe, private, secure consumer AI built around each user personally-positioning Meta Platforms is eager to emulate.

The A20 Pro has new AI accelerators built into the graphics unit, and the size of the neural engine, which is now one of the largest blocks on the chip, has doubled from the previous generation. It allows on-device AI computing to happen without being a big drain on the battery, as it would be on a CPU or graphics unit.

But the most exquisite symphony of hardware, software, and services design in this year's devices comes from a new AI-powered Apple Watch feature called Siri Rewind. It creates a high-level summary of the wearer's conversations throughout the day without compromising on privacy. To pull this off, the S11 chip in this year's Apple Watches has a new walled-off block called the Secure Exclave, which communicates over a secure channel with the Secure Exclave on a paired iPhone's chip. Though it wasn't publicly revealed until last week, iPhone chips have had this unit since 2024, lying dormant until needed for this release. As a result, Siri Rewind will work with the iPhone 16 or newer phones.

Clips of audio are recorded on the Watch Secure Exclave, encrypted and then sent to the iPhone where they are transcribed and condensed. The audio and transcription are then disposed of, and the condensed text is sent to Private Cloud Compute to put it all together into a usable summary and add any additional context, like corresponding appointments on the user's calendar. This one feature is the result of years of privacy-enhancing chip, software and cloud services development and design.

It is Apple's AI strategy in a nutshell. Unlike the "hyperscaler" cloud companies spending over $700 billion on new AI data centers this year, Apple has been much more stingy, and arguably more pragmatic for a company that makes most of its money from hardware. Instead of burning free cash flow on the AI arms race, it is focusing on building an ecosystem of AI services that rely on chips that enable on-device AI computing, paired with Private Cloud Compute for more demanding tasks, and software which ties it all together and harnesses the abundant data on user's devices to provide genuinely personalized information and insights.

This distinct vision is showing up in the market. For more than a month now, Apple stock has been trading as an AI-alternative to data center plays like chip stocks. In the 26 trading days since Aug. 11, Apple shares and the iShares Semiconductor ETF have traded in opposite directions on 19 of them. So far in this duel, Apple is winning, with its shares rising 8%, and the chip ETF down 5% through Wednesday.

Write to Adam Levine at adam.levine@barrons.com

This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

September 18, 2026 02:00 ET (06:00 GMT)