Apple Inc.’s three new iPhones contain a cluster of chips that the company designed itself, including an artificial-intelligence chip that literally will make its smartphones even smarter, as the world’s most valuable public company thrusts itself into leadership in a new area.
Inside Apple’s AAPL, -1.24% three new sleek and glossy iPhones that it launched Wednesday is a system-on-a-chip module dubbed A12 Bionic. Made up of a core central processing unit, a graphics processing unit and a neural net chip that handles machine-learning instructions, the A12 Bionic chip module has a total of 6.9 billion transistors and is manufactured using the most advanced process technology in the industry.
“What the team has done is truly, truly breakthrough,” Apple’s senior vice president of worldwide marketing, Phil Schiller, said at the company’s event in Cupertino, Calif., hailing “the industry’s first 7-nanometer chip.”
While Chinese hardware giant Huawei announced last month a 7-nanometer process technology chip, called the Kirin 980, Apple will be first to market, with its iPhones set to be available on Sept. 21. Huawei’s products with the Kirin 980 are due sometime in October. And while some in the industry may quibble about that difference, Apple — not Intel, not Advanced Micro Devices Inc. — will be first to market with products using the new process technology, thanks to its manufacturing partner Taiwan Semiconductor Manufacturing Co. Ltd. 2330, +0.19% which is also making Huawei’s chips.
The Apple event, which started out with a parody of a scene from the latest “Mission: Impossible” movie, with an employee racing through the company’s new spaceship campus, was an important demonstration of its growing custom semiconductor prowess. While investors sent shares down 1.2% on the day as they digested the company’s new pricing strategy for iPhones, the reaction to the new technology — especially the chips — was roundly positive.
Apple’s CPU for the iPhone is based on an ARM Holdings design, the most popular chip for mobile phones, but it is customized, and its graphics chip and its AI chip appear to be designed in-house by its growing chip-design team, built up by Apple’s senior vice president of hardware technologies, Johny Srouji, over the past decade.
In typical Apple fashion, the company is quite secretive about what’s under the hood of its products. “They never go into a lot of detail,” said Kevin Krewell, a principal analyst at Tirias Research. “Apple doesn’t go to conferences like the ISSCC or Hot Chips. You never really get all the details.”
But with Apple touting its use of the 7-nanometer process technology, which is more advanced than Intel Corp.’s INTC, +0.00% current manufacturing methods, it made a leap forward that could raise more questions about whether Apple will continue to use Intel chips to power its Mac personal computers. The use of smaller processing nodes means that more transistors and processing power can be squeezed onto a silicon wafer. Intel, once the world’s largest semiconductor maker, is now moving to 10 nanometers from 14 nanometers, but it is not yet in full production of its 10-nanometer process technology.
“I don’t see this endangering Intel, if Intel can hit its 2019 10nm commitments,” said Patrick Moorhead, president and principal analyst of Moor Insights & Strategy. “I equate Intel 10nm and TSMC 7nm. The characteristics are quite similar.”
But Apple has also shown it is ahead in the use of machine learning, incorporating these features in its chipset module for the iPhone. Schiller said in his demonstration that Apple’s system on a chip uses machine learning in a range of ways, from helping consumers take their best photos, to creating immersive augmented-reality experiences, to new applications that will be created in the future by the huge ecosystem of software developers.
Many large tech companies have begun to start development of their own chips dedicated to machine learning, but Apple is showing actual results.
“Apple has a lot planned for that neural net processor in the future,” Krewell said. “We are getting to a new phase of compute technology…Neural processing is becoming a more mainstream type of computing element that you will see in more and more processors going forward. So far Intel has not made a move to put a neural net in any of their CPUs…This is where the PC may fall behind because the neural net processing is going to become more important.”
But developers need to create even better applications that take advantage of the artificial-intelligence capabilities of the new chips, in order for consumers to justify the higher prices for the new iPhones. And they will need to be better applications than just being able to pick out the best cat photo on your iPhone.
Want this type of analysis sent to your inbox? Subscribe to MarketWatch's free MarketWatch First Takes newsletter. Sign up here.
Also on Market Tamer…
Follow Us on Facebook