Apple’s soaring stock price over the past two decades has been driven by its iconic consumer devices. It started with the iPod and iMac. Then came the iPhone and iPad. And more recently, the Apple Watch and AirPods.
But there’s a lot more to the biggest U.S. company by market cap than just gadgets. At its Silicon Valley headquarters, in a non-descript room filled with a couple hundred buzzing machines and a handful of engineers in lab coats, Apple is designing the custom chips that power its most popular products.
Apple first debuted homegrown semiconductors in the iPhone 4 in 2010. As of this year, all new Mac computers are powered by Apple’s own silicon, ending the company’s 15-plus years of reliance on Intel.
“One of the most, if not the most, profound change at Apple, certainly in our products over the last 20 years, is how we now do so many of those technologies in-house,” said John Ternus, who runs hardware engineering at Apple. “And top of the list, of course, is our silicon.”
That change has also opened Apple up to a new set of risks. Its most advanced silicon is primarily manufactured by one vendor, Taiwan Semiconductor Manufacturing Company. Meanwhile, smartphones are recovering from a deep sales slump, and competitors like Microsoft are making big leaps in artificial intelligence.
In November, CNBC visited Apple’s campus in Cupertino, California, the first journalists allowed to film inside one of the company’s chip labs. We got a rare chance to talk with the head of Apple silicon, Johny Srouji, about the company’s push into the complex business of custom semiconductor development, which is also being pursued by Amazon, Google, Microsoft and Tesla.
“We have thousands of engineers,” Srouji said. “But if you look at the portfolio of chips we do: very lean, actually. Very efficient.”
Unlike traditional chipmakers, Apple is not making silicon for other companies.
“Because we’re not really selling chips outside, we focus on the product,” Srouji said. “That gives us freedom to optimize, and the scalable architecture lets us reuse pieces between different products.”
Powering iPhones since 2010
Srouji came to Apple in 2008 to lead a small team of 40 or 50 engineers designing custom chips for the iPhone. A month after he joined, Apple bought P.A. Semiconductor, a 150-person startup, for $278 million.
“They’re going to start doing their own chips: that was the immediate takeaway when they bought P.A. Semi,” said Ben Bajarin, CEO and principal analyst at Creative Strategies. With its “inherent design focus,” Bajarin said, Apple wants “to control as much of the stack” as possible.
Two years after the acquisition, Apple launched its first custom chip, the A4, in the iPhone 4 and original iPad.
“We built what we call the unified memory architecture that is scalable across products,” Srouji said. “We built an architecture that you start with the iPhone, but then we scaled it to the iPad and then to the watch and eventually to the Mac.”
Apple’s silicon team has grown to thousands of engineers working across labs all over the world, including in Israel, Germany, Austria, the U.K. and Japan. Within the U.S., the company has facilities in Silicon Valley, San Diego and Austin, Texas.
The primary type of chip Apple is developing is known as a system on a chip, or SoC. That brings together the central processing unit (CPU), graphics processing unit (GPU) and other components, Bajarin explained, adding that for Apple there’s also a neural processing unit (NPU) “that runs the neural engine.”
“It is the silicon and all of the blocks that go on to that silicon,” Bajarin said.
Apple’s first SoC was the A series, which has advanced from the A4 in 2010 to the A17 Pro announced in September of this year. It’s the central processor in iPhones, as well as some iPads, Apple TVs and the HomePod. Apple’s other major SoC is the M series, first released in 2020, which now powers all new Macs and more advanced iPads. That product is up to the M3 line.
Launched in 2015, the S series is a smaller system in package, or SiP, for Apple Watch. H and W chips are used in AirPods. U chips allow communication between Apple devices. And the newest chip, the R1, is set to ship early next year in Apple’s Vision Pro headset. Dedicated to processing input from the device’s cameras, sensors and microphones, Apple says it will stream images to the displays within 12 milliseconds.
“We get to design the chips ahead of time,” Srouji said. He added that his staffers work with Ternus’s team “to exactly and precisely build chips that are going to be targeted for those products, and only for those products.”
The H2 inside the 2nd generation AirPods Pro, for instance, enables better noise cancellation. Inside the new Series 9 Apple Watch, the S9 allows for unusual capabilities like double tap. In iPhones, the A11 Bionic in 2017 had the first Apple Neural Engine, a dedicated part of the SoC for performing AI tasks totally on-device.
The latest A17 Pro announced in the iPhone 15 Pro and Pro Max in September enables major leaps in features like computational photography and advanced rendering for gaming.
“It was actually the biggest redesign in GPU architecture and Apple silicon history,” said Kaiann Drance, who leads marketing for the iPhone. “We have hardware accelerated ray tracing for the first time. And we have mesh shading acceleration, which allows game developers to create some really stunning visual effects.”
That’s led to the development of iPhone-native versions from Ubisoft‘s Assassin’s Creed Mirage, The Division Resurgence and Capcom‘s Resident Evil 4.
Apple says the A17 Pro is the first 3-nanometer chip to ship at high volume.
“The reason we use 3-nanometer is it gives us the ability to pack more transistors in a given dimension. That is important for the product and much better power efficiency,” Srouji said. “Even though we’re not a chip company, we are leading the industry for a reason.”
Replacing Intel in Macs
Apple’s leap to 3-nanometer continued with the M3 chips for Mac computers, announced in October. Apple says the M3 enables features like 22-hour battery life and, similar to the A17 Pro, boosted graphics performance.
“It’s early days,” said Ternus, who’s been at Apple for 22 years. ”We have a lot of work to do, but I think there’s so many Macs now, pretty much all Macs are capable of running Triple-A titles, which is not what it was like five years ago.”
Ternus said that when he started, “the way we tended to make products is we were using technologies from other companies, and we were effectively building the product around that.” Despite a focus on beautiful design, “they were constrained by what was available,” he said.
In a major shift for the semiconductor industry, Apple turned away from using Intel’s PC processors in 2020, switching to its own M1 chip inside the MacBook Air and other Macs.
“It was almost like the laws of physics had changed,” Ternus said. “All of a sudden we could build a MacBook Air that’s incredibly thin and light, has no fan, 18 hours of battery life, and outperformed the MacBook Pro that we had just been shipping.”
He said the newest MacBook Pro with Apple’s most advanced chip, the M3 Max, “is 11 times faster than the fastest Intel MacBook Pro we were making. And we were shipping that just two years ago.”
Intel processors are based on x86 architecture, the traditional choice for PC makers, with a lot of software developed for it. Apple bases its processors on rival Arm architecture, known for using less power and helping laptop batteries last longer.
Apple’s M1 in 2020 was a proving point for Arm-based processors in high-end computers, with other big names like Qualcomm — and reportedly AMD and Nvidia — also developing Arm-based PC processors. In September, Apple extended its deal with Arm through at least 2040.
When its first custom chip came out 13 years ago, Apple was unusual as a non-chip company trying to make it in the cutthroat, cost-prohibitive semiconductor market. Since then, Amazon, Google, Microsoft and Tesla have tried their hand at custom chips.
“Apple was sort of the trailblazer,” said Stacy Rasgon, managing director and senior analyst at Bernstein Research. “They sort of showed that if you do this, you can have a stab at differentiating your products.”
‘Modems are hard’
Apple isn’t yet making every piece of silicon in its devices. Modems, for example, are one big component the company has yet to conquer on its own.
“The processors have been remarkably good. Where they’ve struggled is on the modem side, is on the radio side in the phones,” Rasgon said. “Modems are hard.”
Apple relies on Qualcomm for its modems, although in 2019, the two companies settled a two-year legal battle over intellectual property. Soon after, Apple bought the majority of Intel’s 5G modem business for $1 billion, in a likely move to develop its own cellular modem. That hasn’t happened yet, and in September, Apple signed on with Qualcomm to supply its modems through 2026.
“Qualcomm still makes the best modems in the world,” Bajarin said. “Until Apple can do as good of a job, I have a hard time seeing them fully jump to that.”
Apple’s Srouji said he couldn’t comment on “future technologies and products” but said “we care about cellular, and we have teams enabling that.”
Apple is also reportedly working on its own Wi-Fi and Bluetooth chip. For now, it has a fresh multibillion-dollar deal with Broadcom for wireless components. Apple relies on third parties like Samsung and Micron for memory.
“Our aspiration is the product,” Srouji said, when asked if Apple will try to design every part of its chips. “We want to build the best products on the planet. As a technology team, which also includes the chips in this case, we want to build the best technology that would enable that vision.”
To deliver on that objective, Apple will “buy off the shelf” if it means the team can focus “on what really, really matters,” Srouji said.
Regardless of how much silicon Apple eventually designs, it still needs to manufacture its chips externally. That requires massive fabrication plants owned by foundry companies like TSMC.
More than 90% of the world’s advanced chips are made by TSMC in Taiwan, which leaves Apple and the rest of the industry vulnerable to the China threat of invasion.
“There is obviously a lot of tension around, like, what would plan B be if that happened?” Bajarin said. “There isn’t another good option. You would hope that Samsung is also competitive and Intel wants to be there. But again, we’re not right now. It’s really all at TSMC.”
Apple is at least looking to bring some of that manufacturing to the U.S. It’s committed to becoming the largest customer at TSMC’s coming fab in Arizona. And on Thursday Apple announced it will be the first and largest customer of the new $2 billion Amkor manufacturing and packaging facility being built in Peoria, Arizona. Amkor will package Apple silicon produced at TSMC’s Arizona fab.
“We always want to have a diversified supply: Asia, Europe and the U.S., which is why I think TSMC building fabs in Arizona is great,” Srouji said.
Finding talent
Another concern is the shortage of skilled chip labor in the U.S., where advanced fabs haven’t been built for decades. TSMC says its Arizona fab is now delayed to 2025 because of a lack of skilled workers.
Whether or not it has to do with a shortage of talent, Apple has seen a slowdown in the release of new chips.
“Generations are taking longer because they are getting harder and harder,” Srouji said. “And the ability to pack more and get power efficiency is also different than 10 years ago.”
Srouji reiterated his view that Apple has an advantage in that regard because “I don’t need to worry about where do I send my chips, how do I target a larger customer base?”
Still, Apple’s actions underscore the competitiveness in the market. In 2019, Apple chip architect Gerard Williams left to lead a data center chip startup called Nuvia, bringing some Apple engineers with him. Apple sued Williams over IP concerns, before dropping the case this year. Qualcomm bought Nuvia in 2021, in a move to compete in Arm-based PC processors like Apple’s.
“I can’t really discuss legal matters, but we truly care about IP protection,” Srouji said. “When certain people leave for certain reasons, that’s their choice.”
Apple has additional macro challenges in its core business because smartphone sales are just recovering from their lowest levels in years.
However, demand for AI workloads is leading to a surge in orders for silicon, especially for GPUs made by companies like Nvidia, whose stock has jumped more than 200% this year tied to the popularity of ChatGPT and other generative AI services.
Google has designed a tensor processing unit for AI since 2016. Amazon Web Services has had its own AI chips for the data center since 2018. Microsoft released its new AI chip in November.
Srouji said his team at Apple has been working on its machine learning engines, the Apple Neural Engine, since years before it was launched in the A11 Bionic chip in 2017. He also pointed to embedded machine learning accelerators in its CPU and “highly optimized GPU for machine learning.”
Apple’s Neural Engines power what it calls “on-device machine learning features” like Face ID and Animojis.
In July, Bloomberg reported that Apple built its own large language model called Ajax and a chatbot called Apple GPT. A spokesperson declined to confirm or deny the accuracy of the report.
Apple has also acquired more than two dozen AI companies since 2015.
When asked if Apple appears to be falling behind in AI, Srouji said, “I don’t believe we are.”
Bajarin is more skeptical.
“It’s doable on Apple’s last year chip, even more capable on this year’s chip with M3,” Bajarin said, regarding Apple’s position in AI. “But the software has got to catch up with that, so that developers take advantage and write tomorrow’s AI software on Apple Silicon.”
He anticipates improvements, and soon.
“Apple had an opportunity to really get on that from day one,” Bajarin said. “But I think everyone expects it’s coming in the coming year.”
Watch the video to learn more.