Apple’s AI-Focused Shift: Now Controlling All Core iPhone Chips

The iPhone Air, the latest entrant in Apple’s mobile lineup that debuted this Friday, isn’t just about sleek design. Nestled within its slim profile lies a hardware innovation signaling Apple’s intensified focus on artificial intelligence.

The Apple A19 Pro chip marks a significant architectural leap forward, integrating neural accelerators directly into each GPU core to exponentially boost computational power. This move is complemented by the introduction of the N1, Apple’s inaugural wireless chip for the iPhone, and the second generation of its in-house modem, the C1X. Industry analysts view these developments as a strategic maneuver by Apple (AAPL) to gain comprehensive control over the core silicon components of its smartphones.

“That’s where the magic truly happens. When we have end-to-end control, we can push the boundaries far beyond what’s achievable by simply purchasing off-the-shelf components,” stated Tim Millet, Apple’s Vice President of Platform Architecture, in an exclusive interview with CNBC at Apple Park this September, marking the first U.S. discussion on the revolutionary new silicon.

Historically, Broadcom (AVGO) served as the primary supplier of wireless and Bluetooth chips for iPhones. Although, Apple has a history with nearly a decade of making networking chips for the AirPods and Apple Watch. With the introduction of the N1, Apple now integrates its own wireless chip across the entire iPhone 17 series, including the iPhone Air.

Arun Mathias, Apple’s Vice President of Wireless Software Technologies and Ecosystems, highlighted the improvements driven by the N1’s Wi-Fi capabilities.

“It’s often overlooked, but Wi-Fi access points contribute significantly to a device’s location awareness, reducing reliance on power-intensive GPS,” Mathias explained. “By streamlining this process in the background, minimizing the need to wake the application processor, we achieve a significant efficiency gain.”

For iPhone modems, Qualcomm (QCOM) has been the exclusive provider since 2020. However, this changed in February with the unveiling of the C1 in the iPhone 16e. This move was set in motion back in 2019, when Apple acquired Intel’s modem business for $1 billion. Qualcomm had been proactively warning investors of this shift.

While Qualcomm modems remain in the iPhone 17, 17 Pro, and 17 Pro Max, Apple’s C1X is featured in the iPhone Air.

“Though Apple’s modem may not match Qualcomm’s peak throughput and raw performance just yet, the critical advantage lies in control and power efficiency. This translates directly into better battery life,” explained Ben Bajarin, CEO of Creative Strategies, a prominent technology research and consulting firm. Bajarin anticipates that Apple will “completely phase out” Qualcomm in the “next couple of years.”

Mathias stated that the C1X is “up to twice as fast” as the C1 and “uses 30% less energy” than the Qualcomm modem in the iPhone 16 Pro.

Despite the announcement, neither Qualcomm nor Broadcom experienced major market volatility, and both companies are expected to maintain licensing agreements with Apple for selected foundational technologies.

Apple’s new chip strategy surfaces amidst increasing pressure from Wall Street surrounding the company’s overarching AI strategy.

“It’s unlikely that they will ever develop a proprietary AI model akin to Google’s (GOOGL) or OpenAI’s,” Bajarin said. “They will still host those services on the iPhone. Their core focus will be optimizing the iPhone as the premiere platform for developers to deploy their AI applications.”

Since the launch of the A series with the iPhone 4 in 2010, Apple has increasingly focused on designing its own Systems on a Chip (SoC). The latest A19 Pro generation introduces a revamped chip architecture centered around accelerating AI workloads. The addition of neural accelerators to the GPU cores exemplify this focus.

“We are constructing the most efficient on-device AI capabilities currently available,” Millet told CNBC. “Our laser focus is to ensure that these phones now shipping, or soon to be shipped, are fully equipped for all the most critical on-device AI workloads on the horizon.”

Apple emphasizes privacy as a primary driving force behind its on-device AI approach, but Millet highlighted an important second factor.

“It’s exceptionally efficient for us. It delivers superior responsiveness. We have far greater control over the overall user experience,” he asserted.

Millet cited the new front camera as a notable example of “built-in AI.” The enhanced camera leverages AI to intelligently detect faces, automatically switching to landscape orientation when capturing group photos. “This feature harnesses the full power and range of capabilities within the A19 Pro,” Millet said.

Apple originally introduced its dedicated AI hardware, the Neural Engine, back in 2017, however, integrating compute power into GPUs has been accelerated over time.

“This level of neural processing integration now rivals MacBook Pro-class performance on your iPhone,” Millet explained. “This marks a pivotal step forward in machine learning computing. By embedding capabilities previously absent, such as dense matrix math in the Neural Engine, we have given developers access to advanced AI without significantly changing the underlying system.”

Bajarin told CNBC that Apple’s neural accelerators could mirror the functionality of tensor cores found in Nvidia’s (NVDA) AI chips, such as the H100.

“We’re integrating neural processing in a way that allows developers to write a program extending the instruction set, and have access to a new class of computer. This allows seamless switching between 3D-rendering and neural-processing instruction, all seamlessly inside the same microprogram,” explains Millet.

The iPhone 17 base model includes Apple’s previous generation A19 SoC, while the iPhone Air, iPhone 17 Pro, and 17 Pro Max incorporate the advanced A19 Pro.

Following past overheating concerns in the iPhone 15, Apple is now introducing a “vapor chamber” cooling solution into the Pro models to manage thermal performance of its custom silicon.

“The strategic placement of the vapor chamber directly alongside the A19 Pro SoC is critical,” noted Kaiann Drance, Apple’s Vice President of Worldwide iPhone Product Marketing. “We carefully considered how all the elements intersect, including the forged unibody aluminum design. This design promotes exceptional thermal conductivity and facilitates efficient heat dissipation. The laser welding into the design creates a secure metallic bond promoting effective heat transfer away from the processor core.”

Apple continues to source smaller components from third-party vendors, such as memory chips from Samsung and analog chips from Texas Instruments. However, the company is pushing vigorously to design all major core chips for its entire iPhone range in-house.

“We predict that bespoke Apple modems will find their way into the Mac and iPad lines and we will see ‘N’ variants of the networking chips in future Macs as well. This seems to be the next strategic step to unifying the hardware ecosystem across the entire Apple portfolio,” Bajarin stated.

When asked by CNBC whether neural accelerators will be incorporated into the GPU cores of the M5, the anticipated next-generation Mac SoC, Millet responded, “We have a unified approach to silicon architecture across the board.”

Apple plans to manufacture at least some of its custom chips in the United States, using facilities such as Taiwan Semiconductor Manufacturing Company’s (TSM) new campus in Arizona.

The Apple A19 Pro is presently manufactured on TSMC’s leading-edge 3-nanometer production node. While TSMC aims to commence domestic 3nm production in Arizona by 2028.

“If access to the absolute bleeding-edge in process technology is non-negotiable, then Taiwan remains the only viable option for the foreseeable future,” Bajarin asserted.

Moves toward chip production in the US have been hastened by recent policy.

That same day, Apple augmented its U.S. spending commitment to $600 billion over the next four years. CEO Tim Cook noted that a portion of this investment will be allocated to establish an “end-to-end silicon supply chain right here in America.”

“The impact of tariffs on the silicon supply chain is a critical concern for a lot of industry analysts. It is obviously why Apple and Tim Cook are on their mission and out there talking about investing in America,” said Bajarin.

Bajarin proposed that Apple could give struggling U.S. chipmaker Intel (INTC) “serious consideration if 14A really does deliver on all of its promises.” Although, he added, it’s “going to be awhile” before Intel “becomes a viable option.”

For now, Apple is committed to manufacturing chips at TSMC Arizona.

“We are excited about TSMC’s plans to grow in US manufacturing. Obviously it will help us from a time zone perspective, and we also greatly appreciate the diversification of the supply which is very important,” said Millet.

When asked about the potential amount of Apple’s $600 billion U.S. spend allocated to custom silicon, Millet said, “I truly hope it’s earmarked for silicon.”