← Back to feed

Apple and Intel have reached a preliminary agreement for Intel to manufacture chips for Apple, according to reporting from Reuters and the Wall Street Journal confirmed by Bloomberg’s Mark Gurman. The deal marks the first chip-making relationship between the companies since Apple completed its transition away from Intel x86 processors with the M2 launch in 2023.

The agreement is preliminary. No production orders have been placed. But Apple has signed an NDA with Intel, received process design kit samples from Intel’s 18A-P node for evaluation, and Apple executives have made site visits to a Samsung fab under construction in Taylor, Texas — indicating the company is pursuing dual alternatives to its current exclusive reliance on TSMC.

Why Now

The trigger is not geopolitical risk, though that context is real. The trigger is supply. TSMC’s 3nm and beyond advanced nodes are saturated by AI hyperscaler demand. During Apple’s Q1 2026 earnings call in late April, outgoing CEO Tim Cook acknowledged that Mac mini and Mac Studio supply constraints would persist for months — and specifically cited agentic AI workloads as the demand source pushing both products into shortage. The bottleneck was not memory. It was advanced node capacity at TSMC, which AI data center orders now command ahead of consumer device production.

Apple shipped most of its M-series silicon on TSMC’s 3nm node through 2025. The A20 Pro and M6, expected in late 2026 and early 2027, will likely be the first Apple chips to reach 2nm. TSMC will almost certainly produce those. But Intel’s 18A-P process — Intel’s first node to support Foveros Direct 3D hybrid bonding and chiplet stacking — is now at commercial yield for Core Ultra Series 3 production, making it a credible candidate for lower-end M-series chips targeting 2027 volume.

Analysts at GF Securities and DigiTimes have reported since early 2026 that Apple was evaluating Intel’s 18A-P for its lowest-end M-series chips and non-Pro iPhone processors targeting 2028. The preliminary deal confirmed this week aligns with that timeline.

The Intel Angle

This is a significant commercial signal for Intel’s foundry recovery. CEO Lip-Bu Tan has been rebuilding Intel Foundry Services since taking over in early 2024 following Pat Gelsinger’s exit. Apple as a customer would be the highest-profile external win Intel has had in years, and the volumes involved — even for lower-end M-series parts — would run into the tens of millions of units annually.

Intel’s challenge is yield consistency. Industry analysis through 2026 confirms that both Intel and Samsung have narrowed the gap to TSMC on leading-edge nodes, but neither has fully closed it. For Apple, which requires consistent performance across high consumer volumes and has historically tolerated little margin for product-experience variance, even a small yield gap compounds at scale.

The most realistic scenario, by current reporting: Intel produces Apple’s lower-performance M-series chips (MacBook Air, iPad Pro) from 2027, while TSMC retains the highest-performance Apple silicon. Samsung’s Texas fab, not expected to reach high-volume 2nm production until 2028, is a longer-horizon option.

Broader Context

Apple is the most TSMC-concentrated major chip customer in the world. Any diversification of that relationship matters for the global semiconductor supply picture, but the driver here is specifically AI demand creating a queue problem at the world’s dominant advanced foundry. Apple is not choosing Intel because Intel got better faster — though it has. Apple is choosing Intel because TSMC’s capacity is being bid away by hyperscalers spending $130B+ per quarter on AI infrastructure.

That is a structural shift in how advanced node capacity gets allocated. Consumer devices now compete with data centers for foundry time, and data centers are winning.

Key Numbers

• Node: Intel 18A-P (1.8nm equivalent, supports Foveros Direct 3D bonding)
• Target products: Lower-end M-series (MacBook Air, iPad Pro class), non-Pro iPhone — targeting 2027+
• Samsung Texas fab: 2nm capacity expected 2028
• TSMC constraint cited by Tim Cook: advanced node capacity, not memory
• Apple annual M-series unit volume (reference): tens of millions per year
• Intel foundry milestone: first external Apple chip relationship since 2023