Intel, Apple's Chips Are Close: Testing for 18A-P Has Already Started for iPhone, iPad, and Mac
Since 2016, TSMC has effectively held a monopoly on the production of Apple chips: a ten-year exclusive relationship that has solidified alongside the rise of Apple Silicon and the departure of Intel processors from Macs. Following last week's news and recent rumors, new details have emerged from analyst Ming-Chi Kuo of TF International that further confirm how this monopoly is about to come to an end.
Kuo: Testing Started on 18A-P with Foveros
Yesterday, Kuo published on X the results of his latest checks on Apple's supply chain: Intel has begun small-scale production of processors for low-end iPhone, iPad, and Mac devices, using the 18A-P process with Foveros packaging technology. The roadmap outlined by the analyst spans four years: small-scale testing in 2026, production launch in 2027, growth in 2028, and gradual slowdown in 2029, consistent with the expected lifecycle for the 18A-P node.
It's worth noting that current orders are concentrated approximately 80% on iPhone processors, a proportion that reflects the weight of the iPhone in the Apple ecosystem. The chips in question are classified by Kuo as "low-end/legacy": processors intended for lower-end models or devices still marketed but no longer at the forefront of the production process. Kuo did not specify the exact designations of the chips involved, although WCCFTech, citing the same supply chain checks, indicates the standard A21, the non-Pro version, as the main candidate for the most significant volumes around 2028.
Yield and Timing: The Critical Node
On the production yield front, Intel aims to stabilize the yields of the 18A-P process between 50% and 60% by 2027. This is the minimum threshold expected to initiate shipments in commercial volumes but not an especially ambitious goal for a production line that aims to compete with the industry's benchmark. TSMC has significantly surpassed similar levels on its advanced nodes for years. Intel, with its 18A-P, aims to achieve them in a still multi-year horizon, and the start of testing in 2026 leaves little margin if the goal is to ship significant volumes already in 2027.
According to TechConstant, Apple has already obtained PDK samples of the 18A-P process from Intel, with internal simulations producing results sufficient to justify awaiting versions 1.0 and 1.1 of the PDK, released in the first quarter of 2026. The estimated annual volume for chips produced by Intel is placed, according to TechConstant, between 15 and 20 million units.
Foundry, Not Architecture Supplier
It's important to reiterate that Intel's role in this architecture is strictly limited to manufacturing. Apple designs the chips; Intel produces them, a setup entirely similar to the one governing its relationship with TSMC and radically different from the era of Intel Macs, in which Cupertino purchased processors designed and produced by Intel with x86 architecture. Apple started the transition to Apple Silicon in 2020 precisely to free itself from architectural dependence on Intel, and that dependence is not part of this agreement.
The Weight of Industrial Policy
The context in which this agreement is embedded is far from neutral. The pressure from the Trump administration to shift manufacturing production to American soil has been an explicit factor in Kuo's analyses. Intel is a U.S. company, also supported by a direct intervention from the federal government, which acquired a 10% stake in the group to prevent its financial deterioration. The agreement with Apple gives Intel a long-term contract and provides Apple with grounds for negotiation with Washington, at a time when the White House has already requested direct manufacturing investments from leading American companies. The preliminary agreement between Apple and Intel was reported by the Wall Street Journal on May 8, following over a year of negotiations. An official announcement from Apple has yet to arrive.
However, we must note that the operational scope of this diversification remains limited. TSMC will maintain, according to Kuo's projections, over 90% of Apple's chip supply even when Intel is running at full capacity. Any assessment of the impact on dependence on TSMC must contend with this figure.
In short, the partnership with Intel is real and operational, but for now confined to a precise perimeter: legacy chips, 18A-P node, and marginal volumes compared to Apple's total scale. On the level of geopolitical risk diversification, it is a first step, but in terms of actual rebalancing among suppliers, the story is more nuanced: a residual exposure of 90% on a single manufacturer, with over 60% of TSMC's volumes concentrated in Taiwan, remains a vulnerability that no Intel contract, in its current form, can cover.