Intel's foundry operations are experiencing multi-front breakthroughs after a prolonged period. On April 28, according to TrendForce, Apple is reportedly evaluating the use of Intel's 18A-P process for manufacturing its M-series chips, while Google is considering leveraging Intel's EMIB advanced packaging technology for its TPU v8e project. Concurrently, Tesla has confirmed it will adopt Intel's next-generation 14A process for its Terafab AI complex located in Austin.
The convergence of these positive signals makes the recovery narrative for Intel's foundry business increasingly clear. As mentioned in a Wall Street Insights article, during last week's earnings call, Intel CEO Lip-Bu Tan noted that as inference and agent AI scale up, demand for server CPUs is rebounding. The ratio of CPUs to GPUs is shifting from approximately 1:8 towards 1:1, a structural change that is directly boosting external interest in Intel's foundry platform.
Financially, the latest earnings report shows that the operating loss for Intel's foundry segment narrowed to $2.4 billion in the first quarter, an improvement of $72 million sequentially, with yield improvements being the primary driver.
Reports indicate that Apple and Google are evaluating Intel's platform. Media reports citing supply chain sources state that Apple is assessing the migration of M-series chips to Intel's 18A-P node, while Google is exploring the use of Intel's EMIB (Embedded Multi-die Interconnect Bridge) advanced packaging technology for its TPU v8e project. Related products could materialize as early as 2027.
It is important to note that this information currently remains at the media report level and has not been officially confirmed by Apple or Google. Uncertainty remains regarding whether these evaluations will ultimately translate into orders.
Nevertheless, the report itself carries significant symbolic weight. The report points out that some analysts believe the global surge in AI chip demand is pushing customers to seek second supply sources, a trend that significantly elevates the strategic position of Intel's foundry services.
For large tech companies long reliant on a single foundry source, the need for supply chain diversification is becoming increasingly urgent against the backdrop of current geopolitical and capacity competition.
In contrast to the evaluation phase for Apple and Google, the cooperation between Tesla and Intel has reached a confirmed state. According to a Reuters report, Elon Musk stated that Tesla plans to adopt Intel's next-generation 14A process to produce specialized chips for its Terafab AI complex in Austin, Texas.
This collaboration makes Tesla the first known external customer for Intel's 14A process, marking a significant milestone for the commercialization of Intel's foundry business. The report indicates that the 14A process is viewed as Intel's next technology node after 18A. Tesla's early commitment not only provides substantial order validation for Intel Foundry but also signals to the market the increasing credibility of the 14A technology roadmap.
The 18A process is a core pillar of Intel's foundry recovery strategy. According to TrendForce citing media reports, customers have already begun test chip validation on the 18A node, with core design flows largely complete, while development of the enhanced 18A-P process is progressing concurrently.
Intel CFO David Zinsner stated during the earnings call that yield improvements in the first quarter spanned multiple process nodes, including Intel 4, Intel 3, and 18A, directly contributing to the narrowing losses in the foundry segment. For internal high-volume manufacturing, Intel's flagship mobile processor, Panther Lake, and the Clearwater Forest server CPU both utilize the 18A process.
According to a prior Tom's Hardware report, Intel initiated low-volume production of Core Ultra 300 series "Panther Lake" CPU wafers at its Oregon development facility at the end of 2025, while simultaneously ramping up high-volume production at Fab 52 in Arizona.
Market observers note that better-than-expected progress on 18A yields and the readiness of the PDK (Process Design Kit) are gradually rebuilding external confidence in Intel's foundry business.
The potential recovery of Intel's foundry business is closely linked to structural shifts in AI computing architecture. Lip-Bu Tan pointed out on the earnings call that with the rise of agent AI, physical AI, and edge AI, the CPU is re-emerging as the scheduling and control core of AI systems. The convergence of the CPU-to-GPU ratio towards 1:1 signifies a substantial rebound in demand for server CPUs.
The strategic significance of this trend for Intel Foundry is that the full-blown explosion in AI chip demand is driving tech giants to actively seek diversified foundry options. Intel, with its technological advances in the 18A and 14A processes, is positioning itself as a credible alternative to TSMC.
Analysis suggests that beyond internal capacity absorption, onboarding external customers is the key path for Intel Foundry to achieve profitability at scale. Currently, from yield improvements to customer validation, and now with Tesla's formal agreement, the commercialization of Intel's foundry operations is progressing step by step along this path.
Comments