Intel begins 18A-P chip production for foundry clients

Intel has begun limited production of its enhanced 18A-P process node, marking a step forward in its foundry ambitions. The company disclosed the development at the 2026 VLSI Symposium in Hawaii, revealing that 18A-P delivers a 9% performance boost at the same power level as the base 18A node—or 18% lower power consumption for equivalent performance. The improvements stem from optimizations in transistor design, interconnects, and co-optimized manufacturing techniques, while maintaining full design rule compatibility with 18A. This compatibility allows customers to transition existing 18A designs to 18A-P with minimal adjustments, a factor that may lead some to skip the base node entirely.

The 18A-P node is the first of several planned enhancements to Intel’s 18A process, which debuted commercially in January 2026 with the "Panther Lake" Core Ultra Series 3 processors. Intel’s foundry business, Intel Foundry, had initially planned to offer the more advanced 14A node as its first mainstream commercial process, but CEO Lip-Bu Tan reportedly shifted focus to 18A and its variants. Industry speculation suggests Apple may be among the first customers to adopt 18A or 18A-P for some of its silicon, though neither company has confirmed this.

Performance and compatibility

Intel’s 18A-P node is designed to address the needs of high-performance computing and AI workloads, where power efficiency and transistor density are critical. The company claims the enhancements are achieved without requiring redesigns from customers, thanks to the node’s backward compatibility with 18A. This could accelerate adoption, particularly among foundry clients looking to leverage Intel’s manufacturing capabilities without overhauling their existing designs.

Alongside 18A-P, Intel is developing 18A-PT, a variant optimized for through-silicon vias (TSVs). This technology enables chiplet stacking, allowing memory and logic components to be integrated vertically. Analysts believe 18A-PT will appeal to AI accelerator designers, who often require high-bandwidth memory solutions. Intel’s roadmap also includes longer-term innovations like CFET (Complementary FET), which stacks NMOS and PMOS transistors vertically to increase density, and the integration of gallium nitride power devices with silicon logic for mixed-signal applications.

Challenges and outlook

Intel’s progress with 18A-P comes after delays in the base 18A node’s production timeline. In a recent conference, Intel CFO David Zinsner acknowledged the company had "tried to do too much at once" with 18A, balancing performance improvements and yield optimization simultaneously. The shift to 18A-P suggests Intel is prioritizing incremental enhancements over a single, high-risk leap to 14A, at least in the near term.

For foundry customers, the availability of 18A-P could provide a more stable and predictable path to advanced manufacturing. Intel’s commitment to long-term process innovation was underscored by Naga Chandrasekaran, EVP of Intel Foundry, who stated at the VLSI Symposium: "Our updates and presentations at VLSI signal to Intel Foundry customers and partners that we are fully committed to leading edge process innovation over the long term."

The next 12 months will be critical for Intel as it seeks to convert risk production into commercial volumes. Success with 18A-P could help the company regain momentum in the foundry market, where it faces stiff competition from TSMC and Samsung. Meanwhile, the development of 18A-PT and other advanced nodes will determine whether Intel can meet the evolving demands of AI and high-performance computing customers.