Intel starts production of advanced 18A-P chip node

Intel has begun risk production of its 18A-P process node, an optimized version of the 18A technology that already entered high-volume manufacturing earlier this year. The move marks another step in Intel’s aggressive “five nodes in four years” strategy to reclaim its position at the cutting edge of chipmaking.

Intel’s 18A-P delivers over 9% higher performance at the same power levels, or alternatively, 18% lower power consumption at equivalent performance compared to the base 18A node.

What 18A-P actually brings to the table

The process introduces new RibbonFET transistor technology and enhanced logic VT pairs, offering designers more flexibility in balancing speed against power consumption.

The 18A-P delivers up to a 50% increase in thermal conductivity and a 20-40% reduction in thermal resistivity. Via resistance drops by 10-30%. Front-end frequency at iso leakage improves by roughly 12%.

Intel unveiled these enhancements at the 2026 VLSI Symposium, with Synopsys and other leading EDA partners confirming full design tool support.

The foundry play

The base 18A process reached high-volume production at Intel’s Fab 52 in Arizona and its Oregon facilities, leading to the launch of Panther Lake chips showcased at CES 2026.

Primary fabrication for 18A-P will run at Intel’s D1X Mod 3 facility in Oregon and the Fab 52/62 complex in Arizona. Customer deliveries are targeted for 2026, following the risk production phase currently underway.

Rumors of Apple showing interest in Intel’s improved capabilities suggest the pitch might be landing. Apple, which famously ditched Intel’s processors for its own Arm-based designs starting in 2020, would be a symbolically powerful foundry customer if it ever signed on for manufacturing.

What this means for investors

The power-performance improvements are particularly relevant for AI and high-performance computing applications. An 18% reduction in power consumption at equivalent performance translates directly to lower operating costs for data center operators running thousands of processors simultaneously.

Government incentives under programs like the CHIPS Act were designed to support domestic manufacturing of advanced chips at facilities like Intel’s Arizona and Oregon fabs.

The risk for investors is execution. Risk production is not the same as high-volume manufacturing. If major design houses start committing tape-outs to 18A-P, that’s the inflection point.