After years of fumbles and PR gloss, the move was revealed at the company’s latest Intel Vision 2025 event.

The 18A process is a do-or-die moment for Intel Foundry Services, which has stumbled more than strode in recent years. If this node flops, there may not be many curtain calls left. But for now, the company’s making noises about progress and claims it’s ready to scale once the trial wafers behave.

Risk production is essentially a form of chip roulette—limited runs are used to identify yield issues and manufacturing quirks before committing to full-scale production.

 If 18A holds up, Chipzilla plans to unleash it en masse by late 2025, with the first real-world guinea pigs being the Panther Lake SoCs due in 2H 2025. Other adopters might jump in by 2026, assuming the process doesn’t go off the rails again.

This isn’t just another die-shrink, either. The 18A node introduces BSPDN (Backside Power Delivery Network), a design twist that shifts power delivery to the rear of the wafer, freeing up routing space on the front.

Intel is boasting a macro bit density of 38.1 Mb/mm² on its high-density variants, although how that translates into real-world performance and efficiency remains to be seen.

Still, Chipzilla is betting the foundry farm on 18A, with hopes it can claw back relevance from TSMC and Samsung before Nova Lake hits in 2026. Whether it’s a comeback or just another stumble depends on what happens next in the silicon trenches.