Intel’s 18A-P Wakes Up: Faster Chips, Less Consumed Power, New Buyers
Intel said Tuesday that the newest member of its 18A manufacturing family, called 18A-P, has entered risk production — the early, small-batch stage chipmakers use to prove out a process before full volume. The company tied the milestone to surging demand for its central processors and a renewed push to win outside foundry customers.
Key Takeaways
- 18A-P delivers 9% higher performance at the same power level (iso-power) or 18% lower power at the same speed (iso-performance) versus Intel 18A, with better thermals and more design flexibility.
- The process is fully design-rule-compatible with 18A, so customers can reuse existing intellectual property and design flows without redrawing their chips.
- Intel now pitches 18A as a product for external clients and forecasts second-quarter revenue of $13.8 billion to $14.8 billion, above the $13.07 billion analysts expected.
Risk production is a critical stage in the semiconductor manufacturing process. It occurs when a foundry begins to produce silicon wafers for a specific chip design, but the process is not yet fully validated.
A Faster, Leaner 18A
The upgrade is incremental but measurable. Against the base 18A node, 18A-P gives chip designers 9% more performance at the same power draw, or 18% lower power at the same clock speed. Intel also points to improved thermals and added design flexibility.
Just as important, 18A-P stays design-rule-compatible with 18A. Customers can carry over their existing IP and design flows instead of starting from scratch — a practical detail that lowers the cost and risk of adopting the newer process.
Intel announced the step at the VLSI Symposium, against a roadmap it set with partners a year ago. “This is a journey, and while we have more work ahead, we appreciate the opportunity to share the progress we are making with Intel 18A-P and our longer-range R&D,” said Naga Chandrasekaran, who runs Intel Foundry. The 18A-P family builds on the same generation behind Intel’s Panther Lake laptop chips, which began rolling out of its Arizona fab late last year.
Courting Outside Customers
The on-schedule step also feeds Intel’s foundry ambitions. By moving 18A-P into early production when it said it would, Intel tells prospective clients it can deliver what it promises — a sticking point for any company trying to make chips for others.
CEO Lip-Bu Tan has started describing 18A as something Intel can sell to external clients, a reversal of his earlier stance that the process would pay off only through Intel’s own products. Finance chief David Zinsner laid out that change in March.
Intel has been lining up names to back the pitch. Cloud providers such as Google have committed to its server processors across multiple generations, and outside chipmakers have agreed to build custom AI silicon on Intel’s process, with Amazon among the first to say so publicly.
AI Demand Powers the Comeback
The fresh manufacturing push lands as Intel’s processor business finally catches a tailwind. Demand from companies running AI services ran so hot in the first quarter that Intel sold chips it had already written off — old or underperforming inventory it had shelved.
That strength shows up in the guidance. Intel expects second-quarter revenue between $13.8 billion and $14.8 billion, well above the $13.07 billion analysts had penciled in, according to data compiled by LSEG.
Behind the numbers sits a change in what AI workloads need. As companies move from training models to running them — and toward autonomous AI agents — central processors are pulling more weight alongside the GPUs that powered the first wave. “The CPU is reinserting itself as the indispensable foundation of the AI era,” Tan told investors on Intel’s earnings call.
What It Means for the Chip Race
Beyond Intel’s own books, 18A-P pushes the company deeper into a contest for leading-edge manufacturing that TSMC has owned for years. Analysts read the timely milestone as evidence Intel can challenge TSMC’s most advanced nodes, and the progress inches Intel closer to a possible deal to make chips for Apple devices — the kind of marquee customer its foundry has lacked. Rival TSMC, meanwhile, has been turning out advanced chips on U.S. soil in Arizona, keeping the pressure on.
The stakes reach past corporate rivalry. Advanced chipmaking now underpins everything from AI data centers to national security, and governments have poured subsidies into bringing production home. Because 18A-P works as a drop-in upgrade that needs no redesign, Intel can court customers without forcing them to rebuild their chips — a low-friction path that could speed adoption if yields hold up. For a company that spent years losing ground, getting a competitive process into customers’ hands on time is the proof point that counts.
Written by Alius Noreika
