Intel Reallocates 18A Capacity to Xeon as Server Demand Outruns Supply

By Priya Raman

Intel's Data Center and AI business generated $7.3 billion in the third quarter of 2026, up 17% sequentially and nearly 80% year-over-year. The company now plans to convert manufacturing resources in Oregon from Intel 18A to Intel 3, sacrificing client processor output for server chips. The consumer PC market is shrinking and will continue to shrink, according to the company's own outlook.

What did Intel actually announce?

Four capacity moves dominate the plan:

• Expanded production of Intel 3-based Xeon 6 'Granite Rapids' processors at Fab 24 near Leixlip, Ireland, online only by the end of 2027 unless accelerated.
• Conversion of Oregon resources from Intel 18A, originally intended for 'Panther Lake' client processors, to Intel 3, entering production in the second half of 2027.
• Accelerated equipment installation at Fab 62 in Arizona, the second high-volume facility for 18A, 18A-P and derivative processes, alongside ramping 18A at Fab 52.
• Expanded advanced packaging operations in New Mexico and Malaysia.

"What we are taking offline to put Intel 3 into Oregon is 18A wafers, which we are generally going to build Panther Lake products," said Dave Zinsner, chief financial officer. "We see the opportunity with some of that capacity freed up to go produce more data center wafers."

Intel has not disclosed the precise client-versus-data-center capacity split at the Arizona fabs. That omission matters. The company simultaneously claims it expects to win client market share with Core Ultra 400-series 'Nova Lake' CPUs, which requires high-volume production.

How severe is the shortage?

Chief executive Lip-Bu Tan framed the situation starkly on the earnings call. "The industry continues to navigate one of the most significant supply shortages in history, driven by the unprecedented demand for Super Intelligence," Tan said. "Against this backdrop, I am pleased by the progress we made growing our server supply in Q3. I am mindful about the ongoing challenges we face in the future quarters to keep pace with server demand."

Granite Rapids is Intel's fastest-ramping server processor family, yet demand still exceeds supply. Management cites yield improvements and shorter production cycles as mitigations. Intel says Intel 3 manufacturing costs fell by approximately 15% since the first quarter of 2026.

Intel expects data center CPU demand to stay strong through 2027-2028, driven by CPU-intensive agentic and inference workloads. Expectation is not shipment; capacity arrives late.

Why are substrates the bottleneck?

Wafers alone do not solve the problem. Intel also faces shortages of advanced organic substrates needed to package multi-chiplet processors, or system-in-packages, that rely on EMIB and Foveros 3D stacking technologies.

"We are not only constrained on the wafer side, we are also constrained on the substrate side," Zinsner said. "So, it is important we get the substrate piece of this ramped up to be able to meet the demand."

Intel expects approximately 40% additional substrate capacity in the second half of 2027. Substrate suppliers must themselves procure and install new tools first. Clearwater Forest Xeons already combine EMIB with Foveros Direct 3D and hybrid bonding, stacking Intel 18A compute dies on Intel 3 base dies. On packaging, Tan reported "significant external customer traction" for EMIB and EMIB-T and "robust progress" on Hybrid Bonding Interconnect for 3D chip stacking in upcoming server products.

What will it cost?

Capital expenditure is set to reach approximately $20 billion in 2026 and rise significantly in 2027, though Intel gave no firm figure for the later year. Firm purchase orders to equipment suppliers rose nearly 40% sequentially. "We have made good progress securing the equipment required to support our growth plans," Zinsner said. Intel is also exploring acceleration of its Ohio complex, where the first major structural truss was recently installed.

When does supply catch up?

Not soon, by Intel's own account. Equipment must be installed and qualified, wafers must pass lengthy cycles, and substrates must scale before finished processors ship. Intel expects supply constraints to remain particularly severe during the first half of 2027, even as the client PC market contracts. Only in the second half of 2027 does the company expect additional capacity, higher yields and better substrate availability to let it capitalize on AI infrastructure demand. Every promise in this plan carries a date at least a year out.