Intel hamstrung by supply shortages across its business (www.tomshardware.com)

Intel says rising demand for client and data-center processors in Q3 2025 was throttled by broad supply shortages — limited internal fab capacity on Intel 10 and Intel 7 nodes plus an industry-wide substrate shortfall — and the company will prioritize high-margin data‑center CPUs, raise prices on older client parts, and expect shortages to persist into 2026. CFO David Zinsner confirmed Intel is diverting wafer capacity to server SKUs (including Xeon 6 “Granite Rapids,” whose I/O die is made on Intel 7) at the expense of client volume, which has already tightened supply and pushed up prices for Raptor Lake (13th/14th‑gen) processors built on Intel 7. For the AI/ML community this matters: hyperscalers and AI infrastructure buyers will be prioritized — helping short‑term availability for large cloud providers — but constrained Intel 7 capacity and substrate bottlenecks raise cloud and on‑prem procurement costs, extend lead times for GPU/CPU server builds, and could slow rollout of training and inference clusters. Technical choke points include Intel 7 capacity (used for high‑end client cores, I/O dies and Xeon parts), continued demand for mature nodes, and a slow 18A ramp; Intel expects shortages to peak in Q1 before easing, but 2026 supply remains tight, likely keeping prices elevated and influencing vendor and procurement strategies.