Mature-Node Pricing Splits: 90nm Rises Fastest, 8-Inch Tight
This round of foundry price pressure is not spread evenly across nodes. In Q4 2026, utilization at mature-node fabs is running high, but pricing moves at different speeds by process node and wafer size: older nodes pulled by AI-adjacent demand are rising fastest, mid-range 28/40nm nodes are still waiting for demand to recover, and the 8-inch line is tighter than 12-inch as power-management and discrete devices flow back onto it. Reading the cycle means identifying exactly which node is tight and who is driving the demand.
High utilization, but pricing splits by node
Sigmaintell's October 19, 2026 forecast puts average utilization at major mature-node foundries at about 91% in Q4 2026; UMC and peers are expected to stay above 91%, while mainland fabs such as SMIC, Hua Hong and Nexchip have been at full capacity since earlier in the year. With energy and material costs rising and demand recovering, the agency expects 12-inch and 8-inch mature-node foundry prices to rise broadly in Q1 2027, up roughly 4%–8% quarter over quarter.
Figures above are agency-forecast foundry quotes (USD per wafer), not retail prices; source Sigmaintell, 2026-10-19.
Why 8-inch is tighter than 12-inch
Within mature nodes, the 8-inch platform is the tighter market. TrendForce data show average utilization at the top ten foundries' 8-inch lines recovered to 88% in 2026, reaching about 90% in the second half. PMICs and power discrete devices have long depended on 8-inch lines, while TSMC, Samsung and others keep shrinking or moving part of their 8-inch capacity; industry estimates put global 8-inch capacity down about 6% in 2026 and another 2% in 2027.
The demand increment comes mainly from AI infrastructure. The shift to 48V power architectures in AI servers is lifting demand for power-management and discrete devices, with order visibility generally around six months; 55nm-and-above nodes, silicon interposers and some FPGA-related demand are also occupying mature-node capacity. Supply shrinking while demand returns is the direct reason 8-inch moved into tight supply first.
Implications for procurement and design
PMIC / discretes / legacy-node parts
- Check lead times and alternate sources for parts tied to 8-inch lines early
- Lock price and capacity for long-cycle projects before tape-out discussions
- Qualify a second source for critical parts to absorb single-line swings
New designs / solution evaluation
- Evaluate multiple process options in new schemes rather than binding to one node
- Separate mature-node from leading-edge pricing; don't rely on a blanket "foundry up" narrative
- Include quarterly foundry quote movement in BOM reviews
Overall, this mature-node cycle is structural: AI power and adjacent demand underpin older nodes and 8-inch capacity, while mid-range nodes still wait for recovery in communications, display drivers and other applications. Going forward, it matters more to track whether 8-inch utilization holds and how new capacity lands in 2027, rather than reading a single blanket price-hike list.

Comments
No comments yet. Be the first to comment.