brain/
sourcestock-market

2026 08 26 Autoresearch Intel 18A Yield NO NEW Primary

Nothing new to graduate the open question. Recycled July KeyBanc ~85% vs TSMC N2 ~90% remains modeled. Intel 10-Q still 'higher-cost' 18A wafers. Industry-standard yields still 2027. Do not re-rate INTC.

Source

Summary

No new Intel-disclosed or teardown yield number since the 08-21 VLSI follow-up. Trade-press hits on 2026-08-26 still recirculate the July 15 KeyBanc ~85% (from ~65%) vs TSMC N2 ~90% model — the same modeled channel already attached on intel-18a-yield-vs-tsmc / intel. Counterpoint's Fab 52 visit confirms Panther Lake mass production in Arizona and Oregon and 18A-P PDK 1.0 to external customers; that is volume, not a yield bake-off vs N2.

IFS Q2 / 10-Q language remains "higher-cost" 18A wafers. Industry-standard yields still targeted 2027. Do not re-rate INTC. Question stays open.

Findings

  • KeyBanc 65%→85% in a quarter: already ingested 08-21 as modeled, not a teardown. Recycled this week (RCR Wireless, TechTimes 07-15, TweakTown Nova Lake-S in-house shift).
  • ASML High NA on a Panther Lake subset of layers, NXE-matched yields: already attached. Not a full-node High NA conversion.
  • Counterpoint Fab 52 visit: Panther Lake HVM in AZ + Oregon; 18A-P PDK 1.0 delivered late 2025; 14A "significantly ahead of 18A at a similar point"; capacity only for committed customers.
  • No new abstract, 8-K, or 10-Q yield % overnight.

Sources fetched

What this does not change

intel-18a-yield-vs-tsmc stays open. us-fab-capacity-bottleneck / tsmc-saturation-to-intel-anchor-stack not re-dated on yield.

Referenced by