Applied Materials: agentic AI chips to hit 7,000+mm² per package, 70x mobile silicon area
BenBajarin · x · 2026-10-13
Sharing a slide from an Applied Materials event, Ben Bajarin compares three chip classes: mobile (A20 Pro, 2nm, 2026) at 98mm² logic silicon, 25B transistors, 17 Cu wiring layers; AI training (MI455X, 2/3nm, 3.5D) at 3,470mm², 320B transistors, 45+ layers; and estimated agentic chips (XPU, 3.5D, 2028E) at 7,000+mm² (70x), 500B+ transistors (20x), 55+ layers (3x). His take: mobile was boring for semis — pushing reticle limits is cooler. $AMAT
Related event: Applied Materials: Agentic AI racks consume 2-4x more wafers(2 posts)→
More from Infra
- Cooler Master launches 300 kW liquid cooling CDU, up to 450 kW, for GB200/GB300 racks — BenBajarin · 2026-10-14
- MSAR brings multi-head depth attention to residuals, beating AttnRes and mHC with 1.58x iso-loss speedup — SonglinYang4 · 2026-10-14
- Cloudflare's pipeline handles 2B events/sec, five years of optimization deferred 2.5 years of hardware growth — irvinebroque · 2026-10-14
- South Korea to deploy 74 portable Starlink systems for forestry by end of 2026 — XFreeze · 2026-10-14
- ThunderSyncRL speeds agentic RL up to 1.9x with zero policy staleness, Stanford paper claims — heghbalz · 2026-10-14
- Agent task costs fall ~90% every 6 months, countering Bloomberg's data center cost alarm — sudoraohacker · 2026-10-14