SEMICON Taiwan 2026: AI Chips Shift From Node Scaling to System Scaling With Six Battlegrounds
pstAsiatech · x · 2026-09-08
Based on SEMICON Taiwan 2026 forums and exhibits, SemiVision argues the industry's key signal is an architectural shift: AI scaling is moving beyond transistor scaling toward system scaling.
- A high-end AI accelerator is now a heterogeneous system of logic dies, HBM, I/O dies, SerDes, optical engines, advanced packaging, power delivery, and cooling
- The next phase spans six interconnected dimensions: compute, memory, packaging, optical I/O, thermal, and test scaling — all accelerating simultaneously
- Advanced packaging, CPO, HBM, FOPLP, and testing emerge as the next key battlegrounds
More from Infra
- Both Nvidia and Google run ARC synthetic data generators — GregKamradt · 2026-09-08
- Memory capex to hit $97B in 2026, $146B in 2027 — 56% of next year's semiconductor spend — Beth_Kindig · 2026-09-08
- TokenSpeed stabilizes Kimi K3 on ~1,000 B300 GPUs for 3+ weeks: GB300 TP8 optimization details — zhyncs42 · 2026-09-08
- KV Cache Math From Scratch: A Deep Workshop on LLM Inference at Scale — AI Engineer · 2026-09-08
- Intel extends High NA EUV lead as TSMC and Samsung confirm adoption by end of decade — BenBajarin · 2026-09-08
- User seeks a custom GGUF quant to fit GLM on a 192 GB RAM Mac between Q2 and Q4 — CentrifugalMalaise · 2026-09-08