Buzzberg Cup Live

Ep. 019 - Inside the STEEL Lab: From Package to Transistor (Teardown Lab)

Watch on YouTube ↗  |  July 16, 2026 at 21:00  |  36:07  |  SemiAnalysis
Speakers
Andrew Wagner — Reporter, The Block
Afzal Ahmad — Analyst, SemiAnalysis

Summary

The STEEL teardown lab team explains how they analyze chips from package to transistor, using the SMIC N+3 process and Huawei Kirin 9030 SOC as a case study. They detail die shots, NPU discovery, cell height shrink, and standard cell libraries, while discussing how export bans forced SMIC to innovate without EUV. The conversation also covers future teardown challenges like backside power, gate-all-around, and advanced packaging.

  • The STEEL lab provides competitive intelligence by tearing down chips to the transistor level.
  • SMIC's N+3 node shrank the M0 layer over 15% and reduced SRAM area 10-20% without EUV.
  • Huawei's Kirin 9030 features an upgraded NPU with two tiny cores, discovered through die shot annotation.
  • Export restrictions forced SMIC to innovate, reaching parity with leading fabs but making process compromises.
  • Cell height reduction from N+2 to N+3 improves transistor density; library choices affect performance and area.
  • Nvidia's A100 used a 4-fin HPC library, showing how specialized libraries optimize high-current GPUs.
  • Backside power, gate-all-around, and hybrid bonding create new teardown challenges and analytical opportunities.
  • Huawei is using logic folding with sub-1.5-micron pitch hybrid bonds to circumvent DUV scaling limits.
Ideas
Andrew Wagner Reporter, The Block 16:14
SMIC scales aggressively despite EUV restrictions
SMIC's N+3 process achieves extremely aggressive scaling without EUV, reaching parity with the world's leading fabs in some aspects while managing yield and performance through innovative process decisions. Export restrictions have forced SMIC to innovate, creating progress despite being unable to use EUV.
Up Next

This SemiAnalysis video, published July 16, 2026, features Andrew Wagner discussing 0981.HK. 1 trade idea extracted by AI with direction and confidence scoring.

Speakers: Andrew Wagner  · Tickers: 0981.HK