In-depth guides to the AI chip supply chain, semiconductor chokepoints, GPU export controls, HBM memory, and China's domestic AI stack.
How raw silicon becomes a deployed GPU — the 17-tier supply chain, its chokepoints, ongoing wars reshaping it, and what a single disruption costs humanity's most consequential technology project.
Inside China's $150 billion response to chip export controls — Huawei's Ascend platform, SMIC's DUV workaround, DeepSeek's efficiency revolution, and what the domestic Chinese AI stack means for US, Japanese, Korean, and Taiwanese companies.
A plain-language guide to US and China export controls in the AI chip supply chain — which chips are restricted, which materials China controls, how Japan and Korea navigate multi-jurisdictional compliance, and what comes next.
Why high-bandwidth memory has become AI's most constrained resource — who makes it, why so few can, how the Iran war compounded Korea's supply chain risk, and when supply might catch up to demand.
A deep dive into the critical single points of failure in the semiconductor supply chain — why they exist, why they persist, and why 2026 exposed more of them than any year since COVID.
Kimi K3's 2.8-trillion-parameter release and a White House allegation about GB300 chips reaching Moonshot AI via Thailand reopened the question of how China trains frontier models under export controls — the Nvidia workarounds, the domestic Ascend/SMIC/EDA stack, and what a 2026 wave of Chinese AI-chip IPOs signals about the trajectory.