The Geopolitics of AI Infrastructure
summary
Huawei builds a Cloud Matrix 384 that achieves what Nvidia’s DGX H100 Ranger couldn’t: massive NVLink optics scaleup across 384 chips, enabled by TSMC wafers routed through a Sofco front and 13 million HBM stacks funneled via Samsung→Coasia→Faraday workarounds, so sanctions don’t stop volume. SMIC hits 7nm phone yields and scales toward AI chips. G42 secures 500,000 GPUs a year and is building multi-gigawatt campuses while the US faces a 63 GW shortfall in data‑center power.
AI-written summary · about
chapters (5)
related talks
27mLive Panel: Building at the Roofline — Planning and Delivering Gigawatt-Scale AI Infrastructure
31mHow to build a sovereign European AI compute infrastructure
24mClosing Keynote
12mAI-powered infrastructure:
24mSilicon Development
28mPowering Europe’s Future: Investment in AI and Quantum Infrastructure
01h 13mEngineering the Cloud for Tomorrow: Performance, Speed and Scale for AI Workloads
13mDemocratizing AI with Open Source Compute