SemiAnalysis Weekly

Dan, a SemiAnalysis contributor who reports conference takeaways and explains CPO and optical interconnects. Ivan Chiam, an analyst modeling AI compute demand and memory dynamics. Sravan Kundojjala, a foundry and memory market analyst tracking TSMC allocation. They map the AI silicon shortage, TSMC capacity choices, memory/HBM pressure, GPU rental pricing, and Nvidia's co‑packaged optics roadmap in short, punchy conversations.

Joey Brookhart, SaaS and market analyst who tracks AI adoption and monetization. Malcolm Splitter, economist focused on macro and labor-market measurement. Ray Wang, memory-market specialist modeling DRAM/HBM supply and pricing. They discuss memory shortages and hyperscaler capex, liken memory to oil in scarcity dynamics, explore enterprise-driven token demand and regional pricing, and debate how AI breaks GDP and macro models.

Jeremie Eliahou Ontiveros, Head of Datacenter & Energy Infrastructure Research, explains how AI data centers reshape electricity markets and grid reliability. They cover market design differences like PJM versus ERCOT, thermal accreditation and coal retirements, behind-the-meter deals and customized tariffs, and how hyperscaler commitments, financing mismatches, and demand signals drive grid investment.

Copper Wei (Wega), an accelerator and AI supply‑chain expert, walks through Rubin’s system‑level design, PCB and power delivery, and memory procurement. The conversation covers HBM4 bandwidth and supplier tradeoffs. They discuss NVLink 6, cable‑less PCB trays, thermal innovations like microchannel lids, chiller‑less datacenters, and Rubin’s power and deployment timelines.

Cam Quilici, an AI/ML systems practitioner focused on inference performance and benchmarking. He discusses InferenceX’s evolution from its predecessor and the move to multi-node DeepSeq. Conversation covers major hardware wins, nightly large-scale benchmarking, software tuning complexities, multi-token prediction benefits, cost/TCO modeling, and roadmaps for TPU, multimodal, and multi-turn benchmarks.

A deep dive into Claude Code adoption and how the team tracked its rise. A resurgence of CPUs driven by RL, agent orchestration, and infrastructure demands. A deep look at historic memory tightness, HBM vs logic tradeoffs, and when suppliers might add capacity. Discussion of fast-model economics, Chinese video model advances, and how generated video could explode storage demand.