技术博客与 newsletter feed · SemiAnalysis · Newsletter

Is NVIDIA Over? Export Controls, RASA, Open Source Implications, Supply Chain Implications, Affair, US Still Dependent On China

A wind-powered 800MW campus, 120MW built within a year, a 5MW SST-native 800V DC pilot and a cooling design without central chillers

Viable Alternative of CUDA Moat, NVIDIA scared and working nights & weekends, Jensen’s Counter Submission, Kimi K3, Muse Spark Open, Qwen, GLM5.3, TorchTPU, vLLM, SGLang, llm-d, Ray

Making 13GW in 1.5 Years A Reality, Repurposing Rocket Ship Campuses, A Datacenter at TeraFab?

A frontier benchmark spanning RSI, hardware design, cyber, bio, finance, legal, quant trading, chip design, software engineering, materials science, and alignment

2027 is largely locked in. 2028 is less secure: a prolonged Texas permit freeze is now the biggest single risk to the US buildout, and both parties are running on tighter rules

Trainium Best for Inference? But now good for Training too, Native Trainium PyTorch, High Quality CI, TorchTitan, GPU Impossible Profiler

Huawei Ascend 950PR Deep dive & µbenchmarks, DSv4.1 Flash Analysis, DeepSeek support for Ascend

60GW Under Construction, SemiAnalysis Leading Indicators, 1.3T of Lease Contracts Signed YTD, The Shortage Could Move to Chips

ASML is building a hybrid bonding tool! The company has not announced anything yet, but its patent filings, a news report and its own roadmap all suggest so.

AI safety is on fire.

Limit testing every AI subscription plan from Anthropic, OpenAI, Meta, SpaceXSI, MiniMax, Moonshot, Z.ai, Cursor, and Cognition

GLM-5.3, KV Cache Offloading, HiSparse, DeepSeek Sparse Attention, IndexShare, Single-rollout Asynchronous Optimization, Cybersecurity

Looking inside Intel 18A and Intel’s Panther Lake. A free SemiAnalysis STEEL teardown.

1,000+ facilities across 60+ operators mapped, built retail-first and flipped by AI, largest hyperscaler leases 1/5 national capacity, 100MW in 12 months, Eastern Data Western Compute

In gory detail: reliability, performance, support, pricing—and, of course, security—in our most thorough analysis of GPU cloud providers globally.

Mapping MoE models onto inference hardware: structure, flow, and efficient serving

New Model Architecture Implications for TAM of DRAM/NVMe, DeepSeek V4.1 Flash, AgentX, InferenceX, NVMe experiments

20GW sits inside a restricted local boundary, 1,525MW actually slips, 2.3GW nationwide including New York

Jensen Sandbagging Performance Again, 2x more Annual Profit Per GigaWatt, The More you Buy, The More you Earn, AgentX, InferenceX, Extreme Co-Design