Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-03-22 · source-backed
SemiAnalysis CEO details how long-context inference's KV Cache has exploded HBM demand, with memory requiring 4x the wafer area of DDR. NVIDIA locked up TSMC N3 allocation early — explaining why H100 pricing is higher today than three years ago despite more supply. Source
Each link below shares sources, entities, or timing with this story.
NVHBM, announced August 26, relocates NVIDIA's custom memory controller from the compute chip into the HBM base die, claiming up to 30% more bandwidth than standard HBM4E, 15% lower HBM power, and up to 25% more freed area on the XPU compute die (NVIDIA). Next-generation Train...
Amid a week of pricing and commerce stories, here's hard tech you can actually download. Google released DiffusionGemma on June 10, a 26B-parameter Mixture-of-Experts model (3.8B active) that generates text by diffusion instead of left-to-right decoding. The architecture is th...
NVIDIA launched Nemotron 3 Super — a 120B total / 12B active parameter hybrid Mamba-Transformer MoE, open, designed specifically for multi-agent workloads, and delivering 5x higher throughput than Nemotron 2 at the same active parameter count (NVIDIA Newsroom). It ships with a...
NVIDIA is building a new inference processor integrating Groq's Language Processing Unit technology (acquired December 2025). The chip uses on-chip SRAM instead of HBM, delivering up to 80 TB/s memory bandwidth (~10x H100). OpenAI committed to 3 GW of dedicated inference capac...
On Dwarkesh Patel's August 25 episode, the SemiAnalysis founder argues the two labs are consolidating capacity at an accelerating rate and will control the majority within about two years, on the basis that they monetize inference better than anyone else can. Specific, dated a...
Bloomberg reported August 22 that Nvidia notified its largest customers that servers with its AI chips will cost more than 15% more in many cases, driven by memory prices, effective on systems shipped early next year and covering flagship Vera Rubin and Grace Blackwell configu...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.