Long Video Notes: chapter-by-chapter notes, timestamps, key frames and plain-language explanations for keynotes and long explainer videos, written by people who watched the whole thing.https://longvideonotes.com/https://longvideonotes.com/disaggregated-inference-explained/https://longvideonotes.com/disaggregated-inference-explained/What disaggregated inference is, why NVIDIA splits prefill, attention and FFN across Vera Rubin GPUs and Groq LPX, and what it means for latency and cost.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/gtc-2026/https://longvideonotes.com/gtc-2026/When and where the GTC 2026 keynote ran, what each segment covers with clickable timestamps, and which of our notes to read first.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/gtc-2026-dlss-5-cuda-20-years/https://longvideonotes.com/gtc-2026-dlss-5-cuda-20-years/Why Jensen Huang opened GTC 2026 with CUDA's 20th anniversary, how the install-base flywheel works, and what DLSS 5 neural rendering actually changes.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/gtc-2026-groq-lpx-disaggregated-inference/https://longvideonotes.com/gtc-2026-groq-lpx-disaggregated-inference/How NVIDIA pairs Groq 3 LPX with Vera Rubin through Dynamo at GTC 2026: token price tiers, the throughput curve, the 35x claim and when LPX ships.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/gtc-2026-keynote-notes/https://longvideonotes.com/gtc-2026-keynote-notes/Chapter-by-chapter notes on Jensen Huang's GTC 2026 keynote: what each segment announced, what the numbers mean, and which parts you can skip.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/gtc-2026-nemoclaw-openclaw-agentic-os/https://longvideonotes.com/gtc-2026-nemoclaw-openclaw-agentic-os/What Jensen Huang said about OpenClaw at GTC 2026, how NVIDIA's NemoClaw and OpenShell add policy, network and privacy guardrails, and who owns what.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/gtc-2026-nemotron-open-models-coalition/https://longvideonotes.com/gtc-2026-nemotron-open-models-coalition/Who joined the Nemotron Coalition at GTC 2026, what the six NVIDIA open model families cover, and why NVIDIA keeps funding open models it gives away.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/gtc-2026-physical-ai-robotaxi-robotics/https://longvideonotes.com/gtc-2026-physical-ai-robotaxi-robotics/What NVIDIA announced for robotaxis and robots at GTC 2026: new Hyperion automakers, the Uber rollout, Isaac Lab 3.0, Newton 1.0, Cosmos 3 and GR00T.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/gtc-2026-roadmap-rubin-ultra-feynman/https://longvideonotes.com/gtc-2026-roadmap-rubin-ultra-feynman/NVIDIA's GTC 2026 roadmap explained: Vera Rubin in production, Rubin Ultra and LP35, Feynman with LP40 and Rosa, copper vs CPO, DSX AI factories and Space-1.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/gtc-2026-token-economics-inference-inflection/https://longvideonotes.com/gtc-2026-token-economics-inference-inflection/Why Jensen says inference hit an inflection at GTC 2026: three AI shifts, the $1 trillion order claim, tokens per watt, SemiAnalysis data, Fireworks' 7x.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/gtc-2026-vera-rubin-platform/https://longvideonotes.com/gtc-2026-vera-rubin-platform/What NVIDIA showed of Vera Rubin at GTC 2026: the NVL72 rack, Vera CPU, NVLink 6, CPO Spectrum-X, BlueField-4 STX and Rubin Ultra, with timestamps.Thu, 01 Oct 2026 00:00:00 GMThttps://longvideonotes.com/tokens-per-watt-explained/https://longvideonotes.com/tokens-per-watt-explained/What tokens per watt and tokens per second per megawatt measure, why NVIDIA made them its headline metric at GTC 2026, and how to read vendor claims.Thu, 01 Oct 2026 00:00:00 GMT