SemiAnalysis: Vera Rubin NVL144 hits ~7x tokens per MW vs Blackwell, over 2x profit per GW
sudoraohacker · x · 2026-09-16
Quoting SemiAnalysis, the post argues Jensen Huang "sandbagged" at GTC 2026: his chart showed VR NVL72 at only 3x performance per MW vs Blackwell for 1-3T parameter models around 200 TPS.
- Real-world pre-release software already shows up to 7x better token throughput per megawatt on Rubin.
- At a 170 tokens/s SLO, Vera Rubin NVL72 delivers 67x GB300 throughput per dollar.
- Estimates suggest Rubin earns over 2x more profit per gigawatt than Blackwell even on early software builds; Ian Buck will present real AgentX numbers at the AI Infra Summit in Santa Clara.
Related event: SemiAnalysis Finds Rubin NVL72 Far Exceeds Nvidia's Claimed Gains(3 posts)→
More from Infra
- Tailscale called a 'single point of failure' for startups as acquisition chatter splits devs — TheZachMueller · 2026-11-02
- Gary Marcus: AI compute is headed for a glut, just like fiber in the dot-com bust — GaryMarcus · 2026-11-02
- One H200 Isn't Enough for DeepSeek 4.1 Flash: Two GPUs at $10.59/hr — daluoseo · 2026-11-02
- Databases to distributed systems: every systems skill is merging into AI inference and hardware — Abhishekcur · 2026-11-02
- AWS not "sold out" until 2027: expert says capacity headline needs context — DavidLinthicum · 2026-11-02
- Foxconn chairman Liu: AI is not a bubble, demand for AI products remains clear — pstAsiatech · 2026-11-02