Stanford study: NVIDIA B200 delivers 4x the dollar value per unit of energy vs AMD MI350X
rohanpaul_ai · x · 2026-11-05
Stanford researchers ran the same model (Qwen3.6-27B) on 4 different single AI chips across 220 real professional tasks, measuring useful work produced per unit of energy.
Key finding: the NVIDIA B200 delivered over 4x the dollar value and over 9x the expert time saved per unit of energy compared with the AMD MI350X. In other words, with the model unchanged, the chip alone can decide how much economic return you get from your power budget — chip choice is itself an economics variable.
More from Infra
- Snowflake adds Business Ontology and Cortex AI Gateway governance for enterprise agents — shashib · 2026-11-05
- Ricursive debuts from-scratch Ripple-65B: open-weight agentic coding SOTA at 1/10 cost — HazyResearch · 2026-11-05
- LexiSuite: an all-Rust local LLM stack with training, LoRA and trimming that claims to beat llama.cpp in spots — W61k3r · 2026-11-05
- Gary Marcus predicts the data center binge will collapse, with many bankruptcies ahead — GaryMarcus · 2026-11-05
- Merge's eval claims its MCP connectors beat Microsoft's native servers on tool count and reliability — shensi · 2026-11-05
- TorchTPU compat study: 4,821 of 4,917 community models run crash-free on TPUs — jiqizhixin · 2026-11-05