SemiAnalysis: NVIDIA B200 serving DeepSeek can yield up to $15B annual profit per gigawatt
zephyr_z9 · x · 2026-09-25
SemiAnalysis estimates that NVIDIA vLLM on B200 serving the open DeepSeek v4.1 Flash model at official interactivity and selling prices can generate up to $15 billion in annual profits per gigawatt — calling it a "money printer." Using Engram DRAM offloading on NVIDIA hardware reportedly increases revenue per gigawatt by 50%. A striking take on the economics of serving open models on accelerated infrastructure.
More from Infra
- Altman says OpenAI aims to replace Nvidia in the medium to long term — zephyr_z9 · 2026-11-11
- Chinese University Bought Nvidia H200 Chips Despite US Blacklist — pstAsiatech · 2026-11-11
- Baidu's ERNIE Evolve tops public bests on NanoGPT Speedrun, brings RSI to real inference optimization — xiaohu · 2026-11-11
- Baidu says RSI auto-optimized ERNIE inference, 6.24x decode speedup on Kunxin P800 — xiaohu · 2026-11-11
- Nvidia's supply commitments jump from $119B to $279B in one quarter, locking up 2027-28 memory — tengyanAI · 2026-11-11
- Train 27B models on a 24GB GPU: ternary optimizer cuts memory to 1 byte/weight — oguzeldereli · 2026-11-11