Signal65: CoreWeave up to 65% cheaper on GB200 over 3 years, buys 2.9x the tokens
ryanshrout · x · 2026-10-01
A Signal65 analysis modeled a 5,000-GPU NVIDIA Blackwell deployment over three years on CoreWeave and three hyperscaler clouds. On price alone, CoreWeave came in up to 65% lower on GB200 NVL72 and 52% lower on HGX B300, with no egress, support, or observability fees. Throughput widens the gap: running DeepSeek-R1 on identical GB200 NVL72 hardware, CoreWeave posted the top per-GPU MLPerf Inference v6.0 result, beating NVIDIA's reference and delivering 2.2x the throughput of the only hyperscaler that submitted. That turns a 22% price advantage into 65% lower cost per million tokens — a $1M annual commitment buys 1.33 trillion tokens on CoreWeave vs 466 billion on the competitor. Three-year cost differences across providers run from hundreds of millions to over $2B.
More from Infra
- Mobile AI Factory: agents auto-build on-device SLMs with 19x less GPU time — pmttyji · 2026-11-16
- vLLM serves DeepSeek V4.1-Flash at 954K tokens/s per GPU, $38.96/GPU-hour on GB300 NVL72 — bookwormengr · 2026-11-16
- Colossus 2's 1.2M B200s could generate ~$400B/year in tokens at DeepSeek Flash rates — bookwormengr · 2026-11-16
- BIS issues rule blocking remote access to US AI chips via third-country data centers — teortaxesTex · 2026-11-16
- Qdrant 1.20.1: 50% fewer cold-storage reads, ~2x faster quantization, agent-ready setup — qdrant_engine · 2026-11-16
- One 16GB GPU runs songs, voice, images, 720p video and local RAG — all offline and open source — llama-of-death · 2026-11-16