Liquid cooling yields ~22% more compute per megawatt, NVIDIA-Vertiv DSX deal shows
BenBajarin · x · 2026-10-22
Ben Bajarin of Creative Strategies says the NVIDIA-Vertiv DSX liquid cooling announcement validates the thesis in their April 9 liquid cooling report.
- Their model: one facility megawatt runs 361 accelerators with air cooling at PUE 1.40, 440 at PUE 1.15 (+22%), and 473 at PUE 1.07 (+31%)
- That means roughly 22% more compute and tokens from the same power — free capacity in a power-capped market
- Vertiv claims DSX delivers up to 30% more tokens in year one at the same installed capacity
More from Infra
- Eden Deploys Vercel's Eve Agent to Cloudflare With One Command, Fully Open Source — ritakozlov · 2026-10-22
- Node.js ships emergency fix for high-severity Brotli decoder RCE in v22/24/26 — cramforce · 2026-10-22
- PyTorch's TorchTitan adds DeepSeek-V4, Kimi K3 support, RL training and big scale gains — PyTorch · 2026-10-22
- Epoch AI tracks 95 AI data centers with 14.2 GW — still just 44% of global AI compute — haider1 · 2026-10-22
- Two CVSS 9.2 RCE Vulnerabilities in PHP Push a European Government Site Offline — HankYeomans · 2026-10-22
- Old H100s Now Rent Above Launch Prices as Compute-per-Watt Value Keeps Climbing — abhiadesai · 2026-10-22