Nvidia announced DSX POD on Oct. 21, a reference design that ships seven Vera Rubin racks with cabling and liquid cooling in a container weighing about 28,000 kilograms. Nvidia says this can cut the time for a new AI data center to start serving tokens from six months to 14 days, and it expects partner deliveries to start in early 2027. It also says dry-cooler designs can bring facility cooling water near zero, while CoreWeave and Google Cloud are among the companies it lists as exploring MaxLPS power software.
/ Sources
/ About this story
Compiled by Venture Atlas from the sources above, using automated AI-assisted research. This is a summary of reporting published elsewhere, not original reporting - follow the source links for the full account. See our editorial standards.
Something wrong here? Email apostructura@gmail.com and we will fix it.
/ Related
- OpenAI will sell models through Google CloudWednesday, October 21, 2026
- Nvidia backs a reference humanoid due in early 2027Wednesday, October 21, 2026
- Nvidia makes the OpenShell agent architecture availableWednesday, October 21, 2026
- Google releases llm-d 1.0 for distributed model servingWednesday, October 21, 2026