Enterprise

Nvidia Rubin data centers cut water use with liquid cooling

The chipmaker claims its reference design eliminates nearly all water consumption by running AI servers at higher temperatures.

Omega Editorial· June 23, 2026· 2 min read

Nvidia has unveiled a reference design for its Rubin generation data centers that the company claims can reduce water consumption to near zero through full liquid cooling and higher operating temperatures.

The design addresses growing public concern about the environmental impact of AI infrastructure, though questions remain about construction costs and the broader energy footprint of these facilities.

How the system works

The efficiency gains stem from two key changes: switching to 100 percent liquid cooling and allowing AI servers to run significantly hotter than traditional data centers. Nvidia's design permits temperatures as high as 113 degrees Fahrenheit (45 degrees Celsius).

According to the company, heat is captured directly at the chip level and transported through liquid loops operating at elevated temperatures. This approach enables outdoor dry coolers to reject heat efficiently throughout much of the year, with greater flexibility regarding ambient air temperature conditions.

Water savings at scale

Josh Parker, Nvidia's head of sustainability, stated that the reference design reduces water use "from roughly 2.6 million gallons per megawatt per year for conventional cooling-tower-based systems to near zero — up to a 100 percent reduction."

The company claims that every cloud provider and data center operator building for Rubin is adopting this liquid cooling approach, though Nvidia's announcement did not disclose the capital costs associated with building liquid-cooled facilities compared to traditional air-cooled designs.

Why it matters

Data centers have become flashpoints for community opposition due to their resource demands. While Nvidia's water reduction claims are significant, the announcement leaves unaddressed the substantial power generation requirements and construction-phase environmental impacts of large AI facilities. Amazon has similarly promoted higher heat tolerances as part of efficiency improvements in its primarily air-cooled data centers, suggesting the industry is converging on thermal management as a key sustainability lever.

Remaining concerns

The water savings, while substantial, represent only one dimension of data center environmental impact. The facilities still require massive amounts of electricity, and the construction process itself carries environmental costs that Nvidia's announcement did not address.

These details were first reported by The Verge, which noted that Nvidia's blog post omitted cost comparisons between liquid-cooled and air-cooled infrastructure.

#nvidia#data centers#liquid cooling#ai infrastructure#sustainability#rubin

This is an original analysis by the Omega editorial team. Source reporting: The Verge.

Want systems like this working for your business?

Book a Call

More in Enterprise

Enterprise· 3 min read

NVIDIA Launches DSX Ready Qualification for AI Factory Components

New program certifies battery storage and cooling systems against reference design standards as infrastructure constraints reshape datacenter planning.

Via AI Watch · Sep 21, 2026
Enterprise· 3 min read

Positron IDE Now Runs on Amazon SageMaker for Unified Data Science

Posit's integrated development environment brings R and Python workflows together on managed infrastructure with governed data access and AI assistance.

Via AI Watch · Sep 21, 2026
Enterprise· 3 min read

Benchling Isolates AI Agent Code with Multi-Layer AWS Defense

Life sciences platform runs 600+ daily AI-generated code sessions across 250 tenants using DNS firewall, VPC endpoints, and per-job credentials.

Via AI Watch · Sep 21, 2026