ViqusViqus
Navigate
Company
Blog
About Us
Contact
System Status
Enter Viqus Hub

Nvidia Claims Near-Zero Water Usage for AI Data Centers with Liquid Cooling

liquid cooling data center AI chips Nvidia water consumption sustainability
June 22, 2026
Source: The Verge AI
Viqus Verdict Logo Viqus Verdict Logo 7
Pivotal Infrastructure Shift
Media Hype 6/10
Real Impact 7/10

Article Summary

In response to increasing public scrutiny over AI infrastructure's environmental footprint, Nvidia unveiled a reference design for its data centers (Rubin generation) that employs full liquid cooling. This system significantly increases operating temperatures, up to 113 degrees Fahrenheit (45 degrees Celsius), which allows heat to be captured directly at the chip level. Nvidia claims this method drastically reduces water consumption, projecting a transition from conventional cooling-tower-based systems' use of 2.6 million gallons per megawatt per year to near zero. The technology aims to increase efficiency and flexibility, allowing outdoor dry coolers to reject heat even when ambient air temperatures are challenging. While praising the efficiency gains, the article notes concerns remain regarding the upfront construction costs and the overall power generation requirements of such massive facilities.

Key Points

  • Nvidia’s new design utilizes 100% liquid cooling and raises operating temperatures to improve thermal management efficiency.
  • The transition is claimed to reduce water usage from millions of gallons per megawatt per year to near zero, addressing major environmental concerns.
  • Industry adoption of this highly efficient cooling method is anticipated, despite unanswered questions regarding construction costs and total power draw.

Why It Matters

This announcement is a critical industry response to mounting ESG pressure. By dramatically reducing water consumption, Nvidia is setting a new environmental standard for AI hardware. Professionals should care because this infrastructure shift represents a major CapEx consideration for cloud providers and hyperscalers. While the efficiency gains are real, the market needs more clarity on the total cost of ownership, including the build-out expense versus traditional air cooling, which could dictate immediate deployment timelines.

You might also be interested in