Nvidia says its AI data center design runs hotter to use a lot less water

· Source: The Verge · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure · Depth: Intermediate, quick

Summary

Nvidia has introduced its Rubin generation reference design for a fully liquid-cooled data center, claiming it "eliminated massive amounts of power usage and pretty much all water usage." This design reportedly reduces water consumption from approximately 2.6 million gallons per megawatt per year for conventional cooling systems to near zero, representing up to a 100 percent reduction. A key efficiency gain comes from operating AI servers at higher temperatures, up to 113 degrees Fahrenheit (45 degrees Celsius). Heat is captured directly at the chip and transported through liquid loops, allowing outdoor dry coolers to efficiently reject heat. While Nvidia states that "every cloud provider and data center operator building for [Rubin] is making the transition," the design does not address construction costs or broader power generation requirements.

Key takeaway

For AI Architects evaluating future data center infrastructure, Nvidia's Rubin generation design signals a significant shift towards sustainable operations. You should consider liquid-cooled systems operating at higher temperatures to drastically reduce water consumption and potentially power usage. This approach could influence your long-term planning for AI factory deployments, especially given increasing public scrutiny on resource intensity. Investigate the cost implications and integration challenges of such advanced cooling solutions.

Key insights

Nvidia's Rubin data center design achieves near-zero water usage by employing high-temperature liquid cooling and direct chip heat capture.

Principles

Method

Heat is captured directly at the chip, transported via liquid loops operating at up to 113°F (45°C), then rejected efficiently by outdoor dry coolers, enabling near-zero water use.

In practice

Topics

Best for: CTO, VP of Engineering/Data, AI Architect, MLOps Engineer, Director of AI/ML

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Verge.