NVIDIA unveils water-saving AI data center design
What's the story
NVIDIA has revealed a new Rubin generation reference design for a fully liquid-cooled data center runs hotter and saves a lot of water. The company claims this innovative design has "eliminated massive amounts of power usage and pretty much all water usage." However, it doesn't address all concerns about AI data centers, including their construction and power generation needs.
Water savings
Drastic reduction in water usage
NVIDIA's head of sustainability, Josh Parker, said 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." This is a major improvement over traditional cooling methods. The efficiency gains are partly due to running AI servers at higher temperatures, as high as 45°C (113°F).
Heat management
A promising solution for data center cooling challenges
NVIDIA's system captures heat directly at the chip and transports it through liquid loops operating at much higher temperatures. This allows outdoor dry coolers to reject heat efficiently for most of the year. The design offers more flexibility when it comes to ambient air temperature, making it a promising solution for data center cooling challenges.
Operational efficiency
The new 'closed-loop' cooling system
NVIDIA's new, "closed-loop" cooling system uses a liquid coolant made of three-quarters water and one-quarter propylene glycol. The coolant can remain operational at temperatures of up to 115°F, significantly higher than the temperature limit manufacturers build into hot tubs. This allows the system to prevent chips from overheating without using large amounts of water and fans, resulting in lower operational costs and a reduced ecological footprint.
Industry impact
Attracting the AI industry
NVIDIA's new cooling system, which can produce "up to a 100% reduction in water use," is likely to attract the AI industry. This comes at a time when public concern over AI's ecological impacts is on the rise. A recent Pew Research Center survey found that most Americans familiar with data centers consider them "mostly bad" for the environment and household energy costs.
Cost reduction
Potential to reduce overhead costs
NVIDIA's new cooling system could also help tech companies reduce their overhead costs by cutting back on the power used by their data centers. This could potentially lead to a decrease in the price of tokens, which are the basic unit used to measure AI usage. However, it's worth noting that the effectiveness of this system varies depending on climate conditions and its scalability across the industry is still uncertain.