The growth in GPU-based applications is raising energy consumption and costs at an accelerating rate. Data centers are simply running out of available power. Arranging to have your utility bring in more power is both expensive and time-consuming.

Data Centers Are Energy Hogs

Data centers and cloud providers now devour 6% of global electricity. By 2025 that number could easily reach 20%! ¹

While you’ve probably heard the adage “you can’t squeeze blood from a stone,” it is possible to squeeze a whole lot more compute from your existing power envelope. Data Center infrastructure & operations professionals can turn to liquid-immersion cooling, which can reduce energy costs by up to 50%.

Liquid immersion has an extremely low PUE. In particular, the power requirement for GRC immersion cooling systems is less than that needed for the removal of server fans. GRC’s patented immersion system power requirement amounts to a mere 3% of the “true” compute load versus an industry average of 70%.

Data center infrastructure and density demands are changing dramatically, driven by the advance of next-generation application architectures in AI, IoT, AR along with microservices. New highperformance, more densely packed processors and co-processors consume ever greater power and generate more heat than most existing data centers were designed to handle. Faced with significant time and money expenditures, many operators are struggling to keep up much less stay ahead of the curve.

Liquid immersion shatters the typical limitations of legacy air cooling, raising the bar to support 200 kW/rack densities.² Bottom line: this newfound cooling capacity flat-out solves the pervasive density dilemma.

Here again, GRC has helped many companies around the world find a cure for their density woes through liquid-immersion cooling; and in the process we’ve made energy costs a smaller, less bitter pill to swallow:

For Houston’s top geoscience company, CGG, we doubled the number of servers hosted in their facility (and the per rack heat load capacity) while helping them achieve a 50% reduction in total data center energy usage. Negating the need to purchase new, capital-intensive cooling equipment, we built out the required cooling capacity in 100 kW increments, which eliminated any downtime.

We also delivered super results for the supercomputing facility DUG, mentioned earlier. They operate “Bruce” which, at eight petaflops, is the Southern Hemisphere’s largest commercial supercomputer running one of the world’s largest deployments of Intel® Xeon Phi™ co-processors. With GRC’s help DUG was able to develop their impressive compute while vastly paring back costs along the way. We’re talking a 90% drop in cooling energy consumption and a 35% total energy cost reduction.