As tech giants push the envelope on the capacities of data centers to meet the overwhelming demands of artificial intelligence development, a significant challenge looms on the horizon: the increasing scarcity of electricity. This necessity, as reported by the U.S. Energy Information Administration, is seeing an estimated annual growth of 3% in commercial demand for electricity, concentrated largely in states like Virginia and Texas that are swiftly becoming the epicenters for large-scale computing operations.
Insights from CBRE's "Global Data Center Trends 2024" study, reveals an explosive 24.4% increase year over year in North American data center inventory in the first quarter of 2024, with newly built centers boasting capacities ranging from 100 to 1,000 megawatts. This surge, notably comparable to the power needed to light up 80,000 to 800,000 homes, is spurring a community backlash due to the environmental implications of such high electricity and water demands for AI-driven facilities, as grassroots groups begin voicing their concerns, according to the Oak Ridge National Laboratory official website.
In a scenario where projections from the Electric Power Research Institute (EPRI) suggest data centers could be guzzling up to 6.8% of the total U.S. electricity generation by 2030, innovating in the field of energy-efficient supercomputing has never been more critical. The Oak Ridge Leadership Computing Facility (OLCF), a division at Oak Ridge National Laboratory, has been pioneering this very innovation. In a testament to their progress, OLCF's flagship supercomputer, Frontier, not only leads the TOP500 list of the world’s most powerful computers but also tops the Green500 list as the most energy-efficient, as the Oak Ridge National Laboratory reported.
Despite the commercial sector's burgeoning ventures into high-performance computing (HPC), energy conservation remains a peripheral concern, until constraints necessitate a shift. "Our machines were always the biggest ones on the planet, but that is no longer true. Private companies are now deploying machines that are several times larger than Frontier,” Scott Atchley, chief technology officer of the National Center for Computational Sciences at ORNL, acknowledged in a statement obtained by the Oak Ridge National Laboratory. “That will change once they become more power constrained, and they will want to get the most bang for their buck.”
With a wealth of experience in optimizing HPC energy efficiency, the OLCF stands as a potential paragon for industries at large. “We are uniquely positioned to influence the full energy-efficiency ecosystem of HPC, from the applications to the hardware to the facilities,” OLCF program director Ashley Barker explained in an interview cited by ORNL. In an industry scrambling to address electricity consumption concerns, the strategies and methodologies honed by OLCF through structures like the upcoming Discovery supercomputer are becoming more relevant than ever to create a sustainable path forward for the business of high-power computing.
