WRRC Webinar Explores Tech Innovation and Data Center Sustainability
The data center debate has taken off in recent years as demand for AI increases and communities grapple with the facilities’ environmental consequences and high energy needs. According to David Ebert, U of A’s Chief AI and Data Science Officer, data centers are here to stay, but their design isn’t set in stone; ongoing technological innovation could make data centers more energy efficient, allowing them to conserve water.
Ebert gave a WRRC Water Webinar on September 8 to discuss the history of computing infrastructure, the changing landscape of data center technology, and how Arizona is uniquely positioned to contribute to sustainable data center innovation.
New technologies, like AI-driven predictive cooling and next generation hardware, could be part of the shift toward conservation. Photonic processors, for example — which are being studied at the U of A — use light instead of electricity to process calculations, meaning they use less power than traditional processors, and give off less waste heat.
That said, there’s no guarantee that more efficient data centers will translate into data centers that use less water and energy; in fact, the opposite may be true. Ebert referenced “Jevons Paradox,” which holds that increasing efficiency increases the overall use of a resource. The paradox emerged out of 1860s England, when economist William Stanley Jevons posited that finding more efficient ways to use coal would effectively decrease the cost of coal energy and, consequently, drive up demand.
Ebert’s work focuses on data science, not policy, but he suspects that tech alone may not be sufficient to prevent unsustainable data center growth — regulation will likely be necessary. In some European countries, environmental sustainability requirements are a part of data center projects from the start, and these countries could offer models for US regulatory frameworks.
Ultimately, the U of A is poised to be a leader in the data center sustainability space, because it is home to experts on all aspects of the data center ecosystem, including AI, photonic processing, and water policy.
“We really need to do collaborative co-design,” said Ebert. “We need to not just look at the hardware and the software, but we need to bring the operations and the water issues in and the electrical issues into the final design, so that you get the optimization that makes the most sense and... you get the best design for your situation.”