Energy strategist Katie McGinty highlights that roughly one-third of data center energy is wasted on cooling and non-compute work, even as the sector's electricity demand is projected to double by 2030. Rather than relying solely on new power generation, she argues that efficiency improvements and thermal recovery—which can be deployed in months rather than years—represent an immediately accessible energy supply of 3,000 terawatt-hours annually globally, enough to power hundreds of millions of homes if captured.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Energy strategist Katie McGinty argues that global waste heat—roughly 3,000 terawatt-hours annually—represents an untapped energy resource equivalent to about three-quarters of annual U.S. electricity consumption. Modern data centers operate at average power usage effectiveness (PUE) of 1.5–1.6, meaning roughly one-third of total energy is wasted on non-compute work; leading operators have demonstrated <1.3 annualized PUE targets, cutting non-compute energy by 50%.
Why it matters
Data center electricity demand is projected to more than double by 2030 to around 945 terawatt-hours—equal to about the entire electricity consumption of Japan today—yet a meaningful share of that demand is overhead rather than intrinsic computation. Improving efficiency and thermal recovery can unlock capacity immediately, without waiting years for permits and construction. Technologies like absorption chillers can reduce chiller electricity by 90% while converting waste into usable cooling.
What to watch
Waste heat recovery and heat pump technology have been shown to slash energy bills 32% while cutting emissions by 60%. In Europe, waste heat from process industries is roughly equivalent to total EU building heat demand, showing the scale of untapped thermal resources across regions.
The article reframes energy supply in the context of accelerating AI infrastructure demand. Rather than treating efficiency as a secondary concern, McGinty positions waste heat recovery and improved cooling systems as a primary energy resource—one that can be accessed far more quickly than building new generation and transmission capacity. This reflects a shift from the traditional "build more" logic that has dominated energy strategy for decades.
The scale of the opportunity is substantial: global waste heat alone could theoretically satisfy a significant portion of new demand without new infrastructure. Data centers exemplify this tension—as AI workloads drive electricity demand to more than double by 2030 (projected at around 945 terawatt-hours), the efficiency losses within those same facilities represent an enormous hidden capacity reserve. Technologies like absorption chillers and heat pumps are already available and can be deployed within months, whereas conventional infrastructure projects take years or decades to permit and construct.
McGinty's argument carries operational and economic weight beyond the engineering: waste heat recovery can reduce operating costs, increase available capacity, and improve resilience, with global estimates suggesting tens of billions in annual savings from heat recovery alone. For businesses planning infrastructure expansion, this implies that prioritizing efficiency and thermal integration alongside new generation may deliver faster results and lower costs than traditional expansion alone.
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
