The global pursuit of energy to fuel an expanding world economy often focuses on building new infrastructure, from power plants to transmission lines. Yet, a significant, often overlooked, energy source already exists, embedded within the very systems designed to generate and distribute power. Katie McGinty argues that this embedded capacity, largely in the form of wasted energy, represents a critical, immediate opportunity for meeting escalating demand, particularly as technologies like artificial intelligence drive unprecedented consumption spikes.
Consider the sheer scale of the energy currently produced but never effectively utilized. Globally, nearly half of all energy input is lost, primarily as heat, across various sectors including power generation, industrial processes, and commercial buildings. Industrial systems alone squander between 20% and 50% of their energy as heat, a colossal inefficiency that constitutes a parallel energy system, vast in its scope but largely ignored. Each year, over 3,000 terawatt-hours of usable waste heat goes untapped worldwide. To put this into perspective, that volume of energy is roughly equivalent to three-quarters of the United States’ annual electricity consumption, enough to power hundreds of millions of homes. This isn’t a marginal loss; it’s a fundamental design flaw with immense potential for recovery.
The urgency to address this waste is amplified by the burgeoning demands of the digital age. The International Energy Agency projects that electricity consumption by data centers will more than double by 2030, reaching approximately 945 terawatt-hours. This figure alone matches the entire current electricity consumption of Japan. In advanced economies, data centers are set to account for over 20% of electricity demand growth this decade. The conventional response often points to a need for massive expansion of energy generation. However, a substantial portion of this demand isn’t inherent to computation itself; it’s overhead, largely driven by legacy assumptions in facility design, particularly concerning cooling.
Today’s data centers, for instance, still operate with an average Power Usage Effectiveness (PUE) of around 1.5 to 1.6. This means that roughly one-third of the total energy consumed is not dedicated to actual computing work but is instead lost to non-compute functions, predominantly cooling. This inefficiency is not an unavoidable consequence but rather a design choice. Leading operators have demonstrated the ability to achieve PUE targets of less than 1.3 annually, while also eliminating the need for water evaporation in their cooling processes. Such advancements suggest that non-compute related energy consumption could be halved, freeing up significant capacity without drawing on community resources like water. This immediate availability of saved energy can act as a new supply, bypassing the lengthy processes of permitting, financing, and construction associated with new generation.
The same principles apply to thermal energy, which constitutes nearly half of the global economy’s final energy consumption. Yet, once generated, heat is rarely treated as a valuable resource. Technologies like absorption chillers, for example, can utilize heat, rather than electricity, for cooling purposes. Integrating these into systems, such as data centers, can reduce chiller electricity consumption by as much as 90%, transforming what was once waste into a productive asset. The potential here is not about incremental improvements but about fundamentally enhancing the productivity of existing energy systems. In Europe, the waste heat generated by process industries is roughly equivalent to the total heat demand for all buildings across the EU.
The advantages extend beyond technical feasibility to include compelling economic and operational benefits. While constructing new power plants and expanding grids can take years, even decades, efficiency improvements and thermal integration projects can be deployed in a matter of months. This speed of deployment offers a significant competitive edge in a system experiencing rapid growth. Capturing and reusing energy translates directly into reduced operating costs, increased available capacity, and enhanced resilience. Global estimates suggest that waste heat recovery alone could generate tens of billions of dollars in annual savings. Furthermore, advancements in heat pump technology have shown the capacity to cut energy bills by 32% while simultaneously reducing emissions by 60%.
While the need for large-scale infrastructure development will persist to meet long-term demand, the competitive landscape in energy is undeniably shifting. Success will increasingly favor those who combine traditional scale with agility, delivering capacity more quickly, efficiently, and intelligently. This necessitates treating efficiency and thermal energy not as secondary considerations, but as core components of energy supply. The current model often prioritizes measuring energy production, while largely overlooking the energy lost. In this emerging era, speed, precision, and adaptability will dictate success, making efficiency and thermal energy paramount resources. The future belongs to those who move decisively to unlock the energy already generated, turning what was once waste into a vital resource.
