A new research framework suggests that AI data centers, which consume enormous amounts of electricity, could be transformed into engines for removing carbon dioxide from the atmosphere rather than simply contributing to climate change. Scientists have developed a thermodynamically integrated system that captures the low-grade waste heat generated by AI data centers and upgrades it via heat pumps to power direct air capture (DAC) systems, which pull CO2 directly from the air.

Why Is Data Center Waste Heat Such a Missed Opportunity?

The scale of AI infrastructure growth is staggering. Global data center electricity consumption reached approximately 415 terawatt-hours in 2024 and is projected to rise to 945 terawatt-hours by 2030, nearly equivalent to Japan's entire annual electricity consumption. In the United States alone, data centers accounted for 4.4% of total electricity consumption in 2023, with projections suggesting an increase to 6.7 to 12% by 2028.

Yet nearly all of this electrical energy is ultimately dissipated as heat. Cooling systems already account for 30 to 40% of total data center electricity consumption, and this burden is intensifying as computing power density increases. Modern AI data centers, particularly those running graphics processing units (GPUs) for artificial intelligence tasks, generate large volumes of low-grade waste heat, typically in the form of 50 to 60 degrees Celsius return water from liquid cooling systems. This represents a substantial thermal resource that has been largely underutilized.

Current approaches to valorizing this waste heat primarily focus on integrating it into district heating systems, which has achieved notable success in some regions. For example, data center waste heat has achieved up to 36% CO2 reduction in Finland and supplied up to 80% of local heating demand in specific deployments. However, these approaches are inherently constrained by geographic proximity, climate conditions, and the limited transportability of low-grade heat.

How Could Integrated DAC Systems Work at Scale?

Direct air capture systems operate through cyclic adsorption and desorption processes, where ambient air is drawn through contactors and CO2 is selectively captured by solid absorbents, followed by thermal regeneration to release a concentrated CO2 stream for utilization or storage. The primary energy bottleneck in DAC systems is the regeneration step, which typically requires temperatures between 80 and 120 degrees Celsius.

The research team assessed whether AI data center waste heat could be upgraded via heat pumps to supply this thermal demand across different regions of the United States, accounting for data center capacity, server composition, local climate, electricity prices, and grid carbon intensity. The findings suggest that integration could substantially improve net CO2 removal and lower the levelized cost of capture, which is the average cost to remove one ton of CO2 from the air.

Under a 2030 scenario with more GPU-intensive data centers and cleaner electrical grids, several states could achieve removal ratios above 1, indicating that integrated systems could offset their own operational emissions and deliver additional carbon removal. In carbon-intensive regions, integration can flip direct air capture from net-positive emissions to net-negative, meaning the system removes more carbon than it produces.

Steps to Implement Data Center Carbon Integration

• Regional Assessment: Evaluate local climate conditions, grid carbon intensity, and waste heat availability to determine feasibility and optimize system configuration for specific geographic areas.
• Heat Pump Integration: Install heat pump systems that can upgrade low-grade waste heat from 50 to 60 degrees Celsius to the 80 to 120 degrees Celsius required for direct air capture regeneration.
• Co-Design Optimization: Conduct careful co-design and optimization of system configuration and operation under region-specific conditions to maximize carbon capture benefit and operational efficiency.

What Are the Broader Implications for AI Infrastructure?

The research addresses a fundamental tension in the AI era: the very infrastructure enabling AI-driven decarbonization across sectors is itself emerging as a significant and rapidly growing source of emissions. Training a single large-scale language model such as GPT-3 has been estimated to emit 552 tons of CO2-equivalent. A bottom-up assessment indicates that 2,132 data centers in the United States emitted over 105 million tons of CO2-equivalent in 2023, accounting for 2.18% of national emissions, comparable to the annual tailpipe emissions of more than 22 million passenger vehicles.

Continued expansion of AI server infrastructure is expected to contribute an additional 24 to 44 million tons of CO2-equivalent between 2024 and 2030. This trajectory has prompted discussions about how to reconcile rapid AI infrastructure growth with climate mitigation goals. The integrated DAC-data center approach represents one potential pathway, though the research emphasizes that maximizing carbon capture benefit requires careful consideration of region-specific conditions.

The key unanswered question, according to the research team, is whether waste heat from AI data centers can be effectively harnessed via heat pump integration to drive direct air capture systems, thereby reconciling the rapid expansion of AI infrastructure with climate mitigation goals. Given the substantial spatial heterogeneity in climate conditions, grid carbon intensity, and waste heat availability across different regions, the answer appears to be yes, but with important regional variations in effectiveness and economic viability.

This framework could reshape how technology companies and policymakers think about data center expansion. Rather than viewing data centers purely as energy consumers, they could be designed and operated as integrated systems that generate both computational value and environmental benefit. As AI infrastructure continues to expand at unprecedented rates, approaches that leverage waste heat for carbon removal may become increasingly important for achieving broader climate goals.