Researchers at Ulsan National Institute of Science and Technology have designed an integrated energy system for AI data centers that turns waste heat into electricity and pairs it with carbon capture. The setup uses solid oxide fuel cells for on-site power, an organic Rankine cycle to recover heat from server cooling, and CO2 capture. In a modeled 20,000 GPU facility, the team reports a 48% drop in total operating costs versus a conventional grid and air cooling setup. Cooling accounts for roughly 39% of data center electricity use, so the savings are concentrated where power demand is highest. The study looked at 24 scenarios across South Korea, the US, and the EU. ORC-based cooling saved up to $19 million per year in cooling power costs, while greenhouse gas emissions fell by 80,000 to 150,000 tons annually. Factoring in emissions trading added $5 million to $10 million a year in avoided carbon costs. One important caveat: the best power source depends on local energy prices. In the US, cheap natural gas made on-site SOFC generation the economic winner. In South Korea, higher gas prices meant sticking with grid power was cheaper. The findings were published in Chemical Engineering Journal.
