University of Houston Researchers Design Modular Carbon Capture Systems for AI Power Plants

Researchers at the University of Houston have designed modular carbon capture units built to fit inside standard 40-foot (12-meter) shipping containers, offering a scalable method to manage emissions from power plants supporting artificial intelligence infrastructure, according to a report by TechXplore (https://techxplore.com/news/2026-10-shipping-containersized-power-energy-efficient.html).

Detailed in a study published in Environmental Science & Technology, the system—dubbed small modular electrochemical containers (SMECs)—seeks to replace the sprawling, capital-intensive infrastructure required by traditional carbon capture facilities. The technology comes amid rapid growth in AI workloads; U.S. Department of Energy figures cited in the report indicate data centers accounted for about 4.4% of national electricity consumption in 2023 and could reach up to 12% by 2028, driving increased reliance on continuous natural gas generation.

Conventional carbon capture systems depend on large absorption towers, significant thermal energy, and extensive plant modifications. "With a conventional system, you're building a 130-foot-tall (40-meter-tall) structure designed for a specific power plant. With our technology, we can adjust the capacity by simply turning containers on or off," said UH engineering professor Mim Rahimi. "That flexibility is particularly important for data centers, where electricity can come from a combination of natural gas and renewable sources."

Unlike thermal systems, the SMEC architecture relies on electricity to strip carbon dioxide from exhaust gases before release. By integrating capture and release into a single electrochemical cell, the setup eliminates tall absorption columns and specialized filtration membranes. The units are built around membraneless electrochemically mediated amine regeneration (EMAR) cells developed in Rahimi's laboratory and first detailed in Nature Communications. "The modular, symmetric architecture of membraneless EMAR provided the foundation for the SMEC design," said co-author Ahmad Hassan, a former UH doctoral student and current postdoctoral associate at Harvard University.

In laboratory testing using simulated exhaust from natural gas and coal plants, the system achieved approximately 90% carbon dioxide capture efficiency under optimized parameters and sustained operation across 300 cycles over 50 hours.

Beyond bench-scale testing, the researchers modeled deployment at a hypothetical 360-megawatt natural gas power plant associated with OpenAI's Stargate AI data center project in Abilene, Texas. Their analysis projected that approximately 20 container-sized modules could capture 98% of the plant's carbon dioxide emissions, occupying a minor portion of the facility's 875-acre (354-hectare) site. The team is also working on adapting the technology for direct air capture.