How Amazon Is Using AI to Clean Up the Cloud

Amazon’s AI-designed carbon capture tech promises low-cost, on-site CO₂ removal, hinting at a climate-smart future for data centers

13 Dec 2024

Amazon is extending its reach beyond commerce and cloud services into an increasingly urgent arena: carbon removal. In a new pilot project in the UK, the company is testing an artificial intelligence–designed material that captures carbon dioxide directly inside an operating data center.

The trial is taking place at a facility run by cloud provider Civo and uses technology developed by climate-tech startup Orbital Materials. Unlike conventional climate strategies that rely on offsets or external mitigation, the system is designed to remove emissions at the point of generation, embedding carbon capture into the daily operations of digital infrastructure.

The material functions like a carbon sponge, absorbing CO₂ from the surrounding air. Its defining feature is not only where it operates, but how it was created. Orbital Materials used thousands of AI simulations to rapidly test and refine chemical structures, dramatically shortening the development cycle compared with traditional laboratory methods.

Cost is central to the project’s appeal. The company says deploying the material costs less than renting a high-performance computing chip for a single hour. If that claim holds at scale, carbon capture could shift from an expensive add-on to a practical tool that data center operators can deploy without redesigning their facilities.

The timing is deliberate. Demand for cloud computing and AI services continues to surge, and data centers are among the fastest-growing sources of energy consumption. For companies like Amazon, solutions that reduce emissions without constraining growth are becoming essential.

Significant hurdles remain. Regulators will require clear evidence that captured carbon is stored securely over the long term. The material’s durability across different climates and operating conditions is still unproven. Policy support, including incentives such as the US 45Q tax credit, could also shape whether the technology spreads beyond pilot projects.

Still, the broader signal is hard to miss. This experiment reflects a shift away from reactive climate measures toward tools built directly into the systems that drive the digital economy. If successful, AI-designed carbon capture may help data centers become not just engines of growth, but quieter participants in the climate equation.