AWS' Mission to Reduce the Environmental Impact of Buildings

The global carbon footprint of buildings surpasses any other single source of emissions – prompting AWS to apply its technological expertise to building efficiency.

At its flagship re:Invent conference in Las Vegas, AWS highlighted AI-powered systems designed to reduce the environmental impact of its facilities.

In a collaboration with HVAC specialist Trane Technologies and BrainBox AI, AWS has developed a system to make its grocery fulfilment network smarter and more energy efficient. The new virtual assistant for commercial buildings integrates directly with heating, ventilation, and air conditioning systems using AI to manage energy consumption.

Running on Amazon Bedrock, the system aims to continuously optimise temperature, airflow and energy use, moving beyond the limitations of fixed schedules and manual adjustments. Each facility is assigned a digital operator that interprets sensor data, predicts demand and adjusts equipment settings throughout the day to maintain conditions while reducing energy waste.

Piloting AI for energy optimisation

To test the system, AWS and Trane Technologies implemented the technology in three of Amazon’s North American grocery fulfilment centres. These sites present a complex energy management challenge due to their refrigeration loads, automation and 24/7 operations.

The pilot programme yielded positive results, with the AI assistant reducing the facilities' energy consumption by almost 15% – more than double the initial projection. These findings suggest that intelligent control systems can provide substantial energy savings before any physical retrofits are undertaken, reinforcing the business case for applying advanced analytics to existing building infrastructure.

The reductions correspond to lower operating costs and fewer carbon emissions.

Scaling for network-wide efficiency

Following the successful pilot, Amazon announced plans to roll out the AI-powered system across more than 30 grocery fulfilment and distribution centres in the US.

This expansion will take the virtual assistant from a small-scale test to a core component of Amazon’s grocery logistics footprint.

The move could multiply the potential carbon and cost savings as more facilities adopt the technology. The partners also plan to begin pilots in grocery stores next year, applying the system to customer-facing environments where climate control is critical for comfort and food safety.

Enhancing facility management with AI

The system offers more than just automated adjustments. By analysing vast streams of building data, the AI can assist facility managers in identifying, diagnosing and resolving issues more rapidly than traditional building management systems.

This predictive capability can flag problems from malfunctioning dampers to failing chillers, providing an early warning that helps reduce downtime and maintenance costs. It also ensures that energy savings are not compromised by hidden faults that might otherwise go undetected for long periods.

AWS presents this initiative as part of a wider strategy to address emissions from the built environment, a key area for climate action. Because the AI layer is software-based and runs on Amazon Bedrock, it can be updated and refined over time, learning from various facilities and potentially expanding to other building types beyond grocery fulfilment.

If applied to offices, warehouses, data centres, and retail spaces during the construction phase, similar efficiency improvements could lead to considerable carbon reductions on a global scale without needing to wait for deep retrofit cycles.

"These announcements reflect a broader truth," says Kara Hurst, Chief Sustainability Officer at Amazon. 

"AI and cloud technology are helping us build better businesses and a better future by enabling and accelerating climate action. And this is just the beginning."