Schneider Electric: AI building controls cut energy use by up to 22%
Schneider Electric has published research claiming that AI-enabled building management systems can reduce whole-building energy consumption by up to 22% compared with traditional controls. Released at Climate Week NYC 2026, the study quantifies annual utility savings of between $13,600 and $49,300 per building at current commercial rates, with the company arguing those savings recur year after year and compound further across larger property portfolios.
The report, titled AI for Climate: Quantifying the Energy and Carbon Impact of Building Optimization, uses energy modelling validated against real-world pilot deployments across Australia, India and the United States. It focuses specifically on AI-driven HVAC optimisation layered on top of existing digital building management systems, finding that the AI layer alone can contribute 7.2 to 12.7 percentage points of additional energy savings beyond what smart building controls achieve unaided. In some building scenarios, the company says, annual savings exceed 200 MWh. Carbon avoidance is reported at up to 60 metric tons of CO2 equivalent per building per year, which Schneider Electric compares to the benefit of planting 2,700 mature trees.
What the AI layer does
The optimisation model works by ingesting data that has historically been siloed across building systems: occupancy patterns, weather forecasts, equipment performance logs and operational inputs. The AI layer analyses those streams continuously and automates HVAC adjustments in real time, reducing the cognitive load on facility management teams and, the research claims, enabling smaller buildings to access a level of energy intelligence that was previously cost-prohibitive.
Pankaj Sharma, executive vice-president for software and services at Schneider Electric, said: "An AI layer on top of existing business systems can turn complexity into intelligence and intelligence into action, reducing emissions, lowering costs and improving performance simultaneously." Sharma also highlighted the democratisation angle, noting that small and mid-sized buildings under 100,000 square feet historically lacked the in-house expertise to justify sophisticated energy management; AI automation, he argued, changes that calculus. The study found that both cloud and edge AI deployment models can deliver meaningful energy and carbon benefits, giving operators flexibility in how they architect their systems.
Market context and competitive landscape
Schneider Electric is not alone in pursuing AI-powered building energy management. Johnson Controls, Siemens Smart Infrastructure and Honeywell Building Technologies all offer competing platforms with varying degrees of machine-learning integration, and a wave of specialist startups, including companies backed by significant venture funding in recent years, are targeting the same HVAC optimisation use case. The addressable market is substantial: buildings account for approximately 37% of global energy-related carbon emissions according to widely cited figures, and regulators in the EU, UK and several US states are tightening requirements on commercial building performance.
The EU's Energy Performance of Buildings Directive, which entered a revised phase in 2024, mandates progressive decarbonisation of commercial building stock and encourages adoption of smart-readiness indicators. In the UK, the Minimum Energy Efficiency Standards regime is under ongoing consultation for tighter thresholds. These frameworks are likely to accelerate enterprise procurement of precisely the kind of AI-enabled management layer Schneider Electric is promoting. However, the company's research is self-funded and has not been independently peer-reviewed, which limits the weight buyers and policymakers can place on the specific savings figures without third-party validation.
Caveats and next steps
The "up to 22%" headline figure represents the upper bound of a modelled range across differing building types and geographies; real-world results will vary by building age, climate zone, existing system baseline and quality of integration. Schneider Electric has not disclosed the names of pilot sites used to validate the modelling, nor the duration of the pilots. The full report is available for download via the company's website.
For enterprise buyers, the compelling case is the claimed ratio of carbon avoided to AI system footprint, which the research puts at greater than 100 to one. If that figure holds up under independent scrutiny, it provides a direct counterargument to the growing narrative around AI's own energy consumption, and could prove a useful data point in corporate sustainability reporting frameworks such as GHG Protocol Scope 2 accounting.