Why AI won’t decarbonise real estate

Real estate has never had better tools for understanding building performance and where to cut carbon. Building energy management systems paired with AI analytics find inefficiencies faster than ever. Yet most utility optimisation programmes still fail to turn insight into results.

The technology and the recommendations are rarely the problem. It’s what happens next: who owns the work, whether it reflects operational reality, and whether the right people are involved. Asset managers focus on value, property managers on occupier satisfaction, facilities teams on equipment uptime, occupiers on comfort and occupancy costs, and sustainability teams on carbon and compliance.

Each act rationally, but nobody owns building performance end to end. In 16 years in real estate, I’ve seen programmes fail for the same reasons repeatedly.

AI helps teams reach the right conclusion faster, but data in does not mean decarbonisation out. Technology flags a faulty temperature sensor or equipment running 24/7 unnecessarily. It does not consider lease conditions, operational constraints or the priorities of the people running the building.

Used well, AI is valuable for improving data quality and speeding up analysis. I recently saw an AI-supported review of 1.18 million data points from 1,800 sub-meters in a commercial office. A four-hour process took 10 minutes. That is a real productivity gain, but on its own it delivers nothing to the customer. Someone still needs to interpret the findings and drive implementation.

Technology should accelerate decisions, not become a substitute for making them. The priority is getting the right information to the right people in a format they will use, without adding yet another platform or login.

Good decisions start with granular data and a detailed knowledge of building requirements. On one site, sub-meter data showed heavy overnight consumption in an empty building. The cause was a security guard on patrol triggering an occupancy sensor, which switched on the central HVAC, unnecessarily. The technology worked exactly as designed. It just hadn’t accounted for how the building was used.

Personnel change can be just as disruptive. On an office project in Germany, getting the BMS to control the ventilation units correctly was a simple fix. However, it took over 18 months to execute because the property, facilities and BMS contracts were all re-tendered. The overarching energy management programme kept ownership intact, and the work was eventually delivered.

Delivery must be designed in from day one. That means working alongside site teams, naming who implements each recommendation, and setting measurable targets such as annual energy reduction targets and bonus structures for the individuals responsible for implementation. A recommendation made once almost never lands. Successful programmes stay close to the people who turn insight into action, and later, into results.

A programme that relies on just one internal champion is vulnerable by design. When that person leaves, the knowledge, relationships and momentum leave too. Building performance must be embedded across individual KPIs, lease agreements and capex and opex budgeting processes. Whilst this is not always simple, the impact of this alignment is often the difference between success and failure.

Embedding performance and leveraging data also means building it into how landlords and tenants work together. A leading UK REIT has implemented a comprehensive metering maintenance programme which includes service level agreements (SLAs) on meter reliability and data quality, giving them a sound basis for compliance, decarbonisation decisions and providing tenants a clear view of their energy use and occupancy costs. The result is greater trust, more constructive engagement and a stronger case for investment in decarbonisation projects including the replacement of gas boilers with air source heat pumps.

Transparency builds trust, but lease terms still limit the ceiling of what is possible. I’ve seen many site teams required by a lease clause to run heating and cooling around the clock, even when the building is largely empty. That is a structural problem, not an engagement problem. One fix is linking rent to energy use. Higher consumption then costs more, savings benefit both parties and unnecessary 24/7 operation of equipment largely disappears.

Decarbonising buildings does not depend on better technology or bolder targets. It depends on systems where the right action makes commercial and operational sense, and where responsibility doesn’t sit with one person or team. As the UK Green Building Council recently highlighted, the industry doesn’t have an innovation problem. It has a scaling problem. The solutions exist, and they now need embedding across thousands of assets and stakeholder groups.

The industry does not need to wait for the next technological breakthrough. It needs to make the solutions already available work at scale by aligning incentives, embedding accountability and empowering the people who operate buildings every day. The potential of AI/ML is enormous, but only people working within well-designed systems can turn it into decarbonisation at scale.