Companies say emissions cut from AI systems will offset enviornmental cost of running AI.
Montréal startup BrainBox AI and its parent company Trane Technologies officially marked the opening of an AI innovation lab that could dramatically cut energy waste in the world’s buildings on Wednesday.
Speaking to a packed atrium with a perfectly regulated temperature, Trane chair and CEO Dave Regnery said that most buildings waste about 30 percent of the energy they pay for, and that “the greatest opportunity for agentic AI is smarter, more resilient buildings.”
“An intelligent building is great, but a building that you can trust is even better.”
Jean-Simon Venne,
Brainbox
Trane has long provided software to manage energy consumption in buildings. Now, the company is turning to autonomous AI tools to help automatically cut consumption when it’s not needed.
Trane acquired BrainBox AI at the start of 2025, incorporating its AI solutions for building temperature management into its HVAC technology portfolio. Since then, it kept its Montréal presence of roughly 200 employees, half of whom are working at the new innovation lab.
The lab will help Trane and BrainBox further their goals of reducing the energy waste from buildings by 30 percent, executives told reporters, and connect more buildings to their network of more than 68,000.
An intelligent building you can trust
Founded in 2017, BrainBox launched its HVAC monitoring AI software in 2019 to help lower energy consumption in buildings such as apartment complexes and office buildings. The company claims its AI solution can reduce carbon emissions by up to 40 percent and utility costs by up to 25 percent. Its AI assistant for building managers and technicians, called Aria, is designed to make it easier for technicians to manage and troubleshoot HVAC systems.
Jean-Simon Venne, founder and president of BrainBox, told BetaKit on Wednesday that the lab’s product development will help Trane scale its technology across thousands, or even millions, of commercial buildings. Venne envisions AI agents, which can take action without human intervention, automatically adjusting temperature controls and energy loads to cut costs and energy consumption for building managers. He added that the focus now is to ensure that such AI systems are developed responsibly—for example, ensuring that as agentic AI is implemented, the same level of data security and encryption is maintained for customers.
RELATED: BrainBox AI CEO Sam Ramadori joins Yoshua Bengio’s LawZero as co-president and executive director
“An intelligent building is great, but a building that you can trust is even better,” Venne said.
In a showroom tour at BrainBox offices, Trane Technologies employees gave demos for the various ways Aria’s underlying algorithms work to control temperature in client buildings. For example, one algorithm monitors weather conditions to adjust internal temperature controls. If a sweltering day is forecast, HVAC systems kick in earlier to balance out energy consumption for cooling systems.Another automatically adjusts water chiller systems based on building occupancy—so if a crowd disperses after an office party wraps up, the system will reduce its cooling efforts.
Wraparound screens at the offices displayed videos and statistics tracking how much in emissions BrainBox’s technology had saved its corporate clients (typically over 20 percent) as the media tour was shuttled from station to station. It was clear that the traditionally hands-on HVAC industry is closely tracking new technology; a reporter from a US-based publication read by HVAC technicians had come to see the showroom.
The environmental costs of AI
Deploying AI agents comes with its own environmental costs, however. The energy consumption of data centres is expected to grow by 175 percent from 2023 levels by the end of the decade, and electricity used in AI data centres tends to be dirtier on average.
Venne claimed that the energy efficiency gains BrainBox creates more than offset the cost of running AI models that are processing HVAC data.
“Let’s use AI to go get that savings and reduce the impact,” Venne said. “But let’s also design AI models that are consuming less energy when they’re running in the cloud.”
Trane Technologies has roughly 46,000 employees worldwide and approximately 900 in Canada. The company, which is listed on the New York Stock Exchange, has seen its stock rise by 150 percent over the past five years to a market cap of just under $100 billion USD. Regnery told BetaKit that Trane has no plans to lay off staff for AI-related efficiency purposes, despite similar moves from public tech companies in recent months. “We have AI employed for growth,” Regnery said.
Feature image courtesy Madison McLauchlan for BetaKit.