The dialogue around AI and sustainability has a bit of a tug-of-war feel to it.
On the one hand, there are very real concerns about AI’s environmental impact, given its electricity and water use, as well as the embodied carbon tied to the facilities that power AI technology. On the flip side, AI has the potential to help us design smarter, detect inefficiencies faster, and even cut carbon at the source.
In this latest Digital Builder episode, I sit down with Sara Neff, General Manager of Sustainability and ESG at Microsoft Cloud Operations and Innovation, to unpack one of the biggest tensions in the built world today: the AI boom and the explosive growth of data centers.
We discuss
AI is driving massive demand for data centers, and with that comes more energy use, more cooling, and more materials. It very much seems like these goals are at odds with each other, but according to Sara, growth does not mean backing off on climate goals.
“At Microsoft, we have not changed any of our climate commitments,” she says. “It’s not, ‘We’re growing data centers, so no more 2030 goals.’ Microsoft is still committed to being carbon negative, water positive, zero waste, and protecting more land than we use by 2030.”
In other words, scaling up means stepping up. “If we’re building more, we need to do more and we’re up to that challenge.”
That includes securing enough carbon-free energy to meet demand and pushing the market forward on materials. “We know we need more low embodied carbon concrete, which is why we are doing deals for low carbon concrete, same for low carbon steel. Everything. It’s just more of it.”
The balance, she explains, comes down to discipline. Keep the commitments. Plan carefully. Keep going.
There is also a less visible part of the story that does not get enough attention. “More than half of a typical data center’s embodied carbon is in its HVAC equipment, in the cooling equipment itself,” Sara says. That means decarbonizing data centers is not just about power sources. It is about rethinking cooling systems from the ground up.
Microsoft recently collaborated on a lifecycle assessment of HVAC equipment, and the conclusion was encouraging. “The equipment that most leads to energy efficiency is also, over its lifecycle, the lowest embodied carbon.”
It may not seem like the most thrilling use case for AI, but Sara is most excited about its ability to find practical, measurable reductions in carbon, water, and energy.
AI can monitor water consumption, spot when something is off, and flag leaks before they turn into major waste. It can adjust cooling set points in noncritical spaces like admin areas to cut unnecessary energy use.
She is especially excited about AI for concrete. By analyzing a ready-mix supplier’s historical mix designs, AI can identify the lowest-carbon option that still meets performance and schedule requirements. That means fewer emissions without slowing projects down.
“These are examples of AI saving tons of carbon, saving gallons of water, saving energy. It’s all happening right now,” she says.
Sara continues, “We're all good at AI for process efficiency, and that's really important because it frees up more time to help do regular efficiency work. But I really get really excited about AI tools now for finding those reductions.”
Here’s the reality: you can’t decarbonize in silos. Collaboration is critical to getting sustainable solutions adopted at scale.
Sara shares a great example of this in action: Microsoft signed a deal for steel with 95 percent reduced embodied carbon for a new facility in Sweden. But as she put it, “I’m Microsoft. I don’t buy steel. I buy general contracting services.”
That decision forced deeper collaboration across the supply chain. “Now we have to work down our supply chain radically. How do we work with suppliers making rebar? How do we work with our HVAC suppliers?” What started as a procurement deal quickly became a coordination effort that stretched five steps downstream.
Sara explains, “We did this amazing procurement deal, and now we're having to radically collaborate with places in our supply chain. To be able to really teach everybody how to get green steel and make more of it in the market… that to me is wonderful and transformational.”
Concrete is no different. “There’s us, the owner, then the general contractor, then the concrete subcontractor, then the ready-mix supplier, and then the cement manufacturer. And all of those teams need to be on board for it to work.”
When Sara looks ahead, she’s not just thinking about better data centers. She is thinking about an AI-fluent industry.
“What I’m really, really excited about is the idea that everybody throughout the AECO world, as well as all of our suppliers, gets high levels of AI fluency,” she says. “We’re just in the early days of knowing what AI can do for us.”
In her view, the future is not about a single innovation. It is about widespread capability. Designers, contractors, operators, and suppliers would all understand how to use AI in their own sphere, whether that is design optimization, operational efficiency, or preventative maintenance.
“I really see, and I don’t think it’s going to be that far away, where everybody who helps us site, build, design, operate, deconstruct, and decommission a data center knows how to do it with a clear focus on climate and with AI making it go faster.”
Digital Builder is hosted by me, Eric Thomas. Remember, new episodes of Digital Builder go live every week. Listen to the Digital Builder Podcast on:
or wherever you listen to podcasts.
