The explosion of data centres across Virginia is sparking debate amongst residents, who now live with near constant reminders of artificial intelligence’s (AI’s) impact on both communities and their environment. While AI is being hailed by businesses for its transformative potential on industry, the communities on the front line of the AI boom may be more concerned with the technology’s environmental footprint and legacy.
Lalitha Krishnamoorthy PhD is vice president of AI & digital at Stantec
AI’s Carbon Footprint
The business world’s rush to adopt AI across nearly every aspect of their operations has come at a significant cost: powering these powerful AI systems depends on massive data centres, filled with high-temperature computers that require near constant cooling and consume vast amounts of electricity.
The Lawrence Berkeley National Laboratory estimates that US data centres already account for 4.4% of national electricity use, a figure projected to triple to 12% by 2028. Electricity demand from data centres is already the fastest-growing of any sector.
Although AI currently contributes only 5-15% of data centres’ electricity use, it is the primary driver of expected future growth in electricity demand. By the end of the decade, AI workloads are expected to account for 35-50% of all data centre electricity demand.
For grids that still depend on traditional fossil fuels, this increase in demand will only increase greenhouse gas emissions, exacerbating the climate crisis while simultaneously raising costs for average citizens. It’s one of the biggest criticisms of AI: for seemingly little gain, Americans risk both their environment and their wallets. It’s a valid concern – one that’s especially hard to argue when it comes to the families whose homes now live in the shadow of these massive facilities.
But when assessing the overall environmental impact of AI, it’s important to recognise that it can help solve some of the very problems it contributes to.
It’s a claim that’s often met with scepticism: AI utopian technocrats love to claim that a self-realised GenAI system will invent answers to the world’s most pressing issues – climate change included.
What the technologists have wrong is the idea that the climate crisis will be solved with one, single AI-enabled answer; climate isn’t that simple. But AI is already proving useful in tackling some of the most urgent sustainability and engineering challenges.
For example:
- AI Wind Power Forecasting: Google and DeepMind have applied AI to improve the predictability and value of wind energy by forecasting power output 36 hours in advance using neural networks trained on weather and turbine data. This allows wind farms to make more accurate delivery commitments to the grid, boosting reliability and economic value, increasing the value of wind energy by about 20%.
- Smart traffic lights: The City of Pittsburgh rolled out an AI-based traffic signal system that dynamically adjusts light timings to actual traffic flows, reducing idling and shortening commutes. It’s a shift that’s enabled the city to cut vehicle emissions by 21% on average and shortened travel times by ~26% in pilot areas.
- AI-optimised flight routes: Alaska Airlines used an AI platform that analyses real time data such as weather, turbulence and air traffic to recommend safer and more efficient flight paths, saving 4.5M litres of fuel and nearly 12,000t of CO₂ in one year.
These small, AI-enabled shifts are huge steps towards reducing emissions and creating a more sustainable future. But maximising AI’s potential will require especially high-impact sectors like architecture, engineering and construction (AEC) to take responsibility for how they deploy AI.
What AEC companies can do
The AEC industry is uniquely positioned to lead the way in developing and deploying AI solutions to reduce carbon emissions. From infrastructure planning to building design, AEC firms possess the technical expertise needed to tackle complex environmental challenges through innovation.
But using AI effectively means using it responsibly. As AEC companies adopt AI, it is critical to evaluate and manage the environmental impact of these digital tools, focusing on four key pillars: Track, Avoid, Reduce, Balance.
- Track emissions: You can’t reduce what you don’t measure. Tools offered by cloud providers like Microsoft Azure, Google Cloud and AWS can help companies monitor the carbon footprint of their data and AI workloads to set clear reduction targets.
- Avoid unnecessary emissions: Not every task needs a large language model or compute-heavy system. When possible, opt for lightweight models, traditional algorithms, or even non-AI solutions that meet the need with lower energy use.
- Reduce what you can: Choose cloud regions powered by renewable energy and schedule high-intensity workloads during periods of low grid carbon intensity. Cloud providers typically offer these options as part of their solutions.
- Balance your impact: Beyond managing their own footprint, AEC firms can lead by creating AI-driven solutions that reduce emissions. Whether it’s optimising energy use in buildings, reducing material waste in construction, or improving the efficiency of transportation infrastructure, the potential is vast.
AI represents one of the most transformative technologies of our time, with the potential to accelerate climate solutions and redefine what’s possible across industries. But that power comes with responsibility.
To unlock AI’s full benefits without accelerating the climate crisis, we must focus on building and using AI in ways that are not only smart, but also sustainable, ensuring the long-term health of communities, no matter where they are.
- Lalitha Krishnamoorthy PhD is vice president of AI & digital at Stantec
Like what you’ve read? To receive New Civil Engineer’s daily and weekly newsletters click here.
