There’s no denying AI’s warp-speed integration into most digital workflows over the last few years. Its ability to drive efficiency, innovation and competitive differentiation are just a few of its advantages. For MSPs, it transformed customer-facing services through automation, threat identification and predictive maintenance through real-time data collection and analysis. So far, so good.
But as AI grows in scale and power, so does its environmental impact, because while AI may live in the Cloud, its carbon footprint—data centers and their energy and cooling needs—is very much on Planet Earth.
MSPs are prime facilitators as customers take advantage of AI-powered solutions. But, as 100 Spider-Man reboots have told us, with great power comes great responsibility. MSPs now have a unique opportunity to lead from the front and create competitive advantage while defining a new stage of AI—a future where AI is not only powerful, but also more environmentally responsible.
Powering AI
AI and sustainability have… let’s call it a complex relationship! The MSP GLOBAL community already knows the huge advantages it offers in terms of driving efficiency and optimizing resources: energy use, streamlining supply chains, and intelligent management of natural resources, for example.
But AI’s biggest and most immediate problem is its own environmental footprint, spanning raw-material extraction for chips and data centers to operational energy consumption.
AI is HUNGRY!
The manufacturing cycle for AI hardware—especially GPU microchips and specialized accelerators—requires rare-earth elements and critical minerals like lithium, cobalt and nickel, which are often extracted through mining processes that are harmful to both the environment and communities. A 2kg computer requires 800kg of raw materials. Some of those raw materials also serve as geopolitical levers in ways that can be unpredictable. At Cloud-level scale, that’s just not a pretty picture.
Then there’s the energy required by training and running AI models, and how to cool them. In 2023, AI alone consumed 4.5 gigawatts globally – around 8% of all data center energy use. According to the International Energy Agency, electricity demand from AI-optimised data centers is set to quadruple by 2030. And when it comes to water use for cooling data centers, global AI is predicted to account for 6.6 billion cubic meters of water withdrawal by 2027—roughly the same annual water withdrawal of half the UK. And if that wasn’t enough to raise eyebrows, a single ChatGPT question uses 10 times more energy than a Google search (yes, that includes when users send a “thank you” message). With up to a billion prompts a day, that burns through 2.9 million kilowatt-hours of energy every day—around 100,000 times more power than an average US household.
So we see that high performance can mean some seriously high consumption—even more than many may realize. While GPUs offer the parallel processing capabilities that are ideal for AI tasks and can be more energy-efficient per operation compared to CPUs, scaling via GPU clusters can lead to significant total energy consumption. The balance between performance and environmental cost is a critical factor for MSPs to consider when deploying AI services for customers: not just in terms of optics and company ethos, but your actual bottom line.
Why the focus now, and what legislation is coming?
While legislation and regulation differ from region to region, governments and regulatory bodies are increasingly focusing on sustainable technology practices, including those related to AI.
- EU: The European Green Deal and accompanying policies emphasise reducing carbon footprints and improving energy efficiency in technology, in line with broader goals of achieving net-zero emissions by 2050. The Corporate Sustainability Reporting Directive (CSRD) and associated European Sustainability Reporting Standards (ESRS) require companies with activities in Europe to disclose their water use and water resources—a key consideration for MSPs and their customers alike when deploying AI-based tools.
- U.S: In states such as California, stricter energy efficiency standards for data centers and technology infrastructure are being implemented. Industry giants like Amazon, Microsoft and Google are already considering sustainability in their operational models, aiming to be “water positive” by the end of the decade. Google’s DeepMind AI has previously reduced cooling energy use in its data centers by 40%.
- Global: The United Nations Environment Programme (UNEP) has issued recommendations for member states on how to reduce computational complexity and data usage, and adopt green data centers with renewable energy sources. The AI Action Summit in France in February and subsequent Coalition For Sustainable AI, spearheaded in collaboration with UNEP and International Telecommunication Union, aims to align members and supporters on sustainable AI development.
How to build AI sustainability into MSP strategy
MSPs can turn challenge into opportunity by integrating sustainability into not only their AI strategy, but also their wider business strategy. By staying on top of legislation and compliance, anticipating customer needs, and prioritizing sustainable options, MSPs can future-proof their own businesses, give themselves a competitive advantage, and build customer confidence.
Green infrastructure. For most MSPs, developing a green infrastructure is a significant long-term project and investment. But by recommending or partnering with energy-efficient data centers with state-of-the-art cooling technologies and renewable energy sources, MSPs can drive operational efficiency and appeal to clients who are committed to sustainability. And there are some solid options out there: Iron Mountain’s global data centers are powered by 100% renewable energy, while Ark Data Centres in the UK uses advanced cooling technology and runs on 100% renewable energy. Meanwhile, Google aims to power its whole infrastructure with carbon-free energy by 2030. Strategic alliances, potentially with other sustainability-focused MSPs, show forward thinking.
AI models. Several technological tricks and techniques can contribute to energy efficiency in AI deployments. MSPs who develop expertise in this area enable their clients to benefit from high-performing AI while limiting resource consumption.
- Edge devices: Deploying AI models on edge devices, which process data locally and minimise reliance on centralised cloud services, reduces latency and energy consumption associated with data transmission to and from centralised servers.
- Energy-efficient servers: Next-generation servers designed with low-power processors and advanced liquid or evaporative cooling systems can dramatically reduce electricity usage. These servers are optimized for continuous uptime under high-demand workloads while minimizing energy waste.
- Innovative LLM architectures: Instead of trying to directly reduce the power consumption of LLMs, it’s all about running your models more efficiently. Techniques such as model distillation and pruning significantly lower the compute requirements of LLMs without sacrificing performance. By trimming excess parameters and using more efficient architectures, these models consume less energy during both training and rollout.
- Efficient data pipelines: Streamlined data processing architectures ensure that only the data you actually need is processed and stored. Efficient data pipelines reduce unnecessary computational effort, helping to conserve both energy and storage resources.
Consulting and auditing. MSPs can also offer sustainability consulting and auditing to customers. With emerging regulations and rising consumer expectations, many businesses are seeking expert advice on how to transition toward greener AI practices. MSPs can offer tailored consulting services that assess current tech stacks, optimize energy usage, assess the impact of supply chains and guide companies toward sustainability compliance. Partnering with a sustainability-focused MSP is another way to do this.
Digital transformation also entails green transformation
The intersection of AI and sustainability presents a huge challenge—but also a huge and potentially transformative opportunity for MSPs and their customers. MSPs can lead this evolution by integrating innovative, energy-efficient technologies and practices into their business models. Taking it one step further by partnering with specialist sustainability-focused MSPs and engaging with groups or coalitions like the Sustainable AI Coalition is also a serious power move for MSPs who want to show their customers that sustainability is high on their agenda —which can deliver a reputational boost. That way, you as a service provider are still focusing on what you do best, while bringing in a partner expert to do the heavy lifting on sustainability management.
By addressing environmental impacts across the hardware and software lifecycle, aligning with evolving legislation, and embracing next-generation efficiency models, MSPs are in a prime position to usher in a greener, smarter future for the digital world.
Register for MSP Global
Join us in Barcelona this October to get the latest news, meet the people in the know and discover solutions to keep you ahead of the Sustainable AI curve.
