Welcome to LISA
Unless you have been living under a rock, you have probably heard of the growing concerns around the sustainability of AI datacenters. You don’t have to look far to find stories of communities concerned about increasing energy demands, noise pollution, and most famously, AI’s thirst for water. In fact, the U.N released a report that AI could use as much water as 1.3 billion people by 2030.
For those of us in ITAM, the sustainability of AI has been an ongoing topic of conversation. So, when Microsoft CEO, Satya Nadella, walked onto the Build 2026 stage last week, and declared that Microsoft’s newest AI datacenters use “roughly the same amount of water annually as a single restaurant”, many of us breathed a momentary sigh of relief. Satya Nadella’s message was clear: Microsoft has solved the problem. At least, that is the story we have been told.
The announcement is focused on the Fairwater facility in Wisconsin, an AI campus built around a closed-loop cooling system. Unlike traditional cooling systems, a closed loop cooling system is filled once during construction, and then continuously recycled, eliminating the need for ongoing water consumption.
Closed loop cooling is a genuine innovation and there is no doubt that this is an impressive engineering achievement from Microsoft, but the question is whether this announcement represents a meaningful sustainability shift, or a crafted narrative to deflect some of the ongoing criticism.
The uncomfortable truth is that Fairwater is the exception, not the rule. Microsoft have over 500 datacenters across 80 regions, and most of these are using traditional cooling methods, such as evaporative cooling systems – known to use up to 5 million gallons of water per day. Furthermore, the gains made in efficiency, do not offset the volume of new capacity required to meet the demand of AI.
Whilst this is a step in the right direction, the challenge remains that despite plans to both save and replenish water, the absolute water consumption for Microsoft is only going up. This is the unfortunate reality for all the tech giants building datacenters to satisfy AI demand.
To answer the question on whether this is green washing, I’d say, no, but it seems more like a drop in an ever-increasing ocean.
ITAM’s influence has always been at the forefront of asset lifecycle decisions, and optimisation opportunities. This makes for a strong argument that ITAM could be best placed to have a positive impact on sustainability, as well as cost savings, within a business.
For a long time in ITAM, we have noted the correlation between cost savings and sustainability. With growing cost pressures on businesses, it would also be possible for ITAM to align cost-savings with sustainability goals, to optimise the overall benefits.
What does this look like in reality? Well, calculating your Operational AI footprint is likely to become a necessity, especially as AI becomes more ingrained into our technology stack and processes. Tracking AI usage/consumption, along with real time water/energy consumption, is well within ITAM’s grasp, but I think it must go beyond that.
Governance and optimisation are two very important areas of focus within any ITAM function, and there is no reason for that to change when we are talking about AI.
Effective ITAM Leadership can:
We already know that AI poses a business cost risk, with pricing models constantly changing, and a high likelihood of both price/usage increases over time. AI is being built into more and more applications, making it increasingly difficult to track within an organisation. To combat this, many ITAM functions are already putting mechanisms in place to ensure governance around AI. Building on these mechanisms is the key to operational efficiency.
Whilst Microsoft’s announcement does not necessarily fall into the category of green washing, it is far from a solution to the AI datacenter problem. We can only hope that over time, continued efficiency improvements will result in a positive sustainability impact.
Within the ITAM function, however, it is possible to improve efficiency within your organisation, by implementing methods that allow you to measure your operational AI footprint, in addition to other sustainability metrics. By building these calculations into our day-to-day reporting metrics, we can begin to make more informed decisions around efficiency and optimisation of AI within an organisation.
Kelly, LISA Product Manager, is known for transforming complex licensing, cloud economics and governance challenges into clear actionable strategies. With 15 years of ITAM experience spanning numerous vendors such as Microsoft, Oracle, Snow Software, Kelly offers a blend of commercial insight, and creative problem solving to the ITAM community.
