Microsoft Commits to Cover Full Power Costs for AI Data Centers
Key Points
- Microsoft will ask utilities to set electricity rates that fully cover AI data‑center power costs.
- The company supports a new rate structure for "Very Large Customers" in Wisconsin.
- Microsoft argues that it is unfair for the public to shoulder added AI electricity costs.
- A pledge to improve data‑center water‑use intensity by 40% by 2030.
- New closed‑loop cooling design eliminates the need for potable water and is already in Wisconsin and Georgia.
- Microsoft will pay its full share of local property taxes.
- Commitments are targeted to be realized in the first half of 2026.
Microsoft announced that it will ask utilities and public commissions to set electricity rates high enough to cover the full power costs of its AI data centers. The company also pledged a 40 percent improvement in water‑use intensity by 2030, highlighted a new closed‑loop cooling design that eliminates the need for potable water, and affirmed it will pay its full share of local property taxes. These commitments are aimed at addressing rising residential electricity rates and the environmental impact of AI workloads.
Microsoft’s Plan to Fully Fund Power for AI Data Centers
In a recent blog post, Microsoft detailed a strategy to ensure that the electricity required for its AI data centers is fully covered by the company rather than shifting costs to the public. The firm will work with utilities and public commissions to set rates that are sufficient to pay for the full electricity consumption of its data centers, including any infrastructure additions needed to support the load.
Specifically, Microsoft is backing a new rate structure in Wisconsin that would apply to "Very Large Customers," a category that includes its AI data centers. Under this structure, the rate would be set high enough to cover the total electricity required to serve those customers.
Stance on Public Funding and Residential Rates
The company emphasized that while some have suggested the public should help pay for the added electricity needed for AI, Microsoft believes that, given the profitability of tech firms, it would be unfair and politically unrealistic for the industry to ask the public to shoulder those costs.
Microsoft also noted that residential electricity rates have risen in dozens of states, driven in part by inflation, supply‑chain constraints, and grid upgrades. The company expressed concern that communities value new jobs and property‑tax revenue but do not want higher power bills or tighter water supplies.
Water‑Use Reductions and New Cooling Technology
Beyond electricity costs, Microsoft pledged to improve the water‑use intensity of its data centers by 40 percent by 2030. To illustrate the scale of water consumption in AI, the blog referenced an environmental audit of AI model‑maker Mistral, which found that training and running its Large 2 model over 18 months evaporated enough water to fill 112 Olympic‑size swimming pools.
In response, Microsoft has launched a new AI data center design that uses a closed‑loop cooling system. This system constantly recirculates cooling liquid, dramatically cutting water usage and eliminating the need for potable water in cooling. The design has already been deployed in Wisconsin and Georgia.
Property‑Tax Commitment
Microsoft also reaffirmed that it will not ask local municipalities to reduce their property‑tax rates. The company will pay its full share of local property taxes, aligning with its broader commitment to bring these initiatives to life in the first half of 2026.
Overall, Microsoft’s announcements reflect a comprehensive approach to address both the power and water demands of AI data centers while ensuring that the financial burden does not fall on the public.