Microsoft has committed to “paying its way” to ensure its data centers will not ramp up residential utility rates, becoming the first major hyperscaler to publicly commit to a comprehensive framework that ties artificial intelligence (AI) data center growth to cost-recovery rate design. The hyperscaler also pledged to advance utility coordination, directly fund grid infrastructure, pursue efficiency improvements, and advocate for accelerated permitting and interconnection.
In a Jan. 13 policy blog outlining what it calls a “Community-First AI Infrastructure” initiative, Microsoft said it will take a more explicit role in managing the local impacts of rapid AI-driven data center expansion, specifically around power, water, workforce development, and community investment.
“Like major buildouts of the past, AI infrastructure is expensive and complex. Investments are advancing at a rapid pace. Today, these require large-scale spending by the private sector in land, construction, electricity, liquid cooling, high-bandwidth connectivity, and operations,” wrote Microsoft Vice Chair and President Brad Smith. “This revives a longstanding question: how can our nation build transformative infrastructure in a way that strengthens, rather than strains, the local communities where it takes root?”
However, the company also said the effort was necessary given mounting grid constraints, aging transmission infrastructure, supply chain shortages, and long permitting timelines. Notably, it explicitly acknowledged that large-scale data center growth will be sustainable only if hyperscalers assume responsibility for the electricity systems and public resources required to support that growth rather than shifting costs or risks to local communities.
“Some have suggested that AI will be so beneficial that the public should help pay for the added electricity the country needs for it. We believe in the benefits AI will create, but we disagree with this approach,” Smith said. “Especially when tech companies are so profitable, we believe that it’s both unfair and politically unrealistic for our industry to ask the public to shoulder added electricity costs for AI. Instead, we believe the long-term success of AI infrastructure requires that tech companies pay their own way for the electricity costs they create.”
A Blueprint to Match AI Data Center Demand with Grid Investment
A key concern for Microsoft—is that the U.S. is grappling with “electricity challenges” while it prepares for a potential tripling of load growth from data centers.
“Much of the country’s electricity transmission infrastructure is more than 40 years old, and it’s under strain. Supply chain constraints on transformers and high-voltage equipment are delaying upgrades that would enable existing lines to deliver more electricity. New transmission can take more than 7 to 10 years due to permitting and siting delays,” his creates a mismatch with growing electricity demand.
The hyperscaler outlined four steps it said will address the “mismatch.” First, Microsoft said it will ask utilities and state commissions to set rates “high enough to cover the electricity costs for our datacenters,” including both energy and the dedicated infrastructure needed to serve them, so “the electricity cost of serving our datacenters is not passed on to residential customers.”
According to the firm, that approach is exemplified in a 2016 Large Power Contract Service tariff that Black Hills Energy and Microsoft co-designed (and the Wyoming Public Service Commission approved) to serve loads above 13 MW with market-based energy and backup generation. In Wisconsin, Microsoft is backing We Energies’ March 2025 proposed “Very Large Customer” tariff, which is now before the Public Service Commission. “[W]e are supporting a new rate structure that would charge ‘Very Large Customers,’ including datacenters, the cost of the electricity required to serve them,” Smith wrote. “This protects residents by preventing those costs from being passed on. But we recognize the need to ensure that datacenter communities benefit everywhere. We believe this approach can and should be a model for other states.”
Second, Microsoft pledged to collaborate “early, closely, and transparently” with local utilities to plan for additional generation, transmission, and substation capacity and to directly fund grid upgrades where its projects drive the need. In the Midcontinent Independent System Operator (MISO) footprint, for example, it noted it has already contracted for 7.9 GW of new electricity supply—more than double its current consumption in that region—to help ensure capacity keeps pace with AI data center load growth.
And, third, the company said it will “pursue innovation to make our datacenters more efficient.” The firm is using AI “reduce energy use and improve the performance of our software and hardware in the design and management of our datacenters,” Smith wrote. “And we are collaborating closely with utilities to leverage tools like AI to improve planning, get more electricity from existing lines and equipment, improve system resilience and durability, and speed the development of new infrastructure, including nuclear energy technologies.”
Among its efficiency efforts is work that spans its Azure fleet, including a Project Forge scheduler that uses machine learning to boost GPU utilization from roughly 50–60% to 80–90% at scale, “safely harvesting” about 800 MW of unused power from existing datac enters since 2019, and retrofitting sites with liquid cooling and custom “sidekick” units to cut cooling energy while increasing rack density and compute per kilowatt-hour.
Fourth, Microsoft committed to push for federal and state policies that “accelerate project permitting and interconnection of electricity projects, expedite the planning and expansion of the electricity grid, and [support] new electricity rates for large electricity users.” The company’s agenda spans priorities it set in 2022, which involved expanding clean generation, modernizing the grid, and engaging communities. But it acknowledged progress has been “uneven” and said “this needs to change” as AI load builds.
Beyond power, the “Community-First AI Infrastructure” framework lays out parallel commitments on water stewardship, workforce and apprenticeship programs near data centers, paying full local property taxes without seeking abatements, and investing in AI training, library-based learning hubs, small-business upskilling, and nonprofit partners in host communities. The measures are needed to ensure AI infrastructure build-out “strengthens—rather than strains—the local communities where it takes root,” Microsoft argued.
A Rare Voluntary Commitment
Microsoft’s move comes one month after Amazon released a company-commissioned analysis that asserts its existing data centers already “fully pay for their own electricity costs” and generate surplus revenue that utilities can reinvest in the grid.
As POWER has reported, the E3 white paper, Tailored for Scale: Designing Electric Rates and Tariffs for Large Loads, found that the Amazon data centers evaluated produce revenues that meet or exceed their marginal cost to serve and are projected to generate about $33,500/MW in surplus value in 2025—which is roughly $3.4 million for a typical 100‑MW facility. However, the potential surplus could reach about $6.1 million per 100‑MW site by 2030, which would create a downward rate pressure on other customers under current tariffs. As POWER previously reported, however, E3’s analysis is scoped to individual facilities and explicitly does not model the broader transmission and new generation build-out that may be required to serve aggregate data center growth across a utility territory.
While both hyperscalers implicitly suggest that hyperscale data centers need to “pay their way,” Microsoft’s commitment goes further, asking regulators to embed that principle in future rate design through bespoke large‑load tariffs, higher prices, and direct grid investments.
Microsoft’s announcement, notably, arrives amid escalating federal and state scrutiny over who should pay for the grid infrastructure required to serve AI data centers. One target is in PJM Interconnection, where capacity auction prices hit record highs in 2025—driven in large part by data center demand—while nearly 40 GW of conventional generation faces retirement by 2030 and interconnection queues remain backlogged.
On Jan. 15, 2026, all 13 regional governors of PJM states, alongside U.S. Secretary of Energy Chris Wright and Secretary of the Interior Doug Burgum, signed a Statement of Principles Regarding PJM to address “ongoing affordability and reliability concerns” in the region.
The statement explicitly commits to push data centers to “cover their share of the costs of any new resources” and calls for triggering a reliability backstop auction by September 2026—a 15-year capacity procurement mechanism—with costs allocated to load-serving entities (LSEs), while directing governors to use their regulatory authority to establish new large-load rate classes that allow LSEs to pass those costs to data centers rather than to residential ratepayers. Data centers that have self-procured power or agreed to curtail during peak demand are exempted from these backstop costs.
The PJM governors’ principles follow major regulatory action by FERC, which on Dec. 18, 2025, unanimously ordered PJM to overhaul its rules for co-located and behind-the-meter large loads. FERC found that PJM’s existing Behind-the-Meter Generation (BTMG) rules—which allowed load-serving entities to net BTMG output against peak demand and reduce transmission charges—violated cost-causation principles by inappropriately shifting costs onto other transmission customers. It directed PJM to propose a new megawatt threshold (stakeholder discussions have centered on roughly 20 MW) below which netting may continue, paired with a transition period for existing customers.
The order also requires PJM to establish four new transmission service options for co-location arrangements, including traditional firm network integration transmission service (NITS), a new interim non-firm service for customers seeking NITS, and new firm and non-firm “contract demand” transmission services that would require customers to deploy special protection schemes to limit energy withdrawals to a specified megawatt quantity, with energy above that contract demand prohibited and the load separately metered from associated generators. PJM must file revised tariff language by February 16, 2026, with a 60-day public comment period to follow, and the order marks a strong signal that FERC will soon adopt nationwide rules for co-location and large-load interconnection, particularly given its pending Advanced Notice of Proposed Rulemaking on large-load interconnection.
Separately, the Department of Energy has urged FERC to expand its jurisdiction over large-load interconnections at the interstate transmission level, citing “unprecedented electricity demand” from data centers and related domestic manufacturing. FERC is expected to act on that proposed rulemaking by April 30, 2026.
—Sonal Patel is a POWER senior editor (@sonalcpatel, @POWERmagazine).
