Microsoft locks in long-term power supply for Indiana data-center expansion, tying electricity availability to AI growth

Microsoft has agreed to a long-term electricity arrangement with NiSource intended to support power needs for an Indiana data-center campus, a deal that spans multiple decades and is scheduled to run through 2048, according to a report syndicated by Yahoo Finance.

The reported agreement is described as a 20-year energy arrangement connected to 1,100 megawatts (MW) of power capacity. Megawatts are a measure of electrical demand capacity, not actual consumption at a specific moment. The figure matters because large data centers can require steady, high-capacity power to run servers and cooling systems.

While the report characterizes the timing as beginning before the power is fully delivered, it does not provide, in the text available here, the exact in-service date, the staged ramp schedule, or the specific generation or delivery mechanism NiSource will use to supply electricity. Those implementation details appear to be omitted from the portion of the announcement summarized in the syndication post.

The report frames the contract as a response to electricity constraints and the speed of artificial intelligence infrastructure deployment. It suggests that data-center buildouts, driven by AI and cloud demand, are moving faster than utility grid additions, pushing hyperscalers to secure long-horizon power commitments.

NiSource is the operating utility and energy infrastructure provider in the deal. The arrangement is presented as a way for Microsoft to lock in future power availability and reduce risk that expanding compute demand will outpace supply, a concern that has increasingly shaped investment decisions in the data-center sector.

Microsoft, as the provider of Azure cloud services and a major user of specialized data-center hardware, has made power procurement and grid access central themes in its infrastructure planning. However, the syndication report does not attribute any specific procurement rationale to a particular Azure region, campus phase, or AI workload category, beyond its general link to AI-driven spending.

For investors and grid stakeholders, the key takeaway is the length and scale of the commitment. A 20-year span through 2048 reduces uncertainty for both parties, but it also means the contract likely depends on long planning horizons for transmission, interconnection, and generation availability. The report, as provided here, does not disclose pricing terms, curtailment provisions, or penalties that could materially affect economics under changing demand or regulatory conditions.

Still, important elements remain unclear from the syndicated post. It does not state whether Microsoft holds options to expand beyond the stated 1,100 MW, whether the capacity is firm or subject to delivery constraints, or how the agreement accounts for potential shifts in load due to efficiency improvements or changes in AI training and inference demand. It also does not specify whether Microsoft’s obligation is to take power directly, pay for capacity, or a combination of both.

The next signposts to watch are whether NiSource publishes additional details on the agreement, including the delivery timeline and how the utility will build or source the required capacity, and whether Microsoft follows up with disclosures tied to cloud infrastructure and regional development. Those specifics will determine how quickly data-center operations can scale and how much financial and operational risk is transferred through the contract.