DAILY BRIEFING ·
Who pays for Microsoft’s next Indiana power supply?
One new brief published today covers NiSource’s October 16 announcement. Two briefs from October 18 provide recent context. These source dates are distinct from this edition’s publication date.
Microsoft power deal adds 1,100 MW to NiSource’s Indiana plan
The plan
NiSource announced a 20-year electricity agreement on October 16 for Microsoft’s planned campus in northwest Indiana. Service is expected to start in 2028. NiSource says Microsoft will pay for the power-generation equipment and transmission infrastructure needed to serve its demand.
The company’s GenCo business plans to expand its shared supply pool by about 1,100 megawatts, or MW. GenCo also contracted 600 MW of battery storage from an AGP affiliate for a pool serving several large customers. These are planned resources, not evidence that the campus is already operating.
The practical meaning
A megawatt describes a rate of power delivery, not how long that power lasts. Think of the distinction between how fast water flows from a tap and how much water a tank holds. The battery’s 600 MW rating does not tell us how many hours it can supply electricity; the announcement gives no storage duration.
The supply pool is shared. Its expansion should not be read as a measurement of electricity used by Microsoft’s computing equipment alone.
The local picture
For existing NIPSCO electricity customers, the important question is who pays for the new infrastructure. NiSource says the arrangement protects them from those added costs. It raises projected savings across its data-center contracts from $1.4 billion to about $1.75 billion over their lifetimes, estimating as much as $140 per customer annually. Those are company forecasts, not guaranteed bill reductions or savings already received.
Implications for AI
The agreement connects Microsoft’s planned campus with long-term power arrangements rather than just a promise of future electricity. But signing a contract is not the same as having the equipment ready. NiSource identifies regulatory approvals, construction, financing and customers’ building plans among the risks to its expectations.
Milestones to follow
NiSource says it will provide more operational and financial detail on its third-quarter earnings call. Watch for those details and progress toward the expected 2028 service date.
MHI and Ingeteam target the quality of AI power, not just its supply
The development
Mitsubishi Heavy Industries, or MHI, and Ingeteam announced a partnership on October 15 aimed at U.S. data centers. They signed a memorandum of understanding: an initial agreement to work together, not an announcement that a finished system has entered service.
MHI brings expertise in combining power and cooling equipment into a working system. Ingeteam supplies equipment and software that convert and control electricity. The companies plan to make Ingeteam’s products more widely available through MHI’s group and explore combining them into larger energy systems.
How it works
The announcement distinguishes having enough electricity from delivering it in a form equipment can reliably use. AI computing can change its power demand quickly. The companies say these changing demands create electrical-quality problems that can affect equipment reliability and working life.
Think of ordering a delivery: the amount arriving matters, but so does whether it arrives in a usable condition. Here the companies want to coordinate power equipment and controls, not simply promise a larger electricity supply. Their announcement provides no measured performance improvement for a named customer installation.
What changes for nearby people
The companies link their work to keeping the electricity grid stable while data centers grow. People and businesses share that power system, but any effect on their bills or outages remains unclear. It also gives no local job count, water-use figure or named community project from which to judge those effects.
The AI connection
The proposed combination could help operators manage more power-hungry computing equipment as a complete system. The companies describe reliability, efficiency and expansion as goals. They are also considering future work on electricity distribution and energy-management systems, but those options still need technical and commercial evaluation.
Next steps
Watch for specific products, customer projects and published performance measurements. This is an initial U.S.-focused agreement; the announcement does not give a completed deployment’s size or an opening date.
Siemens backs supercapacitors to soften AI power swings
The news
Siemens announced an investment and exclusive partnership with Skeleton Technologies on October 14. It plans to add Skeleton’s supercapacitors—devices for handling rapid changes in electrical power—to its data-center equipment offering. Siemens receives exclusive rights to sell and expand the products in the data-center market.
The companies are targeting existing facilities as well as new AI installations. Siemens says the equipment can be added to existing electrical systems rather than requiring a complete facility redesign. The investment amount remains undisclosed.
Inside the technology
Siemens describes thousands of AI chips changing their electricity demand together. Their demand can rise or fall in a tiny fraction of a second and combine into megawatt-scale changes within seconds. Megawatts measure how fast electricity is being used. That is a problem of sudden changes, not just the total amount of electricity consumed.
The company compares the storage equipment to a shock absorber: it is intended to soften abrupt electrical swings before they disturb equipment or the grid. That comparison describes its proposed role, not a promise that every power problem disappears. The release presents expected benefits, not measured results from a named customer installation.
Costs and community effects
Siemens says the approach is designed to reduce disturbances on the utility side of a data center’s electricity connection. That could matter to the wider power system, but the announcement does not quantify a benefit for nearby households. Household bill savings, fewer neighborhood outages and local job gains remain unproven.
What AI operators gain
The aim is to fit more computing capability into facilities and expand faster while limiting major upgrades. Being able to add equipment to existing buildings could help operators avoid a complete redesign. Those are Siemens’ stated goals; the announcement does not demonstrate a particular increase in computing capacity or a guaranteed construction-time saving.
What comes next
Completion still depends on the usual closing formalities. Watch for completed customer installations and published measurements showing how well the equipment handles rapid demand changes.
Sources linked below.
NiSource company announcement · MHI and Ingeteam joint announcement · Siemens company announcement