🔍 Read the full analysis: How Batteries Could Support Data Centers’ AI Power Needs on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Two October developments highlight approaches to the power constraints facing AI data centers: Vantage reserved more than a gigawatt of factory capacity for Heron Power solid-state transformers, and NiSource agreed to a long-term power deal with Microsoft that includes a 600-megawatt battery. Batteries could buffer rapid changes in demand, while 800-volt DC systems aim to deliver power to dense AI racks with fewer conversion losses. The projects do not establish that these designs will work at scale or lower electricity costs for consumers.
Vantage reserved more than a gigawatt of factory capacity for Heron Power solid-state transformers, while NiSource signed a 20-year power agreement with Microsoft that includes new generation and a 600-megawatt battery, according to a report published October 27. The deals point to a growing role for batteries and direct-current power systems as operators seek to supply energy-intensive AI data centers, though their effects on reliability and electricity bills remain uncertain.
Vantage’s October 21 reservation concerns manufacturing capacity for Heron Power equipment that converts grid electricity to 800-volt direct current. The source report says the transformer includes storage to buffer changes in data center demand. A reservation for factory capacity is not the same as a completed installation; the report does not specify where or when the equipment will be deployed.
Five days earlier, NiSource and Microsoft signed a 20-year power deal that the report says adds about 1,100 megawatts of generation and a 600-megawatt battery. The agreement links new supply and storage to a major technology customer, but the source does not give the project’s location, construction schedule, battery duration or the contract’s cost and allocation details.
The engineering case for 800-volt DC is partly based on vendor estimates. NVIDIA’s reference design converts 13.8-kilovolt grid power to 800-volt DC near a facility’s perimeter, then distributes DC to racks. NVIDIA says the design can use 45% less copper than 415-volt AC distribution and improve end-to-end efficiency by up to 5% compared with current 54-volt systems. Those are company figures, not independent measurements. NVIDIA plans full-scale production with its Kyber systems in 2027.
Infographic 01 · From grid to rack
AI racks near a megawatt are outgrowing the AC power chain.
Why the industry is moving to 800-volt direct current: fewer conversions, less copper, batteries on the main path. The numbers are vendor figures with different baselines, so they cannot be added up.
Today: many conversions
Typical AC chain from the grid to the rack
- Medium-voltage AC from the grid
- Transformer steps it down to low-voltage AC
- UPS converts to DC for the batteries and back to AC
- Distribution units feed the racks
- Every server power supply converts AC to DC again
Each stage loses energy, and almost all of the loss becomes heat the cooling system must remove.
800 V DC: convert once, outside
NVIDIA reference design, full-scale production planned with Kyber racks in 2027
- 13.8 kV AC from the grid
- Converted once to 800 V DC at the edge of the site, batteries on the same DC bus
- 800 V DC busway to the racks
- Stepped down inside the rack, for 1 MW racks and beyond
Bridge for existing halls: a “sidecar” cabinet next to the rack converts AC to 800 V DC, while the old chain stays in place.
Power lost per MW delivered to the rack
kW, SMA white-paper model vendor model
What NVIDIA says 800 V DC changes
NVIDIA’s engineering claims claim
Sources: NVIDIA technical blog (20 May 2025, updated 31 Jul 2025); SMA America, sponsored content in Utility Dive (26 Oct 2026). Rack density: IEA, Key Questions on Energy and AI (Apr 2026) (AI server power density 11× from 2020 to 2025, a further 4× by 2027).
Powered by Thorsten Meyer AI
Batteries Could Ease AI Load Swings
AI data centers place unusually high and concentrated demands on power infrastructure. The International Energy Agency says AI server power density rose 11-fold between 2020 and 2025 and could quadruple again by 2027. It estimates a rack could reach peak power demand comparable to that of 65 households. These projections help explain why operators are examining power delivery and storage alongside computing hardware.
Batteries may help manage short-term fluctuations when computing workloads change quickly, and could supply energy during grid disturbances. The IEA estimates that data centers could have 20–25 gigawatts of batteries by 2030, while saying their value to the wider grid would depend on suitable incentives. Storage is not a substitute for a reliable grid connection or sufficient generation; how much it can help depends on the system design, battery duration and operating arrangements.
The consequences extend beyond technology companies. New generation, wires and storage can affect electricity markets and utility planning, but the cost burden depends on individual contracts and local rules. The source report cites an average U.S. residential electricity price of 18.21 cents per kilowatt-hour in 2026, compared with 17.30 cents in 2025, and says wholesale prices in the PJM region rose 41%. Those figures alone do not show that data centers caused the increases.
Products Worth Considering
Jackery HomePower 3600 Plus Portable Power Station, 3584Wh, 3600W
Anker SOLIX C200 DC Portable Power Station, 192Wh, 200W
Jackery Explorer 300 Portable Power Station, 292Wh
As an affiliate, we earn on qualifying purchases.
From AC Conversion to 800-Volt DC
Traditional data centers receive alternating current, step it down through transformers and other equipment, and convert it to low-voltage DC inside servers. Each conversion loses some energy, much of which becomes heat that cooling systems must remove. Higher rack power can make the equipment, wiring and conversion stages harder to fit and manage.
For a 1-megawatt rack, NVIDIA says a 54-volt design could require up to 200 kilograms of copper busbar. Its proposed 800-volt approach moves conversion closer to the facility’s edge and distributes power at a higher voltage. A Utility Dive article published October 26 described an interim “sidecar” approach, in which a conversion cabinet supports 800-volt racks in a data hall built for AC. That article was sponsored content from inverter maker SMA America, and its modeled energy-loss comparisons should be read as vendor analysis rather than independent test results.
Getting equipment and power to a site can also take time. The IEA puts transformer delivery times at two to three years and grid connections at five to 10 years in many locations. Berkeley Lab reported about 2,300 gigawatts of proposed generation and storage in U.S. interconnection queues at the end of 2024; just 13% of earlier requests were ultimately built. Queue capacity is proposed capacity, not a measure of projects certain to be completed.
“Up to 5% better end-to-end efficiency”
— NVIDIA, in its 800 VDC technical blog
Costs and Deployment Timelines Remain Open
The reported reservation and power agreement do not establish when the equipment, generation or battery will enter service. The source does not provide the project locations, battery duration, contract prices or how the parties will divide the costs of new infrastructure. It is also unclear how much of the planned generation will serve Microsoft’s data center demand at particular times.
Efficiency figures for 800-volt DC systems and modeled comparisons of AC and DC power paths come from companies promoting or supplying the technology. Independent operating data showing performance at large scale are not included in the source material. The proposed systems may also require changes to facility design, safety procedures and existing equipment.
Finally, forecasts of data center electricity use and battery deployments are not confirmed outcomes. The source reports rising residential and regional wholesale prices, but does not establish a direct cause-and-effect link between data center demand and those prices. The local effect will depend on future projects, utility investment and the terms under which customers pay for them.
Watch for Projects Entering Service
The next evidence will come from whether Vantage’s reserved transformer capacity leads to equipment orders and installations, and whether the NiSource-Microsoft agreement progresses into construction and operation. Dates for those milestones were not specified in the source report. NVIDIA’s stated plan is to bring its 800-volt design to full-scale production with Kyber systems in 2027.
As projects develop, useful measures will include verified energy losses, battery performance during rapid load changes, grid connection dates and the costs assigned to data center customers and other ratepayers. Until those details are public, the announced deals show investment and planning—not proof that batteries or DC distribution will resolve the power constraints facing AI expansion.
Key Questions
What did Vantage reserve?
According to the source report, Vantage reserved more than a gigawatt of factory capacity for Heron Power solid-state transformers. The report does not say that more than a gigawatt of equipment has already been installed.
What does the NiSource-Microsoft deal include?
The reported 20-year agreement adds about 1,100 megawatts of generation and a 600-megawatt battery. The source does not give the battery’s duration or project schedule.
How could batteries help an AI data center?
Batteries could provide short-term support when computing demand changes rapidly and help a site respond to grid disturbances. Their actual contribution depends on system design, battery duration and operating arrangements.
Is 800-volt DC proven to be more efficient?
NVIDIA says its proposed system could improve efficiency by up to 5% compared with current 54-volt systems. That is a vendor estimate; the source material provides no independent measurements confirming performance at scale.
Will these projects reduce household electricity bills?
The available information does not establish that. The cost effects depend on project contracts, utility investment and local rules for allocating infrastructure costs. The reported price increases do not by themselves show that data centers caused them.
Source: ThorstenMeyerAI.com
