🔍 Read the full analysis: What Batteries Add To The 800-Volt DC Strategy For AI Data Centers on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Two October developments show how AI data-center power planning is expanding beyond chips: Vantage reserved more than 1 GW of capacity for Heron Power’s 800-volt DC solid-state transformer, and NiSource signed a long-term Microsoft power deal that includes a 600 MW battery. The projects point to batteries serving both as backup and as a way to manage fast-changing loads, but their costs, operating arrangements and effects on customer bills remain unclear.
Vantage reserved more than 1 gigawatt of factory capacity for Heron Power’s solid-state transformer, which is designed to convert grid electricity directly to 800-volt direct current, while NiSource signed a 20-year power agreement with Microsoft that includes about 1,100 megawatts of generation and a 600-megawatt battery. The October deals show how operators are pairing new power-conversion systems with storage as AI facilities require denser, more flexible electricity supplies.
The Vantage reservation, dated October 21, is for manufacturing capacity, not a stated delivery of 1 gigawatt of operating power. Heron’s transformer is designed to connect grid electricity to an 800-volt DC distribution system and includes battery storage, according to the source report. The available details do not specify how much storage the Vantage-linked equipment would provide or when it would be installed.
Five days earlier, NiSource and Microsoft signed a 20-year agreement tied to about 1,100 megawatts of generation and a 600 MW battery. The agreement pairs a large power supply with storage, but the report does not say how the battery will be dispatched, whether it is dedicated to Microsoft’s data-center load, or how costs and any grid services will be allocated.
The engineering case for higher-voltage DC is partly based on reducing conversion steps and the amount of copper needed to deliver power to high-demand racks. NVIDIA’s reference design converts 13.8-kilovolt grid power to 800-volt DC at the facility perimeter and targets 1 MW racks from 2027. NVIDIA claims the approach can use 45% less copper than 415-volt AC distribution and achieve up to 5% better end-to-end efficiency than current 54-volt systems. Those are company figures, not independent measurements.
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
Storage for AI’s Fast-Changing Loads
Batteries could help address two separate operating needs: keeping power available through disturbances and responding to rapid changes in AI workloads. The International Energy Agency says large, fast power swings from data centers make storage important. In an 800-volt DC design, a battery could be placed on the main power path rather than used only as conventional backup, although the exact configuration depends on the facility.
The scale of the potential buildout is material. The IEA estimates that data centers could have 20 to 25 gigawatts of batteries by 2030, and says those systems could serve as grid assets if incentives support that use. That is a projection, not a confirmed deployment total. For utilities and customers, the distinction matters: batteries may help manage a data center’s demand, but who pays for them and who benefits from their grid services will depend on project contracts and regulation.
Power delivery is also becoming a constraint on expanding AI capacity. The IEA says transformer deliveries can take two to three years and grid connections five to ten years in many locations. Batteries and new conversion equipment may help sites operate more flexibly, but they do not by themselves create new grid connections or generation.
Products Worth Considering
Anker SOLIX C200 DC Portable Power Station, 192Wh, 200W
Jackery HomePower 3600 Plus Portable Power Station, 3584Wh, 3600W
DJI Osmo Mobile 8 Gimbal with Mic Mini Transmitter, Infinity Black
As an affiliate, we earn on qualifying purchases.
From AC Conversion to 800-Volt DC
Most data centers receive alternating current and convert it several times before servers use low-voltage DC. Each conversion loses some energy, much of which becomes heat that cooling systems must remove. The traditional setup was built for racks drawing tens of kilowatts; the source report says the growing density of AI equipment is pushing designs toward much higher loads.
The IEA’s April 2026 report says AI server power density rose 11-fold between 2020 and 2025 and is projected to quadruple again by 2027. NVIDIA says a 1 MW rack supplied at 54 volts could require up to 200 kilograms of copper busbar. Its proposed 800-volt system aims to move more power through conductors with less copper, with full-scale production planned alongside Kyber systems in 2027. These figures describe reported analysis and a vendor design target, not measured results from a broad operating fleet.
A Utility Dive article dated October 26 describes a possible intermediate step: “sidecar” conversion equipment installed near racks in existing AC data halls. The article was sponsored by SMA America, an inverter maker, and its modeled energy-loss estimates are the company’s case rather than independent field measurements. The approach illustrates that operators may retrofit facilities rather than replace every part of their power infrastructure at once.
“Its design targets “up to 5%” better end-to-end efficiency than current 54-volt systems.”
— NVIDIA, in its 800-volt DC technical material
Project Details Still to Come
The announcements do not establish when Heron’s reserved factory capacity will translate into installed equipment, how much battery storage its transformer product includes, or which Vantage facilities may use it. The NiSource-Microsoft summary likewise leaves open the battery’s duration, ownership, operating rules and relationship to the generation in the agreement.
Broader performance claims also need careful reading. The projected copper and efficiency gains for 800-volt DC come from NVIDIA, while the loss comparison in the SMA-sponsored article comes from a vendor model. The supplied material gives no independent operating data showing how those systems perform across commercial sites. It also does not establish how much of the expected data-center battery capacity by 2030 will be built or connected to grid services.
The effect on household electricity bills is not settled by these deals. The source report cites a U.S. residential average of 18.21 cents per kilowatt-hour in 2026, compared with 17.30 cents in 2025, and a 41% increase in wholesale prices in PJM. Those figures alone do not show how much data-center demand caused price changes or how the costs of particular projects will be shared.
Contracts and Deployments Will Test the Case
The next indicators are whether Heron converts the factory reservation into orders and deliveries, and whether NiSource and Microsoft disclose more about the battery’s operation and cost arrangements. For 800-volt systems, the 2027 target for NVIDIA’s Kyber production will provide a milestone, though product availability will not by itself establish the system’s performance at scale.
Utilities, regulators and data-center operators will also have to determine how new generation, batteries and grid upgrades are funded and scheduled. Those terms will shape whether storage mainly protects a site, helps manage its demand, or supplies services to the wider grid. Until project details and operating results are public, the October agreements show a direction for investment—not proof of the technology’s costs or its effect on electricity prices.
Key Questions
What did Vantage reserve?
Vantage reserved more than 1 gigawatt of factory capacity for Heron Power’s solid-state transformer. The announcement does not mean that 1 gigawatt of equipment is already installed or operating.
What does the NiSource-Microsoft agreement include?
The reported 20-year agreement is tied to about 1,100 MW of generation and a 600 MW battery. The available details do not specify the battery’s duration or how it will be operated.
Why pair batteries with 800-volt DC?
Batteries produce DC, and an 800-volt design may allow storage to connect directly to the main power path. Batteries can also help manage rapid load changes, but the specific design and operating benefits depend on each project.
Are the efficiency and copper savings independently verified?
No independent operating measurements are supplied for the stated savings. The 45% copper reduction and up-to-5% efficiency improvement are NVIDIA claims about its design.
Will these projects lower electricity bills?
The announcements do not establish that. Effects on bills will depend on project costs, power supply, grid upgrades, contracts and how regulators allocate expenses; the supplied figures do not isolate data centers’ effect on household rates.
Source: ThorstenMeyerAI.com
