🔍 Read the full analysis: How 800-Volt DC And Batteries Are Changing Power For AI Data Centers 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
AI data center power demand is driving interest in 800-volt direct-current distribution and batteries that can buffer rapid load changes. Vantage reserved more than a gigawatt of factory capacity for a solid-state transformer, while NiSource and Microsoft signed a 20-year power deal that includes 1,100 megawatts of generation and a 600-megawatt battery. The deals signal investment, not proof that the newer designs will deliver the efficiency and cost savings vendors project.
Data center power infrastructure is attracting new investment as AI racks draw more electricity: Vantage reserved more than a gigawatt of factory capacity for an 800-volt DC solid-state transformer, and NiSource signed a 20-year power agreement with Microsoft that includes new generation and a large battery. The developments point to a shift in how operators plan to supply and manage power, though projected efficiency gains remain vendor claims and the projects’ broader effects are not yet clear.
Vantage made its reservation on October 21 for factory capacity for a solid-state transformer designed to convert grid power directly to 800-volt direct current, according to ThorstenMeyerAI.com. The report does not give the transformer’s delivery schedule, price or the data center sites where it would be used. A factory-capacity reservation is not the same as a completed installation.
On October 16, Indiana utility NiSource signed a 20-year agreement with Microsoft adding about 1,100 megawatts of generation and a 600-megawatt battery, the source reports. The deal links data center demand to long-term electricity supply and storage. The report does not specify the generation technology, battery duration, project locations or how costs will be allocated among the parties and other ratepayers.
The technical case for DC is that conventional data centers convert electricity several times as it moves from the grid to server components. Each conversion loses some energy, much of which becomes heat. NVIDIA’s reference design calls for converting 13.8-kilovolt grid power to 800-volt DC near a facility’s edge and distributing DC to racks. NVIDIA says the design could 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. These are company projections, not independently verified results from operating facilities.
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
Power Delivery Is Becoming a Bottleneck
The shift matters because increasingly dense AI equipment can require more power in a single rack than older data center layouts were designed to deliver. The International Energy Agency reported that AI server power density rose 11-fold from 2020 to 2025 and projected another fourfold increase by 2027. These figures describe power density and a forecast, not the power use of every facility. The IEA says one rack could then reach peak demand comparable to 65 households.
Higher-voltage DC could reduce the amount of copper and some conversion losses, while batteries could respond to rapid changes in demand and provide backup or grid support. But equipment design alone cannot supply a site: operators also need generation, substations, transformers and grid connections. The IEA puts transformer delivery times at two to three years and says grid connections can take five to ten years in many places. That makes local power access and infrastructure lead times central to data center plans.
Electricity costs are also a public concern, though the effect of a particular data center deal on household bills depends on location, contracts and regulation. The source cites an EIA average U.S. household electricity price of 18.21 cents per kilowatt-hour in 2026, compared with 17.30 cents in 2025, and a 41% rise in wholesale prices in the PJM region. Those figures do not by themselves establish that data centers caused the changes.
Products Worth Considering
Jackery HomePower 1000 Plus v2 Portable Power Station
Jackery Explorer 300 Portable Power Station with 40W Solar Panel
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Server Power to Grid Capacity
Most data centers have historically received alternating current and converted it through transformers, uninterruptible power supplies and distribution equipment before servers convert it to low-voltage DC. At modest rack power levels, that layered arrangement has been practical. As rack demand approaches a megawatt, the conductors and power equipment needed to deliver electricity at low voltage become more demanding.
NVIDIA’s 800-volt design is intended for its future Kyber rack systems, with full-scale production planned for 2027. The company says supplying a 1-megawatt rack at 54 volts could require up to 200 kilograms of copper busbar. SMA America, an inverter maker, has also promoted a “sidecar” conversion cabinet that would let 800-volt racks operate in facilities built around AC distribution. Its October 26 Utility Dive article was sponsored content, and its modeled comparison of energy losses should be treated as a vendor case rather than an independent field test.
Batteries are part of the same infrastructure debate because AI workloads can cause fast changes in a facility’s electricity demand. The IEA has projected 20 to 25 gigawatts of batteries at data centers by 2030. That is a forecast, not installed capacity. The agency said storage could potentially become a grid asset if suitable incentives are in place.
“Leaving no room for compute”
— NVIDIA, describing its 800-volt design
Savings and Local Costs Remain Open
The reported deals show commitments and planning, but do not establish how quickly the equipment and power projects will be built or how they will perform. The source gives no delivery schedule or site details for Vantage’s transformer reservation, and it does not provide key technical details for the NiSource-Microsoft generation and battery projects.
It is also not yet clear whether 800-volt systems will produce the efficiency, copper and maintenance savings vendors forecast when operated at commercial scale. The figures cited for NVIDIA’s design are company claims; SMA’s loss comparison comes from sponsored vendor material. Independent measurements from deployed systems are not included in the source.
Finally, the contract terms do not settle the broader question of who pays for new generation, batteries and grid upgrades. The source notes that costs are being handled contract by contract, but gives no details on the agreements’ rate treatment or effects on local electricity customers.
Projects Must Reach Deployment
The next milestones are project-specific: Vantage’s transformer capacity must translate into manufacturing and installation plans, while NiSource and Microsoft will need to advance the generation and battery projects under their long-term agreement. The source does not report construction dates or a schedule for either development.
For the wider shift, a key test will be whether operators install 800-volt equipment in new data centers or use conversion cabinets to adapt existing halls, and whether independent operating data supports the projected gains. Grid connection timing, transformer supply and decisions about who funds infrastructure will also shape how quickly proposed capacity becomes usable electricity.
Key Questions
What does 800-volt DC mean for a data center?
It means distributing direct current at 800 volts closer to or through data center racks, rather than relying on several stages of AC distribution and conversion. NVIDIA says its planned design could reduce copper use and improve efficiency, but those figures are projections, not independently measured operating results.
What did Vantage and NiSource announce?
Vantage reserved more than one gigawatt of factory capacity for a solid-state transformer intended to convert grid power to 800-volt DC. NiSource signed a 20-year agreement with Microsoft that the source says adds about 1,100 megawatts of generation and a 600-megawatt battery.
Why are batteries being considered for AI data centers?
AI workloads can cause rapid changes in electricity demand. Batteries can store energy and respond to those changes; they may also provide backup or grid support. The scale and economics of those roles depend on project design and incentives.
Are the projected 800-volt efficiency savings proven?
No independent field measurements are included in the source. NVIDIA’s estimates are vendor claims, and the cited SMA comparison comes from sponsored material. Actual results at operating sites remain unclear.
Will these projects lower or raise household electricity bills?
The reported agreements do not establish how local bills will change. The effect depends on the project costs, power contracts, grid upgrades and regulatory treatment; the source does not provide enough information to attribute household price changes to these deals.
Source: ThorstenMeyerAI.com
