🔍 Read the full analysis: Why Data Centers Are Pairing 800-Volt DC With Batteries For AI on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Vantage reserved more than 1 gigawatt of manufacturing capacity for Heron Power solid-state transformers that convert grid electricity to 800-volt DC. Separately, NiSource and Microsoft signed a 20-year power agreement that adds about 1,100 megawatts of generation and a 600-megawatt battery. Together, the developments show growing interest in power systems designed for AI’s dense, rapidly changing loads, though projected efficiency gains are largely vendor claims and grid costs remain a separate issue.
Data center operator Vantage reserved more than 1 gigawatt of factory capacity for solid-state transformers designed to convert grid electricity directly into 800-volt direct current, while utility NiSource agreed to supply Microsoft with new generation and a large battery. The October deals show how operators and power suppliers are pursuing equipment and storage to serve AI facilities’ growing, rapidly changing electricity loads, as conventional data center power systems face limits at higher rack densities, a challenge explored in questions about grid access and capacity.
Vantage’s October 21 reservation is for manufacturing capacity for Heron Power solid-state transformers. The equipment is designed to convert grid power to 800-volt DC and includes energy storage, according to the source report. A capacity reservation is not the same as a completed installation: the report does not specify how many transformers Vantage will deploy, at which sites, or on what schedule.
On October 16, NiSource and Microsoft signed a 20-year power agreement that adds about 1,100 megawatts of generation and a 600-megawatt battery. The agreement links new electricity supply and storage to Microsoft’s data center demand, but the available material does not identify the battery’s duration, project locations, or how the costs will be allocated, issues that reflect the power bottleneck facing AI data centers.
The rationale for 800-volt DC is to reduce the number of power conversions between the grid and computing equipment and to move more power through conductors. NVIDIA’s reference design calls for converting 13.8-kilovolt grid power to 800-volt DC near a facility’s perimeter, then distributing DC to racks. NVIDIA says its design could use 45% less copper than 415-volt AC distribution and deliver up to 5% better end-to-end efficiency than current 54-volt systems. These are vendor estimates, not independent measurements. NVIDIA has said full-scale production with its Kyber systems is planned for 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
Power Design Meets AI Rack Demand
The shift matters because AI facilities are concentrating more electricity demand in each rack, while new grid connections and power equipment can take years to secure. The International Energy Agency said AI server power density rose 11-fold from 2020 to 2025 and is expected to quadruple again by 2027. Its estimate that a rack could reach peak demand comparable to 65 households illustrates the scale, but it is a projection, not a measurement of every planned rack.
At higher power levels, conventional AC distribution can require more copper and leave substantial space for power equipment. Direct-current distribution may reduce conversion stages and allow batteries—also DC devices—to connect more directly to the power system. Storage could help manage rapid load changes or provide backup, depending on system design and operating arrangements. The possible benefits do not establish that every data center will save money or use less electricity overall.
For local electricity customers, the key issue is not only how efficiently a data center uses power, but also who pays for generation, wires, and grid upgrades. A battery at a data center could potentially help the grid, but that outcome depends on how it is operated and on market rules. The source material does not establish whether either October deal will lower local electricity costs.
Products Worth Considering
LG UltraGear 27-Inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDMI 2.1
LG UltraGear 32GX870A-B 32-inch 4K OLED Gaming Monitor
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From AC Conversion to DC Distribution
Most data centers have historically received alternating current, stepped it down through transformers and power systems, and converted it to low-voltage DC near the servers. Each conversion loses some energy as heat. At traditional rack loads of tens of kilowatts, that architecture has been workable; AI racks approaching megawatt-scale demand put more pressure on distribution equipment, space, and conductors.
The IEA estimates global data center electricity consumption will rise from 485 terawatt-hours in 2025 to about 950 terawatt-hours in 2030. Berkeley Lab estimated US data centers used 4.4% of national electricity in 2023 and could account for 6.7% to 12% by 2028. These estimates describe different time periods and geographies; they are not forecasts for a particular facility.
Batteries are also being considered for more than emergency backup. The IEA projects 20 to 25 gigawatts of battery capacity could be installed in data centers by 2030, and says storage could become a grid asset if incentives support that use. Separately, a Utility Dive article published October 26 was sponsored by inverter maker SMA America. Its modeled comparison of AC and 800-volt DC losses and its description of a “sidecar” conversion cabinet should be read as vendor-provided material, not an independent field test.
“Up to 5% better end-to-end efficiency”
— NVIDIA
Costs, Deployment and Grid Effects
The reported Vantage reservation does not establish when Heron transformers will be delivered, where they will be installed, or how much of the reserved manufacturing capacity will become operating equipment. The NiSource-Microsoft agreement’s 20-year term and headline capacity figures are reported, but the available information does not detail construction milestones, battery duration, or the contracts’ cost and risk allocation.
Efficiency and materials savings cited for 800-volt DC come largely from equipment makers’ designs or models. Real-world results will depend on facility layout, conversion equipment, load patterns, cooling, and how existing systems are adapted. It is also unclear how quickly operators will move from pilot or new-build designs to broad deployment, particularly where data halls were built for AC distribution.
Grid constraints remain a separate challenge. The source report cites IEA estimates that transformers can take two to three years to arrive and grid connections can take five to ten years in many places. Berkeley Lab reported about 2,300 gigawatts of capacity in US interconnection queues at the end of 2024, while only 13% of earlier requests were built. Those figures describe queues and historical outcomes, not a guarantee that any specific project will be delayed or canceled.
Tracking Orders, Projects and Standards
The next evidence will be whether Vantage’s reserved manufacturing capacity turns into announced deployments, and whether Heron’s transformers and integrated storage enter service on the reported schedule. For the NiSource-Microsoft agreement, project development and further details on the generation and battery buildout will show how the long-term contract translates into delivered power.
Operators will also need to show how 800-volt DC systems perform outside design models, including their efficiency, reliability, maintenance needs, and compatibility with equipment from different suppliers. NVIDIA has identified 2027 as the planned full-scale production date for its Kyber systems. That timeline is a company plan, not confirmation of broad industry adoption.
As these projects develop, regulators and utilities will face questions about whether storage can provide grid services and how the costs of new generation and network upgrades are assigned. The October deals indicate that power infrastructure is becoming a central part of AI capacity planning, but they do not by themselves resolve connection delays, local rate impacts, or the economics of DC systems.
Key Questions
What happened in the Vantage deal?
Vantage reserved more than 1 gigawatt of factory capacity for Heron Power solid-state transformers designed to convert grid power to 800-volt DC. The reported reservation is not confirmation that all of that capacity has been installed or will be used at a specific site.
Why are data centers considering 800-volt DC?
At very high rack power levels, DC distribution may reduce conversion stages and the amount of copper needed to move electricity. The efficiency and materials improvements cited in this report are vendor estimates and may vary by installation.
What does the NiSource-Microsoft agreement include?
The 20-year agreement adds about 1,100 megawatts of generation and a 600-megawatt battery, according to the source report. The battery’s duration and the projects’ cost allocation were not provided.
Why pair batteries with AI data centers?
Batteries could help manage rapid changes in computing demand, support backup power, or provide services to the grid. Whether they perform those roles depends on their design, operating rules, and available incentives.
Will these projects reduce electricity bills?
The reported agreements do not establish that household bills will fall. The effect on customers depends on local power markets, grid upgrades, contract terms, and how the costs and benefits of new generation and storage are assigned.
Source: ThorstenMeyerAI.com
