🔍 Read the full analysis: What 800-Volt DC Means For AI Data Centers And Their Electricians 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
Vantage reserved more than 1 gigawatt of factory capacity for Heron Power solid-state transformers that deliver grid electricity as 800-volt direct current. The deal, alongside a separate Microsoft-NiSource power agreement, reflects efforts to handle denser AI loads, but performance and savings figures remain largely vendor claims, and grid access remains a major constraint.
Data center operator Vantage reserved more than 1 gigawatt of factory capacity for Heron Power solid-state transformers that convert grid electricity directly to 800-volt direct current, a move aimed at supplying increasingly power-dense AI systems. The reservation, reported October 21, comes as operators examine DC distribution and batteries to manage electricity demand that is outgrowing conventional data center designs, a challenge explored in the AI data center power bottleneck.
The Vantage reservation is a commitment to factory capacity, not confirmation that more than 1 GW of equipment has been installed or is already serving data centers. Heron’s transformer design combines power conversion with energy storage, placing batteries on the facility’s DC power path. The source material does not give a delivery schedule or identify the sites that would use the reserved capacity.
Another development points to the scale of the power capacity challenge facing U.S. data centers. On October 16, Indiana utility NiSource signed a 20-year power agreement with Microsoft that adds about 1,100 megawatts of generation and a 600-megawatt battery. That agreement is separate from Vantage’s transformer reservation; together, the deals show companies pursuing both new electricity supply and changes to how power reaches computing equipment.
Industry designs are moving toward 800-volt DC distribution as AI racks grow more demanding. NVIDIA’s reference design calls for converting 13.8-kilovolt grid power to 800-volt DC near a facility’s perimeter, then distributing it to racks. NVIDIA says the approach could reduce copper use by 45% compared with 415-volt AC distribution and improve end-to-end efficiency by up to 5% against current 54-volt systems. These are company engineering claims, not independent measurements; full-scale production is planned with 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
Why DC Matters for AI Power
Higher voltage can move the same amount of power with less current, reducing the size and weight of conductors needed for very high-demand racks. NVIDIA says a 1-megawatt rack supplied at 54 volts could require up to 200 kilograms of copper busbar. Its proposed design is intended to reduce that burden and the number of conversion stages between the grid and computing hardware.
Reducing conversion losses could also mean less electricity becomes heat that cooling systems must remove. A model cited in an October 26 Utility Dive article, sponsored by inverter maker SMA America, estimates losses of about 80 kilowatts when delivering 1 megawatt through a traditional AC chain, compared with about 20 kilowatts in an 800-volt DC design. Those figures come from a vendor-backed model, not a reported independent test, and real results will depend on equipment and facility design.
The change has consequences for electrical workers and project planners. DC systems require equipment and safety practices suited to their voltage and configuration, while a retrofit may need to connect new DC racks with existing AC infrastructure. SMA describes a “sidecar” conversion cabinet beside a rack as one possible bridge. The source material gives no project-level details about training, staffing or the work practices Vantage customers will adopt.
Products Worth Considering
Jackery HomePower 3600 Plus Portable Power Station, 3584Wh, 3600W
Anker SOLIX C200 DC Portable Power Station, 192Wh, 200W
DJI Osmo Mobile 8 Gimbal with Mic Mini Transmitter, Infinity Black
As an affiliate, we earn on qualifying purchases.
From Server Rooms to Megawatt Racks
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. That architecture was developed for much lower rack demands than those now planned for AI systems.
The International Energy Agency reported that AI server power density rose 11-fold from 2020 to 2025 and is projected to quadruple again by 2027. The IEA says a rack could then have peak demand comparable to 65 households. These figures describe power density and projected peak demand, not average electricity consumption by every rack.
Electricity supply and connections are also limiting factors. The source material cites an IEA estimate that global data center electricity use could rise from 485 terawatt-hours in 2025 to 950 terawatt-hours in 2030. Berkeley Lab data cited there put US grid-connection requests at about 2,300 gigawatts at the end of 2024, while only 13% of earlier requests were built. A queue is not the same as committed construction: many projects may not proceed.
“The company says its design could use 45% less copper and provide up to 5% better end-to-end efficiency than current 54-volt systems.”
— NVIDIA, in its 800 VDC technical blog
Costs, Sites and Field Performance
The Vantage reservation does not establish how many transformers will be delivered, when they will reach operating sites or how much of the reserved capacity will be used. The source material also does not specify the projects, costs or contractual terms associated with Heron’s equipment.
Claims about reduced copper, conversion losses, efficiency and maintenance have not been independently verified in the material provided. The Utility Dive article is sponsored content from SMA America, and its loss estimates should be read as a vendor-backed model. Actual results may vary with the load, equipment and whether a site uses a new DC design or a retrofit.
It is also unclear how costs for new generation, batteries and grid infrastructure will be divided among data center operators, utilities and other customers. The NiSource-Microsoft agreement establishes a long-term power deal, but the source does not provide enough detail to determine how its costs may affect household bills or other ratepayers.
Deliveries and DC Rollouts
The near-term milestones are equipment production, project deployments and the planned 2027 launch of NVIDIA Kyber systems using its 800-volt design. Vantage’s reservation may support future supply, but no installation dates or customer sites are specified in the source material.
Operators and utilities will also need to build or secure generation, storage and grid connections. Whether 800-volt DC becomes common will depend not only on engineering performance but on equipment availability, retrofit costs, safety requirements and power access. More information on these projects and independently measured performance will show whether the proposed gains hold up in operating data centers.
Key Questions
What does 800-volt DC mean for an AI data center?
It means distributing power to high-demand racks at 800 volts of direct current rather than relying on a chain that converts AC power to lower-voltage DC near or inside servers. NVIDIA says its proposed design converts grid power near the facility perimeter.
What did Vantage agree to?
Vantage reserved more than 1 gigawatt of factory capacity for Heron Power solid-state transformers. The reservation is not evidence that the equipment has already been installed or that 1 GW of capacity is operating.
Are the efficiency and copper savings independently verified?
No independent verification is described in the source material. The figures—such as NVIDIA’s claimed 45% copper reduction and up to 5% efficiency improvement—are company claims or vendor-backed estimates.
Why are batteries being considered for AI data centers?
AI workloads can cause rapid changes in electricity demand. Batteries may buffer those swings and support power delivery, but the source material does not establish how much storage will be deployed at specific sites.
Does this mean household electricity bills will rise?
The source material does not establish a direct effect from either deal on household bills. It says the allocation of costs for generation, batteries and grid infrastructure is being determined contract by contract.
Source: ThorstenMeyerAI.com
