🔍 Read the full analysis: Why Electricians Matter In The AI Race For Data Center Power 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
Recent deals by data center operator Vantage and utility NiSource show that AI expansion increasingly depends on power infrastructure as well as chips. The source reports a large transformer-capacity reservation and a long-term Microsoft power agreement that includes generation and battery storage; their wider effects on data center schedules and electricity bills remain uncertain.
Vantage reserved more than a gigawatt of factory capacity for a solid-state transformer, while NiSource signed a 20-year power deal with Microsoft that adds about 1,100 megawatts of generation and a 600-megawatt battery, according to ThorstenMeyerAI.com. The agreements highlight a growing constraint on AI data center expansion: getting enough power equipment and grid capacity to serve facilities with rapidly rising electricity demands.
The source says Vantage’s October 21 reservation is for a solid-state transformer designed to convert grid electricity directly into 800-volt direct current. It does not provide the contract’s price, delivery schedule or the amount of transformer capacity that will be delivered at any particular time. A factory-capacity reservation is not the same as a completed installation or a guarantee that a data center will receive a grid connection.
Five days earlier, NiSource signed a 20-year agreement with Microsoft involving approximately 1,100 megawatts of new generation and a 600-megawatt battery, the source reports. The deal ties new electricity supply and storage to a major technology company, but the account does not specify the generation mix, project locations, construction milestones or how costs will be allocated among Microsoft, the utility and other customers.
The equipment choices reflect the high power demands of AI computing. The International Energy Agency says AI server power density rose 11-fold from 2020 to 2025 and is projected to quadruple again by 2027. NVIDIA’s proposed design would convert power to 800-volt DC at the facility perimeter and distribute it to racks. The company says its approach 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. Those figures are NVIDIA’s vendor claims, 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
Power Equipment Sets AI Expansion Pace
The deals matter because data center growth depends on more than access to processors. Facilities need electrical equipment, a connection to the grid and enough generation to run power-dense computing loads. If transformers or grid hookups arrive later than servers, operators may have equipment ready but be unable to use it at planned capacity.
There may also be consequences beyond the data center. The source cites US Energy Information Administration figures putting the average household electricity price at 18.21 cents per kilowatt-hour in 2026, compared with 17.30 cents in 2025, and says wholesale prices in the PJM region are up 41%. Those figures do not establish that data centers caused the increases. Who pays for new power plants and transmission infrastructure can vary by project and contract, leaving the effect on other electricity customers dependent on local arrangements and regulation.
Storage could help operators manage short-term changes in demand or respond to grid disturbances, but it does not by itself create new generation or guarantee a connection. The practical impact of batteries depends on how they are operated, what services they provide and whether the grid can accommodate the facility.
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
Conventional data centers receive alternating-current power, then use transformers, uninterruptible power supplies and distribution equipment before converting electricity to the lower-voltage direct current used by servers. Each conversion wastes some energy as heat, which cooling systems must remove. At much higher rack power, the equipment and cabling needed for low-voltage distribution can take up valuable space.
NVIDIA has proposed moving conversion closer to the facility’s electrical perimeter and distributing 800-volt DC to high-power racks. The company’s technical material says a 1-megawatt rack supplied at 54 volts could require up to 200 kilograms of copper busbar. NVIDIA plans full-scale production with its Kyber systems in 2027, according to the source. Those are future plans and company estimates, not evidence that the design has already become a standard across the industry.
A Utility Dive article dated October 26, 2026, describes possible migration paths, including conversion cabinets installed beside racks in existing AC data halls. The source identifies that article as sponsored content from inverter maker SMA America. SMA’s comparison of power losses in AC and 800-volt DC systems is therefore a vendor model, not an independent field measurement. The distinction matters when comparing proposed efficiency gains with actual operating results.
“Leaving no room for compute”
— NVIDIA, as described in its 800 VDC technical material
Delivery, Costs and Grid Effects
The source does not provide the Vantage reservation’s price, delivery timetable or final installation plans. It also leaves open when the NiSource-Microsoft generation and battery projects will be built, what energy sources they will use and how their costs will be shared. Those details are needed to judge how much new electricity will become available and when.
The broader figures require care. The IEA’s data center electricity projections and estimates of future battery installations are forecasts, not measurements of completed projects. NVIDIA’s efficiency and copper figures and SMA’s conversion-loss comparison are company or vendor estimates. The source also reports rising household and PJM electricity prices but does not isolate the contribution of data center demand from other influences. The specific effects on local bills, reliability and emissions remain unclear.
Projects and Standards in 2027
The next indicators are project-level: whether the reserved transformer capacity turns into delivered equipment, when the NiSource generation and battery projects reach construction and operation, and how the power agreement assigns costs and supply responsibilities. Grid connection timelines will also shape when new data center capacity can come online.
NVIDIA’s stated target is full-scale production of its 800-volt design with Kyber systems in 2027. Independent operating data would help test the company’s estimates for copper use and efficiency against existing systems. Until projects are built and measured, the October agreements show commitments and plans—not proof of realized savings, lower electricity prices or faster AI data center deployment.
Key Questions
What did Vantage reserve?
According to ThorstenMeyerAI.com, Vantage reserved more than a gigawatt of factory capacity for a solid-state transformer designed to convert grid power to 800-volt DC. The source does not give a delivery schedule or price.
What does the NiSource-Microsoft deal include?
The source reports a 20-year power agreement that adds about 1,100 megawatts of generation and a 600-megawatt battery. It does not specify the generation mix, project locations or cost allocation.
Why are data centers considering 800-volt DC?
High-power AI racks need substantial electricity, and distributing it at higher voltage can reduce the amount of current and cabling required. NVIDIA says its design could reduce copper use and improve efficiency, but those remain vendor claims pending independent operational results.
Will these deals lower household electricity bills?
The available source does not establish that. Electricity prices reflect multiple factors, and the effect of new data center power projects depends on local grid conditions and contract terms, including who pays for generation and infrastructure.
When could the 800-volt design be used at scale?
NVIDIA has said full-scale production with its Kyber systems is planned for 2027. That is a company target, not confirmation of broad adoption or completed installations.
Source: ThorstenMeyerAI.com
