🔍 Read the full analysis: AI’s Data Center Buildout Is Changing The Work Of Electricians 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
Recent deals and engineering plans show AI data centers shifting investment beyond chips to power equipment, batteries and grid connections. The buildout could change electrical work inside facilities, but the pace and scale depend on equipment availability, utility connections and designs that are still being tested.
AI data center projects are changing the electrical systems that electricians and power engineers install, as operators plan for higher-density racks, direct-current distribution and large battery systems. The latest signs include Vantage reserving more than 1 gigawatt of factory capacity for a solid-state transformer and NiSource signing a 20-year deal with Microsoft that adds about 1,100 megawatts of generation and a 600-megawatt battery.
The shift reflects the rising power demand inside AI facilities. The International Energy Agency says server power density rose 11-fold from 2020 to 2025 and is projected to quadruple again by 2027. At that level, a single rack could draw as much electricity at peak as 65 households, according to the IEA. That puts pressure on familiar arrangements of transformers, uninterruptible power supplies, distribution equipment and server-level converters.
One proposed design moves more of the conversion work to the edge of a data center and distributes 800-volt direct current to high-power racks. NVIDIA’s reference design describes converting 13.8-kilovolt grid power to 800-volt DC for racks expected to reach 1 megawatt. The company says the 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 engineering claims, not independently verified results.
For electricians, the planned changes point to more work involving high-capacity power conversion, battery integration, protection systems and equipment designed around DC distribution. Some operators may use conversion cabinets beside racks, often described as sidecars, to connect new DC equipment to halls built for AC. That could allow upgrades without replacing every part of an existing electrical system, though the precise work will vary by facility and design.
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
How New Power Designs Change Site Work
The changes matter because a data center’s electrical infrastructure is becoming a limit on how quickly operators can install computing capacity. Higher rack loads mean larger conductors and more demanding power conversion and cooling systems. Moving to higher-voltage DC could reduce the amount of copper required for some configurations, but it also changes where conversion happens and how equipment is connected, protected and maintained.
Batteries are also moving closer to the main power path. They can help manage rapid changes in demand and support a site during grid disturbances. The IEA estimates that data centers could have 20–25 gigawatts of battery capacity by 2030, while saying the systems could become a grid asset if the right incentives are in place. That is a projection, not a confirmed buildout.
The scale of investment also has consequences beyond construction. More generation, substations, transformers and transmission capacity may be needed to serve new campuses. How those costs are allocated between data center operators, utilities and other customers depends on individual contracts and regulatory decisions. The development is relevant to electricians and contractors planning skills and staffing, and to communities watching electricity demand and bills.
Products Worth Considering
Apple MacBook Pro 16.2-inch with M5 Pro, 18-Core CPU, 20-Core GPU, 48GB Memory, 1TB SSD, Nano-Texture Display, Space Black
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
LG 24G414B-B 24-Inch UltraGear Full HD IPS Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From AC Halls to High-Voltage DC
Most data centers have historically received alternating current and converted it through several stages before supplying low-voltage DC to servers. Each conversion loses some energy as heat, which then adds to cooling demand. Existing systems were designed for much lower rack loads than the megawatt-scale equipment now being planned for future AI systems.
NVIDIA says its 800-volt design is intended for its planned Kyber systems, with full-scale production scheduled for 2027. A separate Utility Dive article published October 26 describes possible approaches to DC distribution, including sidecar conversion cabinets. That article was sponsored content by inverter maker SMA America; its modeled estimates should be understood as a vendor’s case, not independent measurements. The broader electricity demand is also growing: the IEA projects global data center use to rise from 485 terawatt-hours in 2025 to 950 terawatt-hours in 2030.
Grid access can take longer than procuring computing equipment. The IEA says transformers can take two to three years to arrive and grid connections can take five to ten years in many places. Berkeley Lab reported that US interconnection queues held about 2,300 gigawatts of proposed generation and storage at the end of 2024, while only 13% of earlier requests were ultimately built. Queue capacity is not the same as projects that will be completed.
“Leaving no room for compute”
— NVIDIA, describing its 800-volt distribution design
Costs, Rollout and Grid Effects
The deployment schedule and real-world performance remain uncertain. NVIDIA’s efficiency, copper and maintenance figures are company estimates, and the supplied material does not report independent field measurements from operating 800-volt AI data centers. It is also unclear how many operators will adopt the design, how quickly electricians will encounter it across projects, or what training and safety procedures will be required at specific sites.
The financial and public effects are also unsettled. The NiSource-Microsoft agreement specifies generation and battery capacity, but the material does not detail the full cost allocation or how the project will affect bills in particular communities. National residential electricity prices and regional wholesale-price movements do not, by themselves, establish that data centers caused a given increase. The relationship between new demand, utility investment and customer charges will differ by location and contract.
Projects Move Toward Deployment
The next milestones are the development of the transformer capacity reserved by Vantage, the planned production timeline for NVIDIA’s Kyber systems in 2027, and the buildout of generation and storage under the NiSource-Microsoft agreement. Utilities and data center operators will also need to secure connections, complete project approvals and decide how new equipment fits into existing facilities.
As installations proceed, independently measured performance, construction timelines and cost allocation will help clarify whether DC distribution and on-site batteries deliver the projected benefits. Until those details emerge, the confirmed development is a shift in investment and design planning—not proof that one electrical architecture has become the standard.
Key Questions
What is changing in data center electrical systems?
Operators are planning for higher-power AI racks, more direct-current distribution and larger battery systems. One proposed design converts grid power to 800-volt DC near the facility perimeter and distributes it to racks.
How could the changes affect electricians?
Projects may require work on power conversion, high-capacity distribution, batteries, protection systems and upgrades that connect new equipment to existing AC halls. The exact work will depend on each site’s design and rollout.
Are the claimed efficiency and copper savings independently verified?
No independent measurements are provided in the source material. The figures for copper use, efficiency and maintenance costs are NVIDIA’s claims about its proposed design.
Why are batteries being added to data centers?
Batteries can help buffer rapid changes in AI power demand and provide support during grid disturbances. The IEA says they could become a grid asset if suitable incentives are established.
Will data center growth raise household electricity bills?
The effect depends on local generation, grid investment, contracts and decisions about who pays for infrastructure. The cited national and regional price figures do not establish that data centers caused a specific change in household bills.
Source: ThorstenMeyerAI.com
