Technology
AI boom drives adoption of solid-state power transformers to upgrade aging grids
Energy-intensive data centers have become the primary catalyst for modernizing electrical infrastructure with semiconductor-based transformer technology that replaces 150-year-old designs.
Ars Technica
The explosive growth of AI data centers is straining US electrical grids and creating an unexpected opportunity: accelerating development of solid-state power transformers that could fundamentally reshape how electricity reaches both large facilities and everyday consumers.
The bottleneck holding back grid upgrades#
America's power infrastructure faces mounting pressure to meet surging electricity demand driven largely by massive AI data centers. Today's transformers, which adjust voltage levels as electricity moves across distribution networks, rely on designs essentially unchanged since the 1880s. These devices consist of copper wire coils manually wound around steel cores, custom-built by hand for each utility company's specific needs.
The hand-assembly requirement means new transformers cannot be mass-produced. Utility companies now face wait times of several years to receive upgraded equipment, creating a dual problem: expansion of the grid slows to a crawl, and replacement of failing equipment gets delayed indefinitely.
How semiconductor-based transformers work#
Solid-state transformers take a fundamentally different approach, using high-frequency semiconductor switching to convert between voltage levels. Built from materials like silicon carbide, they can be manufactured at scale using industrial processes borrowed from the semiconductor industry. The advantages extend beyond production speed: these devices are physically smaller, lighter, and modular—allowing utilities to upgrade or swap individual components rather than replace an entire unit.
Perhaps most importantly, solid-state transformers can perform multiple functions as all-in-one power electronic systems, potentially eliminating the need for separate conversion equipment.
Why AI data centers are the catalyst#
Tech companies constructing the largest AI data centers have already begun adopting direct current (DC) power architectures to efficiently deliver power to racks of energy-hungry AI processors. Solid-state transformers can directly convert the alternating current (AC) power supplied by local grids into the DC power these facilities require—a significant advantage over conventional transformers, which would necessitate an additional conversion device.
