Infineon Unveils Gallium Nitride Converters for AI Power
New converter designs achieve high efficiency for 800-volt AI data center power systems.
Updated on Oct. 21, 2026 in Data Centers
Live Poll
Do you expect new AI data center power technologies to lower energy costs for your household?
Infineon Technologies has introduced new gallium nitride converters designed to manage 800-volt electricity for AI data center infrastructure. The hardware provides efficient voltage reduction to support the high power demands of next-generation processors.
Why it matters
Higher-voltage power distribution reduces current levels, which helps lower the physical demand placed on electrical conductors. This efficiency is critical as companies prepare for significantly increased power loads in data center environments.
The devices support switching frequencies up to one megahertz and can reduce 800-volt input to 50, 12, or six volts. The 800-to-50-volt design achieves 98% efficiency, while the 800-to-12-volt version reaches 97.1% efficiency.
The players
Infineon Technologies
This German semiconductor manufacturer specializes in power electronics and sensor systems for automotive and industrial applications.
NVIDIA
This technology company is a leader in the design and manufacture of graphics processing units and hardware for artificial intelligence data centers.
This multinational technology company focuses on internet services, cloud computing, and massive data center operations.
Microsoft
This technology corporation develops software, services, and cloud computing infrastructure, including significant AI data center investments.
The details
The new converters lower high-voltage input before power reaches AI processors, using high-frequency switching to allow for smaller, more compact electrical components. Infineon is collaborating with NVIDIA and other industry leaders on these power chain components through the Open Compute Project.
Timeline
May 2025: Infineon and NVIDIA began collaborating on power chain components.
March 2026: Reference designs for the converters were released.
October 21, 2026: Infineon officially highlighted the gallium nitride converter technology.
2027: NVIDIA targets the deployment of power racks drawing one megawatt or more.
The Tech Race
This development marks a shift toward standardized high-voltage architectures necessary to sustain the rapidly growing energy demands of artificial intelligence hardware. It positions these power components to replace legacy distribution methods as the industry moves toward one-megawatt racks.
While these components are internal to massive AI infrastructure, they enable the continued scaling of cloud-based AI services by reducing energy waste. Users can expect more efficient, high-performance computing as data centers modernize their power chains.
The takeaway
The transition to 800-volt systems represents a necessary evolution in hardware design to handle the intense power requirements of modern AI. Businesses should anticipate that data center operators will increasingly prioritize high-frequency, high-efficiency power management to maintain capacity.
What happens next
NVIDIA is scheduled to begin the deployment of one-megawatt power racks starting in 2027.
Further reading
Learn more about the latest innovations in Data Centers infrastructure.
Live Poll
Do you expect new AI data center power technologies to lower energy costs for your household?
