Navitas Developed New 24 kW AI Power Board
The semiconductor company created a high-efficiency power delivery board to optimize infrastructure for AI data centers.
Updated on Oct. 12, 2026 in Semiconductors
Live Poll
Do you believe rapid improvements in AI infrastructure efficiency will lead to a more sustainable future?
Navitas Semiconductor has developed a new 24 kW power delivery board designed to streamline energy distribution in AI data centers. The architecture converts 800 VDC directly to 12 VDC, aiming to increase system efficiency and reduce physical hardware requirements.
Why it matters
This design aims to boost system efficiency and significantly reduce board space in rapidly expanding AI infrastructure. By eliminating the 50 V intermediate bus converter stage, the system simplifies power delivery for voltage regulator modules.
The board achieves a power density exceeding 2,000 W/in³ using 16 GaNFast FETs with 650 V ratings and 18 mΩ resistance. It features a compact 400 mm by 60 mm form factor.
The players
Navitas Semiconductor
A company headquartered in Torrance, California, that specializes in gallium nitride power integrated circuits.
The details
The board utilizes a stacked half-bridge topology to perform single-stage power conversion. By moving directly from high-voltage input to low-voltage output, the system decreases energy loss and footprint compared to legacy data center power units.
Timeline
Navitas Semiconductor announced the power delivery board on October 12, 2026.
The board will be featured at the OCP Global Summit 2026.
The Tech Race
This development represents a major technological shift by replacing the traditional 50 V intermediate bus converter stage with a direct conversion architecture. It positions Navitas to capture greater market share as data centers scramble to improve power density for high-performance AI chips.
While this hardware targets industrial data centers, its improved efficiency helps manage the massive energy demands of modern AI tools. Lower energy loss at the server level ultimately supports the scalability of cloud services and AI applications used by consumers.
The takeaway
Advancements in power architecture are essential for keeping pace with the soaring energy consumption of modern artificial intelligence hardware. Reducing power conversion stages allows for more compact, cooler-running systems that can handle increased compute loads.
What happens next
The power delivery board is scheduled to be showcased at the OCP Global Summit in 2026.
Further reading
Learn more about the latest innovations in Semiconductors.
Live Poll
Do you believe rapid improvements in AI infrastructure efficiency will lead to a more sustainable future?
