Navitas Semiconductor and Microchip Technology have jointly developed an 800V DC-to-6V DC reference design for AI data-center rack power, collapsing two conversion stages into a single direct converter in a bid to raise end-to-end efficiency for next-generation computing infrastructure.
The collaboration, announced on October 5, pairs Microchip's digital power control and security technology with Navitas' GaNFast gallium nitride power devices, aligned with the Open Compute Project's 800V DC standard.
Single-stage conversion for higher efficiency
Conventional rack power architectures step voltage down from 800V to 50V and then from 50V to 6V. The new reference design merges both stages into one direct converter, cutting conversion losses.
The power board uses 16 NV6034 650V, 17-milliohm GaNFast FETs in a stacked half-bridge topology, targeting up to 96% peak efficiency at full load, a 1 MHz switching frequency, and a power density of 2,100 watts per cubic inch.
Security built in
Core components include Microchip's dsPIC33AK digital signal controllers and its TA100 CryptoAuthentication device. The TA100 supports secure boot, trusted firmware updates, TLS and root-of-trust operations, while the dsPIC33AK supports post-quantum cryptography aligned to the CNSA Suite 2.0 algorithms.
Joe Thomsen, corporate vice president of Microchip's digital signal controller business unit, said AI infrastructure optimization is driving one of the most significant power architecture transitions the data center industry has experienced in decades. Vipin Bothra, Navitas' vice president of global solution marketing, called the company's GaNFast technology a critical enabler of higher power density, greater efficiency and improved system performance as AI infrastructure scales.
The board is roughly 20% thinner than a mobile phone, enabling close integration with GPU boards. The reference design, with its board, software and documentation, will be showcased at Microchip's OCP event in San Jose, California, from October 12 to 15.
Sources:
TimesTech: Navitas and Microchip collaborate on 800V AI data center design
BitcoinVersus: Microchip-Navitas 800V-to-6V AI rack power collaboration
The collaboration, announced on October 5, pairs Microchip's digital power control and security technology with Navitas' GaNFast gallium nitride power devices, aligned with the Open Compute Project's 800V DC standard.
Single-stage conversion for higher efficiency
Conventional rack power architectures step voltage down from 800V to 50V and then from 50V to 6V. The new reference design merges both stages into one direct converter, cutting conversion losses.
The power board uses 16 NV6034 650V, 17-milliohm GaNFast FETs in a stacked half-bridge topology, targeting up to 96% peak efficiency at full load, a 1 MHz switching frequency, and a power density of 2,100 watts per cubic inch.
Security built in
Core components include Microchip's dsPIC33AK digital signal controllers and its TA100 CryptoAuthentication device. The TA100 supports secure boot, trusted firmware updates, TLS and root-of-trust operations, while the dsPIC33AK supports post-quantum cryptography aligned to the CNSA Suite 2.0 algorithms.
Joe Thomsen, corporate vice president of Microchip's digital signal controller business unit, said AI infrastructure optimization is driving one of the most significant power architecture transitions the data center industry has experienced in decades. Vipin Bothra, Navitas' vice president of global solution marketing, called the company's GaNFast technology a critical enabler of higher power density, greater efficiency and improved system performance as AI infrastructure scales.
The board is roughly 20% thinner than a mobile phone, enabling close integration with GPU boards. The reference design, with its board, software and documentation, will be showcased at Microchip's OCP event in San Jose, California, from October 12 to 15.
Sources:
TimesTech: Navitas and Microchip collaborate on 800V AI data center design
BitcoinVersus: Microchip-Navitas 800V-to-6V AI rack power collaboration
