ST and NVIDIA Debut 800V Converters to Power AI Infrastructure
STMicroelectronics on March 17, 2026, announced an expansion of its power conversion portfolio for AI data centers, rolling out new 12-volt and 6-volt architectures for 800-volt direct current (DC) systems. Developed in collaboration with NVIDIA and highlighted at the NVIDIA GTC 2026 conference, these solutions are designed to address the escalating power requirements of next-generation AI hardware.
The new offerings complement ST's existing 800V to 50V solutions, creating a complete portfolio for power distribution inside gigawatt-scale computing facilities. "As AI infrastructure compute scale continues to expand fast, it requires higher voltage distribution and greater density, which can only be achieved with system-level innovation," said Marco Cassis, a key executive at STMicroelectronics, emphasizing the need for more efficient power architectures.
New Design Slashes Energy Loss by up to 12%
The collaboration's core innovation is the development of direct 800V to 12V and 800V to 6V conversion paths. This architecture eliminates the traditional 54V intermediate conversion stage, directly reducing system-level energy losses and the amount of copper wiring required. The design moves the power bus closer to the GPU, minimizing resistive losses and improving transient performance, which is critical for large-scale AI training clusters.
This technological shift has significant financial implications for data center operators. Independent industry analysis indicates that moving to 800V DC power can reduce annual energy-related operating expenses by 8% to 12%. Furthermore, it can generate capital expenditure savings of $4 million to $8 million for every 10 megawatts of new capacity by simplifying power delivery systems. This initiative builds on ST's prior success, including a GaN-based converter introduced in October 2025 that achieved over 98% efficiency.
Broader Partnership Signals Deeper AI Integration
This power management collaboration is part of a wider strategic alignment between STMicroelectronics and NVIDIA. On March 16, the companies also revealed plans to accelerate the development of "Physical AI," including humanoid robots. ST is integrating its sensors and actuators with the NVIDIA Holoscan Sensor Bridge and developing high-fidelity component models for the NVIDIA Isaac Sim robotics platform.
By providing a unified foundation for building, simulating, and deploying physical AI systems, the partnership aims to streamline development from algorithm creation to hardware integration. This deeper integration across both data center power and robotics hardware positions ST as a critical supplier within NVIDIA's expanding AI ecosystem, signaling a long-term commitment to solving fundamental challenges in scaling artificial intelligence.
Edgen Stock·Mar 17 2026, 07:20
