Pranav Hotkar
Geneva, Switzerland – June 2, 2026 - STMicroelectronics has expanded its 800 VDC power conversion portfolio for AI data centers, introducing new high-voltage power technologies designed to improve efficiency, scalability, and power delivery for next-generation AI infrastructure.
The company announced a new 12V power conversion solution and additional technologies aimed at supporting the growing electricity demands of hyperscale AI clusters and high-density GPU deployments. STMicroelectronics said the products are specifically engineered for emerging 800 VDC architectures that are increasingly being adopted for future AI data center designs.
According to the company, traditional power distribution systems are becoming less efficient as AI workloads push rack densities and energy consumption to unprecedented levels. By moving toward higher-voltage DC distribution, operators can reduce transmission losses, improve energy efficiency, and simplify power delivery across large-scale AI environments.
STMicroelectronics said the expanded portfolio was developed in collaboration with NVIDIA and targets the next generation of AI factories and hyperscale compute facilities. The solutions are designed to support advanced GPU systems requiring increasingly sophisticated power conversion and voltage regulation technologies.
“AI data centers are driving a major transformation in power infrastructure,” STMicroelectronics said in the announcement, noting that higher rack densities and larger GPU clusters are accelerating demand for more efficient electrical architectures.
The company’s new technologies include high-efficiency silicon carbide and gallium nitride power solutions designed to operate within 800 VDC environments. STMicroelectronics said these systems can help reduce energy losses while supporting the high-current demands associated with AI training and inference workloads.
The move reflects a broader industry shift toward next-generation power architectures as hyperscalers and AI infrastructure providers seek to improve operational efficiency and manage rapidly rising electricity consumption. AI training clusters powered by advanced accelerators are increasingly requiring significantly higher rack-level power densities than conventional cloud infrastructure.
Industry analysts view 800 VDC distribution as a potential long-term evolution for hyperscale AI facilities, particularly as operators move toward gigawatt-scale campuses and larger liquid-cooled deployments. Higher-voltage architectures can reduce the amount of copper cabling required while improving overall system efficiency and power scalability.
STMicroelectronics said its expanded portfolio is intended to support a broad range of AI infrastructure applications, including hyperscale data centers, accelerated computing platforms, and next-generation networking systems.
The announcement further highlights how semiconductor manufacturers are increasingly positioning themselves within the AI infrastructure ecosystem beyond compute processors alone. As AI-driven electricity demand continues to rise globally, power conversion and energy efficiency technologies are becoming critical components of future data center design.
With AI infrastructure rapidly scaling worldwide, companies developing advanced power management systems are expected to play an increasingly important role in enabling the next generation of hyperscale AI deployments.
