The Shift to 800V HVDC

In a landmark move for the technology sector, Google, Microsoft, and Nvidia have joined forces through the Open Compute Project (OCP) to establish 800-volt direct current (HVDC) as the new power standard for AI data centers. As artificial intelligence models grow exponentially in size, the power demands of high-density GPU clusters have outpaced traditional alternating current (AC) infrastructure. The 800V DC standard is designed to minimize power conversion losses and maximize efficiency at the rack level.

Technical Details and Efficiency

Transitioning to an 800V DC architecture drastically reduces the number of AC-to-DC conversion stages required within a data center. By delivering power directly at a higher voltage, facilities can achieve unprecedented power density, supporting next-generation AI accelerators that require massive, uninterrupted power draws.

  • Reduction in copper cabling weight and volume by up to 40%.
  • Decreased thermal output from power conversion units, lowering cooling demands.
  • Enhanced compatibility with renewable energy sources and battery energy storage systems.

Industry Impact and Future Outlook

This collaborative standardization effort signals a major shift in how hyperscalers approach data center design. By unifying the power architecture, hardware manufacturers can develop standardized, plug-and-play power shelves, accelerating deployment times for new AI clusters. Furthermore, this open-source approach ensures that smaller enterprises and colocation providers can also adopt the technology, democratizing access to high-performance AI infrastructure. The transition to 800V DC is not merely an incremental upgrade; it is a fundamental rethinking of data center power distribution that will define the next decade of AI innovation.