API
< Back to all clusters
[TECHNOLOGY] · Japan · 11 sources

started · updated

AI infrastructure drives shift toward 800V DC power architectures

The semiconductor and data center industries are undergoing significant shifts driven by the massive power requirements of AI computing. DIGITIMES Intelligence reports an industry-wide transition toward 800V DC power architectures to improve efficiency and power density in high-density AI racks. This shift is reshaping the power semiconductor landscape, with Gallium Nitride (GaN) and Silicon Carbide (SiC) technologies playing critical roles across the power-delivery chain.

To support these evolving infrastructures, Microchip Technology and Navitas Semiconductor have co-developed a reference design for 800V-to-6V DC conversion specifically for AI data center racks. Additionally, Microchip has introduced the PAC1761 and PAC1861 digital power monitors to support 48V power systems used in automotive and AI applications.

Market projections highlight the scale of this growth. Keysight Technologies notes that the global semiconductor market is expected to reach $2.3 trillion by 2030, with AI infrastructure semiconductor sales alone projected to climb from $330 billion in 2025 to $1.2 trillion by 2030. Meanwhile, Hewlett Packard Enterprise (HPE) has unveiled its next-generation ProLiant Gen 13 systems, designed to handle the intense compute and networking demands of enterprise AI workloads.

Entities

DIGITIMES Intelligence · HPE · Keysight Technologies · McKinsey & Company · Microchip Technology · Navitas Semiconductor

Claims

What the coverage asserts, and how many sources carry each claim.

Sources