Taipei – October 07, 2026 -- A new DIGITIMES Intelligence report finds that rising rack-level power demands from AI computing are pushing data center operators toward 800VDC power architecture, a shift set to redraw competitive lines across the semiconductor and infrastructure supply chain.
Rack power density is turning electricity delivery into a system-level design problem
As AI accelerators and GPUs scale compute density, DIGITIMES Intelligence notes that power delivery is no longer a facility-level afterthought. Operators and server makers must now jointly optimize conversion efficiency, power density, system footprint, and thermal management across the entire path from grid to chip.
GaN and SiC devices are splitting into distinct roles rather than one winner emerging
The report finds no single power semiconductor technology dominating the 800VDC transition. Instead, gallium nitride (GaN) and silicon carbide (SiC) devices are being deployed at different stages of the power-delivery chain based on voltage, switching frequency, power density, and efficiency requirements specific to each system segment.
Supplier competition is expanding beyond chip specs to system integration capability
DIGITIMES Intelligence expects rivalry among power semiconductor vendors to extend past raw device performance. Ability to co-design with server OEMs and ODMs, participate in system-level architecture decisions, guarantee supply stability, and adapt quickly to shifting AI platform requirements is becoming a decisive competitive factor.
Value distribution across the AI power supply chain is poised to shift
For data center operators, server manufacturers, power semiconductor suppliers, and infrastructure investors, the move to 800VDC extends beyond a voltage change. DIGITIMES frames it as a structural redesign of how power moves through AI infrastructure, with implications for where value accrues across the supply chain as the architecture matures.