Skip to main content
Partner Podcast

Powering AI at scale

Maury Wood appears on an Electronic Specifier podcast explaining the challenges of powering AI and unveiling the trends toward wafer-scale processors. Listen to the podcast.

Chiplet based processor image

As AI processor demands skyrocket toward 40kW, traditional multiphase TLVR power networks are breaking down from massive thermal losses. Vicor addresses this critical bottleneck with its innovative Factorized Power Architecture™ (FPA™), which minimizes energy waste by running low current at 48V directly to the point-of-load. Their ultra-thin, 1.5mm current multipliers achieve best-in-class current density, enabling efficient vertical power delivery directly beneath the processor. By transforming power delivery from a legacy constraint into an efficiency driver, Vicor provides the scalability needed for future 1MW AI compute racks.

Listen to the podcast to unpack why traditional power architectures are struggling to keep up with AI, and how new approaches like 48V Factorized Power Architecture could reshape the future of AI infrastructure.