Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about cutting-edge power delivery solutions for AI accelerators in this conference presentation from the Open Compute Project. Explore how the rapid adoption of AI and machine learning has dramatically increased compute power demands, with modern GPUs exceeding 1000A at sub-0.8V and projections surpassing 4000A within 3-5 years. Discover how this growth follows a new Moore's Law where compute capability scales with power consumption, pushing existing power architectures to their limits. Examine high power density VRM solutions leveraging Vertical Power Delivery (VPD) technology that enables high current density, reduced PDN losses, and improved system efficiency for next-generation AI accelerators. Understand the key challenges of VPD adoption including limited space beneath processors, thermal management requirements, and supporting high currents while maintaining low height profiles. Review essential design principles and see a demonstration of a 1500A VPD hardware platform that showcases the potential impact on future AI accelerator development.
