Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about Meta's MGX-OCP system, their deployment of the GB300 AI infrastructure developed in collaboration with Nvidia, in this 15-minute conference talk. Discover how hardware engineers at Meta balanced time-to-market pressures, technical capabilities, and program risks to deliver high-performance AI capacity on schedule. Explore the specific system design choices, development tradeoffs, and considerations that went into adapting existing building blocks for unique Open Compute Project deployment requirements, including integration with ORv3 HPR racks and Rack Management Controller (RMC) for leak handling. Gain insights into the rapid deployment strategies used to implement this next-generation AI system within Meta's OCP data centers, presented by Matt Bowman and Hao Shen, Hardware Engineers at Meta.
