AI Infra @ Scale 2024: Meta’s Infrastructure Chief on Scaling New AI Frontiers
For the second consecutive year, Meta’s engineering and infrastructure teams took the stage at the AI Infra @ Scale conference. The focus: the technical hurdles of operating AI at massive scale, spanning GPU cluster rollouts, open hardware designs for next-generation data centers, and custom silicon such as the Meta Training and Inference Accelerator (MTIA), which is being used for select AI training workloads.
Opening the 2024 event with a keynote, Aparna Ramani — Meta’s VP of Engineering for AI infrastructure, data infrastructure, and developer infrastructure — outlined the current state of the AI landscape, the core technical bottlenecks, and where open models and purpose-built hardware might push the field next. Her full presentation is available for viewing at the conference site.
