Meta to Detail Networking Lessons for Gigawatt-Scale AI Clusters at Hot Interconnects
thoefler · x · 2026-08-14
At the upcoming Hot Interconnects 2026 conference, Meta's network engineering lead will deliver a keynote sharing lessons and challenges learned from networking gigawatt-scale AI compute fleets.
Additionally, the conference agenda covers several cutting-edge interconnect technologies, including:
- Photonic Interconnects: Lightmatter discussing BiDi DWDM as the foundation for high-performance AI training, and scaling inference prefill using high-radix photonic interconnects.
- Ultra-Scale Networking: Multipass random leaf-spine network designs tailored for extreme-scale computing systems.
- Novel Transports: Application-centric Multipath Reliable Connection (MRC), and petabit-class switching breaking the network wall for single-hop MoE training.
More from Infra
- Developer Calls for Bifurcated AI Inference: Cheap Tokens for Background Tasks, Premium for Interactive — jwt0625 · 2026-08-14
- Data centers pay 57% of property taxes, fund $15M aquatic center and sports complex in small Washington town — Chris_Brannigan · 2026-08-14
- Running MiniMax Video Model on RTX 5090 Uses Only 20GB VRAM — BoredHobbes · 2026-08-14
- Mistral AI Pivots to Infrastructure: Plans 1GW Compute, Launches Pre-sale Financing — demian_ai · 2026-08-14
- SK Hynix: US and China to Account for 95% of Global AI Compute Demand by 2027 — nugurimt · 2026-08-14
- AWS Rolls Out Major Updates: DynamoDB Adds Real-Time Vector Search — _jaydeepkarale · 2026-08-14