UALink Consortium
Open scale-up accelerator interconnect vs NVLink
UALink is the open memory-semantic accelerator-to-accelerator interconnect (AMD, Broadcom, Google, Intel, Meta, Microsoft, HPE) explicitly created to be a non-NVIDIA alternative to NVLink for scale-up pods. The UALink 200G 1.0 spec targets up to 1,024 accelerators in a pod and is central to whether the industry can build competitive non-NVIDIA training clusters.
Spec
UALink 200G 1.0
Scale
up to 1,024 accelerators
How it fits the stack
UALink Consortium with what it depends on (above) and what it feeds (below). The figure renders as a crawlable diagram and upgrades to an interactive 3D graph as it scrolls into view.
UALink Consortium in the AI stack. UALink Consortium with its immediate upstream dependencies (top) and downstream dependents (bottom) in the AI value chain. Hover a node in 3D, or read the full relationships below.
Graph data (text) — 4 entities, 3 relationships
- UALink Consortium —supplies→ Broadcom
- UALink Consortium —partners with→ AMD
- UALink Consortium —competes with→ NVLink / NVSwitch
Depends on ↑ · 1
Context — capital, rivals, policy · · 2