Skip to main content

Insights / industry / weak_signal

UALink AI Scale-Up Fabric (GPU-to-GPU) — interconnect standard / switch silicon

Updated 6/12/2026

UALink AI Scale-Up Fabric (GPU-to-GPU) — interconnect standard / switch silicon

Companies committing: astera-labs.

Opportunity: NVLink lock-in is the single biggest non-NVIDIA accelerator pain point; the UALink consortium is the answer and Astera is the first merchant silicon making it ship-worthy.

Earliest stage of the bunch — visible only in Astera JDs so far, but the role seniority (Chip Lead Senior Director, Principal program mgr) implies a real product on the roadmap, not a marketing nod. AMD's chiplet+UCIe role is the related die-to-die layer, not UALink itself.

Companies committing

Subscribe to aiinframap Weekly

Weekly AI-infrastructure signals — one email Friday. Engineers only.