Multi‑Gigawatt Maia Inference Accelerators Shift Scale‑Up/Scale‑Out to a Single Ethernet Fabric2 min read·4h ago·via InfoQ AI/ML/Data