WEKA‑Andromeda and Cerebras‑AMD Announce AI Infrastructure Partnerships

WEKA, an AI data and memory infrastructure company, announced a partnership with Andromeda, a platform that connects AI teams to high‑performance compute. The agreement integrates WEKA’s NeuralMesh storage layer into Andromeda’s managed GPU clusters, delivering consistent, high‑performance storage across more than 50 compute providers worldwide and lowering total innovation costs. Wil Moushey, CEO of Andromeda, said, “Our goal is to enable the global flow of compute…Idle GPUs are throttling the pace of AI innovation.”

Separately, Cerebras and AMD revealed a collaboration to build the world’s fastest disaggregated AI inference solution. AMD’s Helios rack‑scale architecture will handle the compute‑intensive pre‑fill phase, while Cerebras’s Wafer‑Scale Engine provides ultra‑low‑latency decode, delivering up to five times higher tokens‑per‑second‑per‑watt. Julie Choi of Cerebras noted, “It’s a one plus one equals five X in this case.” The joint system targets workloads such as agentic coding, real‑time voice, and multimodal generation, with AMD hardware slated for deployment in Cerebras data centers later this year.