Data Center Communication Wall

The core challenge in AI computing scenarios. When GPU computing power improves rapidly but network transmission capabilities cannot keep pace, computing resources experience significant idle time, creating the so-called “communication wall” problem. This issue becomes particularly prominent in super-large-scale clusters with over 10,000 GPUs. The Panmai 920 effectively breaks through this bottleneck by enhancing networking capabilities and coordinating with computing chips.

Micron 245TB 6600 ION SSD Deep Review: Pioneering the Storage Revolution in the AI Era

by

As AI training datasets expand exponentially, traditional storage solutions face unprecedented pressure. The capacity density and latency bottlenecks of hard disk drives (HDDs), combined with the per-drive capacity limits of traditional enterprise solid-state drives (SSDs), have become the “Achilles’ heel” constraining AI infrastructure expansion. Against this backdrop, Micron Technology’s 245TB 6600 ION SSD, launched in May … Read more

Alibaba T-Head Expands Chip Portfolio with First Smart Network Card “Panmai 920” for Large-Scale AI Clusters

by

On April 28, 2026, in Fuzhou, China — At the 2026 Digital China Construction Summit, Alibaba’s semiconductor subsidiary T-Head officially unveiled its first smart network card product, “Panmai 920.” Specifically designed for AI computing clusters with over 10,000 GPUs, this new product marks T-Head’s completion of a comprehensive “computing-storage-networking” chip layout. Three Years in the … Read more

Gen DAS Dex: Reshaping Data for Embodied AI Hands

by
Gen DAS Dex

At this critical juncture where embodied intelligence is moving towards the real physical world, data quality has become a core bottleneck restricting model generalization. Gen DAS Dex (hereinafter referred to as Dex), officially released by Jianzhi, takes “head-hand full-modality” as its entry point, attempting to solve the long-standing industry challenge of agile behavior acquisition. one … Read more