Tracked topic
A supernode is a set of accelerators wired together tightly enough that software treats them as one very large chip rather than a network of separate ones. This approach is often necessary; a model too big to fit on a single card would otherwise have to shovel data across comparatively slow networking, and that overhead can determine whether trillion-parameter models are practical to serve. One of the best examples is Nvidia's GB200 NVL72, which leverages 72 of its Blackwell GPUs in a single configuration, or AMD's 72-GPU Helios configuration, which offers its own Instinct MI455X GPUs with mor
Rent a supernode by the hour - Alibaba brings frontier-scale AI compute to the public cloud, but only if you live in this remote Chinese province… The underlying Zhenwu M890 processor debuted at the Alibaba Cloud Summit in May 2026, where the company also disclosed a detailed chip roadmap through 2028. comes from T-Head, the group's chip design unit, and carries 144GB of HBM memory with inter-chip bandwidth up to 800 GB/s. …
… Additionally, an apparent subsidiary of Kingsoft Cloud got approval to buy AMD accelerators equivalent to Nvidia's H200, presumably Instinct MI300X-class chips. …
… Wie auch bereits bei früheren Versionen plant Alibaba zudem, die Modellgewichte später öffentlich bereitzustellen. Derzeit steht Qwen 3.8 zunächst als Vorschauversion zur Verfügung. Nutzer können das Modell über Alibabas Cloud-Dienste sowie über ein tokenbasiertes Nutzungssystem ausprobieren. …
… China backs Alibaba's accusations regarding Claude Code, Anthropic Alibaba engineers reverse engineered Claude Code to reveal checks for Chinese system time zones, proxy servers, AI lab infrastructure and network characteristics. …
… The Alibaba i Alibaba Alibaba, founded in 1999 by Chinese entrepreneur Jack Ma, is one of the most prominent global e-commerce companies that operates platforms like AliExpress, Taobao, and Tmall. -backed developer released the software behind its AI model Kimi K3, allowing any government, company,… …
… Inference at this scale depends on extreme co-design across chips, system architecture, and software. …