Huawei Pangu Pro Reportedly Trains 505 Billion Parameters Without Nvidia 报道称华为盘古Pro在无Nvidia条件下训练5050亿参数
A Tech Times report says Huawei's Pangu Pro trained a 505-billion-parameter model without Nvidia hardware, while questioning what the underlying supply chain implies. The claim matters because it turns attention to the practical availability of domestic accelerators, networking, and memory rather than model size alone. Tech Times报道称，华为盘古Pro在未使用Nvidia硬件的条件下训练了5050亿参数模型，同时也追问其背后的供应链含义。该消息的关键不只在模型规模，更在于国产加速器、网络与存储能否获得稳定供给。