344
Productivity & Workflow355
Automation & Workflow225
Software Development251
Marketing & Growth192
AI Infrastructure & MLOps175
Writing & Content Creation203
Data & Analytics142
Photography & Imaging156
Design & Creative170
Customer Support132
Sales & Outreach125
Voice & Speech135
Education & Learning131
Operations & Admin87
Huawei’s Eric Xu says Chinese AI labs should move faster to catch US leaders, as Huawei lays out new Ascend chip timelines and a bigger AI system plan.
In short: Huawei chair Eric Xu said China’s AI labs should accelerate AI development, even as some US AI companies call for a slowdown, and Huawei shared updated plans for its AI chips.
Eric Xu, Huawei’s rotating chair, told reporters that Chinese AI researchers need to move faster to catch up with leading US labs. He said going faster would also help Chinese teams better understand AI risks, similar to what US labs are seeing. He added that the industry will need to balance progress and safety once those risks become clearer.
His comments contrast with recent calls in Silicon Valley for a coordinated slowdown in AI development. Companies including Anthropic and OpenAI have warned about safety concerns, including the chance that very powerful AI could cause serious harm.
Huawei also used the event to give updates on its AI hardware roadmap. The company is trying to build an alternative to Nvidia’s widely used AI computing platform, at a time when the US is restricting exports of the most advanced AI chips to China.
Huawei said its Ascend 960 DT chip, aimed at training AI models (teaching the system by showing it lots of examples), is planned for the first quarter of 2027. It also plans an inference chip for the third quarter of 2027. Inference means running the model to answer questions (like using a trained student to take a test).
AI progress depends heavily on access to powerful chips and the ability to connect many chips together in data centers, which are like giant computer warehouses. Huawei says its next system could link up to 4,096 Ascend processors, up from 1,024 before, which could help Chinese labs build larger AI models at home. The split between “move faster” and “slow down for safety” signals that governments and companies may take very different paths on how AI should be developed and controlled.
Source: Financial Times