Huawei promises to double the power of AI chips every year - ForkLog: cryptocurrencies, AI, singularity, the future

# Huawei Promises to Double AI Chip Power Every Year

Huawei plans to release a new generation of Ascend AI chips annually and double their performance. This was announced by the company's vice president and head of its cloud business in China, Chen Lin.

According to him, the market is hitting a shortage of computing power and rising memory costs. This impacts model training and iterations, increasing expenses. Chen Lin believes the solution should be stable supply of computing resources and a predictable upgrade schedule.

The top manager confirmed the plan of "one generation per year, doubling computing capacity." The Ascend 950DT chip is scheduled to be officially launched in Huawei Cloud's cloud service in August. The company promises a significant increase in vector performance, memory bandwidth, interconnects, and native support for modern formats like FP8.

Huawei claims this will simplify programming and tuning of models, including scenarios for intelligent driving.

Separately, Huawei Cloud's division announced the creation of a global computing and storage network based on its own Ascend chips and a localized supply chain. The plan includes 34 regions and 102 availability zones; large computing clusters are already deployed in Guyana, Uhu, and Inner Mongolia.

The company also outlined goals in the automotive and enterprise sectors. Manufacturers of "smart" vehicles will receive a safer and more reliable cloud infrastructure, as well as sustainable development of computing power. Huawei will continue investing in AI services for digital business transformation.

Recall that in January, Google DeepMind CEO Demis Hassabis suggested that China has almost caught up with the US and the West in AI.