| Mobile Web

Search results for Huawei

AI & Enterprise

DeepSeek unveils development tools for Huawei AI chips, targeting Nvidia CUDA

Chinese AI company DeepSeek is working with Huawei to develop programming tools for Huawei\'s Ascend chips, Reuters reported on Tuesday. The move is seen as targeting Nvidia\'s CUDA software ecosystem. DeepSeek said via its official WeChat account it would open-source programming infrastructure for the Ascend platform, including computing and communications libraries. DeepSeek and Huawei also developed a supernode solution using 128 Ascend 950 chips to boost computing and communications performance.

AI & Enterprise

New AI model emerges as Meta\'s Muse posts early surge

TypeSafe AI, founded by a former OpenAI employee, has put forward a new AI model called Jev that differs structurally from large language models and focuses on making decisions rather than generating text. Experts say it could boost AI agent efficiency, and a report said the company is seeking $1 billion in funding. The article also rounds up new, lower-cost models and agent tools from OpenAI and Anthropic, Meta\'s early traction with Muse, and other moves by global and Korean companies.

AI & Enterprise

DeepSeek bets on training AI models with Huawei chips, plans 8 trillion-parameter model

DeepSeek CEO Liang Wenfeng (량원펑) has told investors that training AI models on domestically made chips is the company’s top priority, according to a report. He expects Huawei to begin supplying training chips as early as the fourth quarter, but did not specify volumes. He called the effort a major bet that must succeed and said Huawei chips could match Nvidia’s within a few years. DeepSeek is training a 2 trillion-parameter model and plans an 8 trillion-parameter model.