Products10:11
Chinese content华为 FreeClip 2 潮玩拼色版开启预售，售价 1299 元华为给 FreeClip 2 出了三个拼色新版本，1299 元开售，耳机里塞了 NPU 芯片，能用小艺翻译和头动播报，单次续航 9 小时。
IT之家Source
Huawei华为
China·Shenzhen
Ascend AI chips and Pangu models
What they do
Founded in Shenzhen in 1987, Huawei's AI business spans the Ascend AI chips and CANN software stack, the Pangu models and Huawei Cloud; it is China's main domestic supplier of AI compute.
Latest coverage
Stories in the feed that mention this companyChinese contentDeepSeek V4.1 低价策略或挤压西方托管与国内对手生存空间
这条帖子把 DeepSeek V4.1 的超低价拆成了一盘棋：压价、锁推理栈、自建 Ascend 算力，读完你会重新理解国产模型怎么打价格战。
PyTorchCon Demo Theater: 13 Sessions on Training, Inference and Agents
The lineup reveals where the PyTorch ecosystem is concentrating effort: inference speed for video models, MoE expert parallelism on NVL72, and fault-tolerant training that survives link and node failures live. Engineers choosing accelerator backends can use these demos to compare AWS Trainium, Intel AI PCs, TPUs and Rebellions hardware on real workloads. The Huawei CRCR CI relay is also a practical pattern for teams maintaining forks against upstream PyTorch.
Chinese content华为 HarmonyOS 7 小艺 AI 快捷栏延期至 11 月初分批推送
本来 10 月底就能用上的小艺 AI 快捷栏，华为客服确认推迟到 11 月初了，鸿蒙 7 用户留意系统更新提示。
IT之家Source
Chinese content独家：DeepSeek 协助华为设计下一代 AI 芯片
The Information 独家：DeepSeek 在帮 Huawei 设计下一代 AI 芯片，华为工程师直接进驻 DeepSeek 办公室，目标是不靠 Nvidia 训练先进模型。
Chinese content北汽星钽 5X 预售 13.99 万元起：方盒子设计，纯电版首搭华为乾崑 ADS 5 Pro
13.99 万起的方盒子越野车，纯电版还是第一款装华为乾崑 ADS 5 Pro 的，座位放倒能躺 1.8 米，预算十几万想兼顾智驾和越野的可以看看。
IT之家Source
Chinese content华为与哈工大（深圳）联合设立计算硬件专项，拟招10名硕士生
华为和哈工大深圳合办硕士专班，招10人，读研期间直接进华为做鲲鹏昇腾硬件课题，毕业还有优先留岗，相关专业的可以关注。
IT之家Source
Industry17:17
Chinese content华为靳玉志谈自动驾驶进入“下半场”：安全、体验与产业体系三大变化华为乾崑掌门人靳玉志讲清了自动驾驶下半场怎么打，还放出一堆硬数据，比如 168 亿公里辅助驾驶里程、680 万次避险，想看行业走向的可以看看。
IT之家Source
Chinese content华为第七届奥林帕斯奖启动征集，单项奖金最高 100 万元
华为拿 100 万征存储难题，这次四个方向全对准 Agentic AI 的数据瓶颈，搞存储和知识图谱的可以看看题目。
IT之家Source
Chinese contentDeepSeek 开源降本工具，适配 Nvidia Blackwell 与华为昇腾
DeepSeek 开源了个省钱工具，能在 Nvidia Blackwell 和华为昇腾 950 上部署模型，自建推理服务的可以看看。
DeepSeek V4.1F shows 5x-7.4x decode edge on Ascend 950DT vs GB200
The cost crossover at 6.8K tokens/s/chip reframes the Ascend-versus-Blackwell debate from raw benchmark losses to workload-dependent economics, suggesting DeepSeek could run its API on a sub-US$1B Blackwell cluster while pivoting post-training to Ascend. Practitioners tracking inference infrastructure or Huawei's CANN ecosystem get a rare concrete per-chip price and throughput comparison. The proposed 128-chip Ascend supernode topology has no Nvidia equivalent, which matters for anyone planning large-scale decode deployments.