此语言仍有 24 条内容或界面任务待翻译;未完成部分显示中文原文。

Insights & News

AI、算力、具身智能机器人与智能硬件的产业动态,以及选型与采购指南。

今日更新0 篇Updates This Week23 篇Published147 篇订阅周报

2026-10-28 · 本站更新

Compute and ServersIndustry Updates

CCF National High-Performance Computing Conference 2026: Sugon8000 100,000-Card Cluster Debuts as Compute Evaluation Shifts to Scaling Efficiency and Density

The 22nd CCF National High-Performance Computing Conference opened in Shenzhen on 18 10, 2026. Sugon8000, a fully domestic 100,000-card AI supercluster, made its first public appearance as a real system, sharing the stage with Lingsheng—which topped TOP500 with an all-CPU architecture—and the Kunpeng supernode. The report argues that the benchmark for evaluating compute is shifting from peak compute toward scaling efficiency, heat load per unit area, and ecosystem adaptation, and the conference released the New Era Supercomputing Industry Collaborative Development Action Plan along with the Superintelligence Collaboration Open Community.

DOIT · 约 10 分钟

大模型与软件Industry Updates

Moonshot AI Releases Kimi K3.1 Pro, Model Weights to Be Open-Sourced Before 4 11

According to its official blog, Moonshot AI has released the first model in its K3.1 series, Kimi K3.1 Pro, along with the high-speed variant Kimi K3.1 Pro Highspeed. Both models support a 1000K-token context window, and the high-speed version runs at 2 to 3 times the speed of the standard version. Official data shows K3.1 Pro scored 65.70% on Terminal-Bench 4.0, with a notably lower hallucination rate than K3. The models are now live on Kimi.ai and other platforms as well as API, with full weights planned to be released before 4 11.

PANews(腾讯新闻) · 约 3 分钟

2026-10-27 · 本站更新

Compute and ServersIndustry Updates

NVIDIA Introduces DSX POD Modular AI Factory Reference Design, With Deliveries Planned for Early 2027

On 21 10, 2026, NVIDIA introduced the DSX POD modular AI factory reference design, integrating the Vera Rubin computing system with power, liquid cooling, and networking infrastructure into shipping containers, with integration and testing completed before leaving the factory. Partners plan to begin global deliveries in early 2027; deployment time is said to be reduced from about six months to as fast as 14 days, though that figure excludes site construction.

硬件之家 · 约 4 分钟

Compute and ServersIndustry Updates

Google and Meta Open-Source TorchTPU: Migrate PyTorch Models from CUDA to TPU by Changing One Line of Device Code

Google and Meta open-sourced TorchTPU at PyTorch Conference North America on 2026 10, 20. Developers can run PyTorch models on TPUs by changing .to("cuda") to .to("tpu") in device initialization. The project is released under the Apache 2.0 license, built on the PrivateUse1 backend, compatible with torch.compile and XLA/StableHLO, and officially validated on 5000 Hugging Face models with numerical consistency of 97%. However, the closed TPU instruction set, single-process limitation, and communication overhead still constrain peak performance.

CSDN博客(开源早知道) · 约 6 分钟

AI Chips & StorageIndustry Updates

Meta生产环境启用Google TPU,PyTorch原生TorchTPU方案落地

Meta、Google与RadixArk合作推出TorchTPU,使SGLang、vLLM等PyTorch推理引擎无需重写代码即可在TPU上运行;Meta确认PySGLang-TorchTPU已进入生产部署,减少对Nvidia生态的依赖。文中还给出Ironwood TPU与GB200的Kimi K3推理对比数据,以及Amazon开源TorchNeuron等相关动向。

同一事件另有 1 篇报道

飞翔科技频道(转载Toms Hardware) · 约 3 分钟

AI Chips & StorageIndustry Updates

Google TPUs Supplied to External Customers for the First Time, Now Running in Meta's Production Environment

Google's in-house Tensor Processing Units have been supplied to an external customer for the first time and are now handling production AI workloads in Meta's data centers. The key to deployment is inference engine adaptation: SGLang has released a PyTorch-native version for TPUs, and vLLM has also published a corresponding open-source version, allowing most models to migrate to TPUs without code rewriting.

IT时代网(腾讯新闻发布页) · 约 4 分钟

AI Chips & StorageIndustry Updates

Yuanjie Semiconductor Announcement: Projected Revenue of 6.5 to 7亿元 for Q3 of 2026, First Batch of Phase II Equipment to Enter Plant Soon

Shaanxi Yuanjie Semiconductor Technology Co., Ltd. issued an announcement on its recent production and operations, projecting Q3 2026 revenue of 65,000.00万元 to 70,000.00万元, up 264.56% to 292.60% year over year; and net profit attributable to the parent of 40,300.00万元 to 45,300.00万元. The company said AI computing power demand is driving demand for high-speed optical modules and upstream laser chips, CPO/NPO-related products are advancing steadily, and the Phase II facility is expected to begin forming capacity from the end of 2026 to early 2027.

东方财富网(转载源杰科技公告) · 约 5 分钟

2026-10-26 · 本站更新

Compute and ServersIndustry Updates

Enflame Technology Makes Its Exhibition Debut at the 2026 Industry Fair NEXT-GEN IC EXPO, Showcasing the Yunsui ESL64-O Supernode and Full-Stack Computing Solutions

Enflame Technology made its exhibition debut at the 2026 China International Industry Fair Integrated Circuit Expo (NEXT-GEN IC EXPO), held at the National Exhibition and Convention Center in Shanghai from 10 12 to 16, showcasing the Yunsui ESL64-O high-performance supernode. The solution, jointly developed with ZTE based on Enflame's self-developed accelerator modules, adopts an orthogonal architecture and zero-cable design, supporting networking of intelligent computing clusters at the 10,000-card level. Its self-developed domestic-chip-and-domestic-model high-efficiency adaptation and optimization system won the 2026 NEXT-GEN IC EXPO Innovation Achievement Award.

法讯(qlfz365.cn 业内资讯) · 约 7 分钟

大模型与软件Industry Updates

OpenAI's Three Models and Codex to Arrive on Google Cloud Model Garden, General Availability Expected Within Weeks

On 21, 10, 2026, OpenAI and Google Cloud announced a partnership under which GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna, and Codex agent workflows will enter Google Cloud Model Garden, with general availability expected within the coming weeks. Customers can purchase via Google Cloud Marketplace, with usage consolidated onto a single invoice and offsettable against committed spend; data is committed to staying within customer-designated boundaries and not used for training foundation models. Microsoft's exclusive distribution agreement expired in 4 2026, meaning OpenAI models now cover all three major US cloud providers.

网易号·大数据文摘 · 约 6 分钟

Compute and ServersIndustry Updates

Google Open-Sources TorchTPU, Making TPU a Native PyTorch Hardware Option

According to a PANews report citing the official Google Cloud blog, Google announced the open sourcing of TorchTPU at PyTorch Conference 2026. The project was jointly developed by Google and the Meta PyTorch team, and the code has been published on GitHub.

PANews(腾讯新闻) · 约 4 分钟

Compute and ServersIndustry Updates

CPC Central Committee and State Council Issue "Opinions on Developing New Quality Productive Forces," Systematically Incorporating the AI Industry Chain for the First Time

According to CFI.cn, the "Opinions on Developing New Quality Productive Forces" were jointly issued by the CPC Central Committee and the State Council, marking the first time the AI industry chain has been systematically incorporated into the national strategic productivity framework, with four pillars clearly defined: the computing power network, large model applications, data elements, and technology finance. The document proposes building a nationally integrated computing power network, establishing industry application pilot-testing bases, and increasing technology finance supply.

中财网 CFi.CN · 约 5 分钟

大模型与软件Industry Updates

Moonshot AI Releases Kimi K3.1 Pro with 1000k Context Support

Moonshot AI released Kimi K3.1 Pro in the early hours of 10/24, calling it the first model in the K3.1 series and its most capable 3-trillion-parameter model to date. It supports a 1000k context window and delivers significant improvements in coding, 3D creation, financial research, and long-horizon Agent capabilities. The model is now available via kimi.com, Kimi App, Kimi Work, Kimi Code, and Kimi API, with the full model weights planned for release before 11/4.

同一事件另有 1 篇报道

每日经济新闻(经凤凰网大风号转载) · 约 3 分钟

Product Selection Assistant

让需求变成可修改的筛选条件

依据已录入参数生成条件,结果可查看和修改。帧率与性能需核对实测,不能仅凭算力数值保证。