Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Li Auto showcases Mach M100 AI chip at TSMC OIP forum

AI summary

Li Auto CTO Xie Yan presented the company’s in-production Mach M100 edge AI chip at the 2026 TSMC OIP ecosystem forum in Beijing on November 17. Li Auto says the 5nm automotive chip uses a dynamic dataflow architecture, delivers 1280 TOPS and has shipped in several models; when running Qwen3.5-35B-A3B, it achieved 2.7 times the prefill speed and 1.5 times the decode speed of NVIDIA DGX Spark. Claims including “world’s first,” utilization above 82% and benchmark superiority remain vendor-reported without independent testing.

In-depth insight

The forum presentation gave Li Auto’s edge-AI architecture and mass-produced automotive silicon a prominent technical platform, while the reported Qwen3.5-35B-A3B comparison highlights its claimed advantage in on-device inference.

2026 年 11 月 17 日,台积电 OIP(Open Innovation Platform,开放创新平台)生态系统论坛在北京举行。作为全球半导体设计、EDA、IP 及封装测试领域最具影响力的年度技术盛会之一,本届论坛汇聚了芯片架构、AI 计算与系统创新方向的顶尖企业与专家。理想汽车首席技术官谢炎受邀发表主题演讲,成为本次论坛唯一的特邀演讲嘉宾,理想汽车也成为台积电 OIP 生态系统论坛举办以来首家进行特邀演讲的大陆企业。

图片 1

谢炎演讲的题目为《高效 AI 推理系统的设计思考与实践》,核心内容围绕理想汽车自研的马赫 M100 芯片展开。从架构设计到量产装车,这颗芯片完整走过了从原创技术到规模化应用的全流程。

#Li Auto#Mach M100#TSMC#ISCA#NVIDIA DGX Spark#Qwen3.5-35B-A3B#Mach VLA 2.5#3D ViT

Original · 4 min readRead the original
Group chat

No comments yet. Start the conversation.