NextFin News (HANGZHOU) -- Alibaba Group's semiconductor division T-Head introduced its next-generation unified training and inference AI chip, the Zhenwu M890, at the 2026 Alibaba Cloud Summit on Wednesday.
The new processor delivers three times the performance of its predecessor, the Zhenwu 810E, and is configured with 144GB of high-bandwidth memory alongside an inter-chip interconnect bandwidth of 800GB/s.
Engineered with native data precision support ranging from FP32 to FP4, the hardware targets large-scale model training and mixed-precision inference tasks optimized for AI agents. Alongside the silicon debut, Alibaba Cloud launched its self-developed ICN Switch 1.0 interconnect chip—enabling full-bandwidth 64-card scaling—and rolled out the Panjiu AL128 supernode server to house the accelerators for enterprise cloud clients.
Explore more exclusive insights at nextfin.ai.
Insights
What are key technical principles behind Zhenwu M890 AI chip?
What historical advancements led to the development of Zhenwu M890?
How does Zhenwu M890 compare to Zhenwu 810E?
What is the current market situation for AI chips?
What feedback have users provided about the Zhenwu M890?
What industry trends are shaping the AI chip landscape?
What recent updates have been announced regarding Alibaba Cloud's products?
What policy changes affect the semiconductor industry currently?
What potential long-term impacts could Zhenwu M890 have on AI workloads?
What are some core challenges faced by AI chip manufacturers?
What controversies exist surrounding the AI chip market?
How does Zhenwu M890's architecture differ from competitors?
Can you provide examples of successful AI chip implementations?
What are the scaling capabilities enabled by ICN Switch 1.0?
What role does mixed-precision inference play in AI workloads?
What future developments can we expect from Alibaba's semiconductor division?
