Alibaba’s New Chip: Now With 3x More Robot Interns
Alibaba just dropped the Zhenwu M890, a chip built for AI agents with goldfish-proof memory and teamwork skills. Think less “faster chatbot,” more “office of tireless robot interns.” It’s 3x speedier than the last one, and there’s a whole sequel plan: V900 in 2027, J900 in 2028—because cliffhangers sell.
They’ve already shipped over half a million Zhenwu chips, and the new ones arrive in 128-pack server racks like Costco for accelerators. Bonus: a fresh LLM, Qwen 3.7-Max, tuned for marathon coding and 35-hour brain sprints.
TL;DR: Alibaba isn’t dodging export rules—it’s building its own AI theme park: chip, model, cloud, the works. Bring your data.
Beijing Buffet: All-You-Can’t-Ship Nvidia, Huawei Takes Seconds
Trump popped into Beijing with Jensen Huang as his plus-one, teased “something” on chip exports, then left with nothing but jet lag. Nvidia’s H200s? Approved by Washington, blocked by Beijing—licenses galore, zero deliveries. Why? The US says “use them only in China,” Beijing says “use them only abroad.” Result: Schrödinger’s GPU.
Meanwhile, the real plot: China’s stacking its chips on Huawei. DeepSeek tuned for Ascend, Tencent says domestic GPUs ramp through 2026, Alibaba’s T-Head hits mass production, Nvidia’s China revenue face-plants to ~5%. CEO diplomacy met policy physics and bounced.
Bottom line: the H200 deal is legally greenlit, practically icebound, and Huawei’s quietly eating the sandwich while everyone watches the tarmac.
*Disclaimer: The content in this newsletter is for informational purposes only. We do not provide medical, legal, investment, or professional advice. While we do our best to ensure accuracy, some details may evolve over time or be based on third-party sources. Always do your own research and consult professionals before making decisions based on what you read here.