Alibaba has unveiled its Zhenwu M890 AI chip and Qwen3.7-Max model, signaling a bigger push for domestic alternatives to Nvidia amid U.S. export restrictions.
Alibaba has unveiled a new artificial intelligence chip, the Zhenwu M890, alongside a fresh version of its flagship model, Qwen3.7-Max, in a move that underscores how quickly China’s technology giants are trying to reduce their dependence on foreign hardware.
The announcements, made at Alibaba’s Cloud Summit, point to a broader strategy: build the chips, the models and the cloud infrastructure needed to power the next generation of AI services inside China. That ambition comes at a time when U.S. export controls have made it harder for Chinese companies to buy the most advanced processors from Nvidia and other American suppliers.
A chip built for AI agents
Alibaba said the Zhenwu M890 is designed for “agentic” AI workloads — systems that can perform complex tasks with limited human supervision. In practical terms, that means software that can plan, act and keep working across multiple steps, rather than simply answering one-off prompts.
The company described the chip as a major step up from its previous Zhenwu 810E, saying it delivers roughly three times the performance. Alibaba also said its chip unit, T-Head, has shipped more than 560,000 Zhenwu chips to date and that more than 400 external customers across 20 industries have already adopted them.
That scale matters. In the global AI race, performance alone is no longer the only measure of success. Reliability, supply, and the ability to deploy at volume are just as important, especially for enterprise customers looking for a domestic alternative to U.S. silicon.
Qwen3.7-Max joins the stack
Alongside the new chip, Alibaba introduced Qwen3.7-Max, the latest version of its large language model. The company said the model is designed for advanced coding and long-running agent tasks, and claimed it can operate continuously for up to 35 hours without losing performance.
That claim places the model squarely in the growing market for AI systems that do more than generate text. Businesses are increasingly looking for tools that can handle software development, workflow automation and customer operations with less human intervention.
For Alibaba, the pairing of a new chip and a new model is strategic. It allows the company to present itself not just as a cloud provider, but as a full-stack AI supplier with hardware, software and services bundled together.
Long-term roadmap
Alibaba also outlined a multi-year roadmap for future chips. It said the V900 is expected in the third quarter of 2027 and should deliver another threefold performance gain over the M890. A further chip, the J900, is scheduled for the third quarter of 2028.
That kind of forward planning suggests Alibaba is trying to reassure customers that its AI infrastructure will keep improving, even as access to foreign hardware remains uncertain. In China’s fast-moving tech market, long-term commitments can be as important as immediate product launches.
The roadmap also reflects the scale of the challenge. Matching the most advanced global AI chips is difficult, but building a credible domestic ecosystem may matter even more for Chinese firms navigating restricted access to overseas semiconductors.
Geopolitics and opportunity
The timing of the launch is significant. Washington’s restrictions on high-end chip sales to China have created what many analysts see as a vacuum in the market, pushing Chinese companies to accelerate homegrown alternatives.
Alibaba’s moves place it alongside Huawei and Cambricon in a domestic competition that is becoming both commercial and geopolitical. Each company is trying to prove that Chinese AI can advance without relying on the most powerful U.S. processors.
At the same time, Nvidia remains an important reference point. The company’s chips still set the benchmark for performance, but China’s market is increasingly shaped by policy, procurement limits and strategic self-reliance.
For Alibaba, the message is clear: the future of AI in China may depend not only on smarter models, but on who controls the chips underneath them.
