Chinese tech giant unveils domestically built processor hours before the US chipmaker's hotly anticipated results
Whether by coincidence or design, Alibaba chose to unveil its latest artificial intelligence chip just hours before Nvidia Corp (NASDAQ:NVDA, XETRA:NVD), the dominant force in AI processors, reports one of the most closely watched sets of quarterly results this earnings season.
The timing puts a spotlight on the intensifying competition Nvidia faces, not just from Chinese rivals building domestic alternatives under the pressure of US export curbs, but from an increasingly crowded field of Western challengers too.
Alibaba's Zhenwu M890, developed by its semiconductor design subsidiary T-Head, delivers three times the performance of its predecessor and is purpose-built for AI "agents," software systems capable of carrying out complex, multi-step tasks with limited human oversight.
The Hangzhou-based company also laid out a multi-year chip roadmap, with successors due in 2027 and 2028, each promising roughly threefold performance gains over the prior generation.
T-Head said it has shipped more than 560,000 Zhenwu units to date, with over 400 external customers across 20 industries deploying the chips.
Alibaba last year committed more than $53 billion to cloud and AI infrastructure over three years, its largest ever investment in the sector.
The chip was announced alongside Qwen 3.7-Max, the latest version of Alibaba's flagship large language model, which it said can operate continuously for up to 35 hours without performance degradation.
Nvidia is not short of rivals closer to home, either.
Cerebras, which sells a fundamentally different form of AI processor, held its initial public offering last week, while AMD is preparing to launch its rack-scale server system later this year.
Amazon's custom chip segment now has an annual revenue run rate exceeding $20 billion and is growing at triple-digit percentages, with major AI customers including OpenAI and Anthropic signing deals for its Trainium processors.
Google, meanwhile, used its annual I/O developer conference this week to announce two new chips, the TPU 8i for running AI software and the TPU 8t for training models, further expanding its in-house silicon ambitions.
None of which is likely to dent Nvidia's near-term dominance, but the direction of travel is clear: the company's customers are also becoming its competitors.