Alibaba Unveils New AI Chip and Plans Much Larger Qwen Model

Alibaba unveiled a new AI chip and plans for a substantially larger artificial intelligence model on Tuesday, September 22, expanding its effort to develop the hardware, software and cloud infrastructure behind its AI services.

Latest News
  • Alibaba unveiled its Zhenwu V900 AI chip on September 22.

  • Mass production is scheduled for the first quarter of 2027.

  • The company plans an AI model with five to ten trillion parameters.

  • Performance claims and expansion targets remain unproven.

September 22, (THEWILL) – Alibaba unveiled a new AI chip and plans for a substantially larger artificial intelligence model on Tuesday, September 22, expanding its effort to develop the hardware, software and cloud infrastructure behind its AI services.

Chief executive Eddie Wu presented the plans at the company’s Apsara conference in Hangzhou, China. The announcement includes the Zhenwu V900 chip, developed by Alibaba’s T-Head semiconductor business, and a planned successor to its flagship Qwen model.

For cloud customers and developers, the commercial promise is a supplier that can coordinate more of the technology needed to deliver AI. The announcement, however, provides no basis for concluding that customers will immediately receive cheaper services.

Ask ZiVA 728x90 Ads

Alibaba

A New Chip With Production Still Ahead

Alibaba says the V900 delivers three times the performance of its predecessor and can support both training AI models and running them after deployment.

Mass production is scheduled for the first quarter of 2027, according to reporting by The Wall Street Journal. Its unveiling therefore precedes the planned manufacturing rollout, while the performance comparison remains Alibaba’s claim.

Customers assessing the chip will need evidence from their own workloads, including operating costs, software compatibility and dependable availability. A headline performance multiplier alone cannot establish the savings a business would achieve.

READ ALSO:

Larger Models Will Need More Infrastructure

Alibaba also plans to train a model containing five trillion to ten trillion parameters, compared with the reported 2.4 trillion in its current Qwen3.8-Max.

Alongside that ambition, the company is targeting more than 20 gigawatts of global data-centre capacity by 2032. Both are future plans, rather than capabilities delivered with Tuesday’s announcement.

The combination puts Alibaba’s next phase of AI development beyond software releases. Achieving those targets will require manufacturing, construction and sustained investment, followed by enough customer demand to justify the expense.

Alibaba

For businesses considering its services, the practical developments to follow are commercial availability, published pricing and independently tested performance.

Tuesday’s announcement sets out what Alibaba intends to build; those later disclosures will show what customers can actually buy and use.

Illustrated portrait of a smiling Black woman with short dark hair (head-and-shoulders).

Joy Onuorah is a business journalist and brand communications specialist covering financial markets, artificial intelligence, digital economy, and the ideas reshaping business across Africa and the global market. Beyond her reporting for TheWill, Joy uses brand strategy, storytelling, copywriting, and high-value SEO to help brands build lasting market authority.

More Articles Like This