Alibaba AI Chip Unveiled as Qwen Targets Up to 10 Trillion Parameters

The new Alibaba AI chip put the Chinese technology group back at the center of the global AI race on September 22, 2026, as it unveiled the Zhenwu V900 processor and outlined plans for future Qwen models containing between 5 trillion and 10 trillion parameters.

The announcements were made at Alibaba Cloud’s annual Apsara Conference in Hangzhou, where the company laid out an integrated strategy spanning foundation models, semiconductors, data centers and cloud infrastructure.

Alibaba’s Hong Kong-listed shares responded by rising about 5.1% to their highest level in a month.

What is the Alibaba AI chip Zhenwu V900?

The Zhenwu V900 was developed by Alibaba’s T-Head semiconductor unit as the latest generation of the company’s in-house artificial-intelligence processors.

CEO Eddie Wu said the new chip delivers about three times the performance of its predecessor, the Zhenwu M890.

Alibaba also says clusters can link up to roughly 500,000 processors to train and run extremely large AI models.

The Zhenwu V900 is scheduled to enter mass production and commercial release in the first quarter of 2027.

Why does the chip matter for China?

The announcement comes as Chinese technology companies accelerate efforts to develop domestic alternatives to advanced U.S.-designed AI processors.

Washington maintains export restrictions on some of Nvidia’s most advanced chips for China, increasing pressure on Chinese companies to build domestic computing infrastructure.

For Alibaba, controlling its chip, foundation models and cloud infrastructure could reduce dependence on outside suppliers while giving the company greater control over AI computing costs and availability.

The development also comes as Washington and Beijing continue negotiations involving technology, trade and artificial intelligence. Profile News has covered the latest U.S.-China talks on AI, trade and technology.

Qwen models could reach 10 trillion parameters

Alibaba’s announcement extends beyond semiconductors.

Wu said the Qwen team plans to train models with between 5 trillion and 10 trillion parameters to tackle more complex and longer-horizon tasks.

The company’s current flagship model, Qwen 3.8 Max, contains around 2.4 trillion parameters.

Alibaba said Qwen 4 is currently being trained, with later Qwen 4.5 and Qwen 5 models expected to scale toward the 5 trillion-to-10 trillion parameter range.

Do more parameters automatically mean better AI?

Not necessarily.

Parameters are variables learned by an AI model during training and are commonly used as a rough measure of a model’s scale.

A higher parameter count alone does not guarantee greater accuracy, intelligence or efficiency because performance also depends on training data, architecture, optimization and inference design.

The 10-trillion figure therefore primarily shows the scale of Alibaba’s computing ambitions and the infrastructure required to support them.

Alibaba is building a full AI stack

Tuesday’s announcements highlight Alibaba’s attempt to control more layers of the artificial-intelligence technology stack.

Its strategy now spans Qwen foundation models, Zhenwu processors, Alibaba Cloud, large-scale data centers and AI applications for businesses and consumers.

Wu said Alibaba intends to build across the full AI technology stack rather than depending on a single external supplier for key computing components.

20 gigawatts of data-center capacity by 2032

AI models of this size require enormous computing capacity.

Alibaba therefore set a target for Alibaba Cloud’s global data-center capacity to exceed 20 gigawatts by 2032.

Wu described customer demand for AI computing as exceptionally strong and said medium- and long-term demand currently exceeds Alibaba’s ability to supply enough infrastructure.

AI revenue is already growing rapidly

Alibaba’s AI strategy is also beginning to generate substantial revenue.

The company says AI-related product revenue reached about $1.8 billion in its latest quarter, extending triple-digit year-over-year growth for a twelfth consecutive quarter.

AI Cloud and Compute Services revenue reached approximately $7.1 billion, up 45% from a year earlier.

Alibaba also spent nearly $10 billion in capital expenditure during the quarter, up 75%, largely reflecting its continued investment in AI infrastructure.

Zhenwu already has a customer base

The V900 is not Alibaba’s first attempt to commercialize its own AI silicon.

The previous Zhenwu M890 processor was launched in May and is now used through Alibaba Cloud by more than 650 customers across more than 20 industries, according to the company.

Those industries include autonomous driving, financial services and internet platforms.

How does Alibaba fit into the Nvidia and AMD race?

The global AI-chip race is increasingly expanding beyond Nvidia.

In the United States, AMD has become a major alternative supplier and crossed a $1 trillion market capitalization this week as investors returned aggressively to AI-related stocks.

Profile News has covered AMD’s surge above a $1 trillion valuation during the AI rally.

In China, Alibaba and other technology groups are attempting to build domestic alternatives spanning chips, models and cloud infrastructure.

Alibaba shares jump 5%

Investors reacted positively to the Apsara announcements.

Alibaba’s Hong Kong-listed shares climbed around 5.1% on Tuesday, reaching their highest level in about a month.

The move reflects growing investor confidence that cloud computing and artificial intelligence could become increasingly important growth engines for a company historically dominated by e-commerce.

China’s push for AI-chip independence

The challenge for Alibaba extends beyond designing a fast processor.

Building advanced AI chips at commercial scale also requires access to manufacturing, memory, networking, packaging and large data centers.

As U.S. restrictions continue, the ability to construct a more independent Chinese AI technology stack has become part of the wider strategic competition between the world’s two largest economies.

Zhenwu V900 and Alibaba’s future Qwen models therefore represent more than a product update. They are part of a global contest over who controls the chips, computing capacity and foundation models powering the next generation of artificial intelligence.

Report Sources: Reuters — Alibaba’s new AI chip and Qwen roadmap; Alibaba Group — AI and cloud business growth; Alibaba Cloud — Apsara Conference 2026.