Alibaba is planning a next-generation artificial intelligence model with between 5 trillion and 10 trillion parameters, marking one of the most ambitious AI development plans announced by a Chinese technology company. The plan was revealed at Alibaba Cloud’s 2026 Apsara Conference in Hangzhou, alongside the launch of the company’s new Zhenwu V900 AI chip.
Alibaba’s announcement expands the competition surrounding increasingly powerful AI models while highlighting a broader strategy: building the models, chips and cloud infrastructure needed to operate advanced artificial intelligence systems.
The company says its next-generation model is intended to handle more complex and longer-horizon tasks and support research toward artificial superintelligence.
What Is Alibaba’s 10 Trillion-Parameter AI Model?
Alibaba’s Qwen team is developing a future AI model at a scale of 5 trillion to 10 trillion parameters.
Parameters are numerical values that an AI model learns during training. They help the model recognize patterns and generate responses. Parameter count is one way to describe the scale of an AI system, although it does not by itself determine how capable or efficient a model will be.
Alibaba’s latest flagship model, Qwen3.8-Max, has 2.4 trillion total parameters. The proposed 10-trillion-parameter model would therefore represent a substantial increase in total model scale.
Alibaba has not announced a final release date or complete technical specifications for the future model.
Why 10 Trillion Parameters Matters
The proposed scale is significant because frontier AI development is increasingly moving toward systems designed to solve complex, multi-step problems.
Alibaba says its next-generation model is intended to improve performance on complex and long-horizon tasks and contribute to research into recursive self-improvement and artificial superintelligence.
However, bigger does not automatically mean better.
Modern AI systems can use architectures such as Mixture of Experts (MoE) to contain extremely large numbers of total parameters while activating only a portion of them for a particular task.
Alibaba’s Qwen3.8-Max provides an example. Although it contains 2.4 trillion total parameters, Alibaba says its architecture activates approximately 95 billion parameters at a time.
This means the future performance of Alibaba’s model will depend not only on parameter count but also on architecture, training data, computing infrastructure, algorithms and inference efficiency.
Alibaba Is Building More Than a Bigger AI Model
The most important part of Alibaba’s announcement may be its full-stack AI strategy.
The company is developing three major components:
- AI models through the Qwen family
- AI chips through its T-Head semiconductor business
- AI cloud infrastructure through Alibaba Cloud
Alibaba describes these components as the foundation of what it calls the era of “Machine Intelligence.”
This strategy could allow Alibaba to control more of the technology stack required to train and operate large AI systems.
Zhenwu V900: Alibaba’s New AI Chip
Alongside its model roadmap, Alibaba introduced the Zhenwu V900, a new AI accelerator developed by its T-Head semiconductor division.
Alibaba says the V900 delivers approximately three times the performance of its previous-generation Zhenwu M890 chip.
The company also says a V900-based cluster can support up to 500,000 cards for frontier AI model training and inference.
Alibaba expects the Zhenwu V900 to enter mass production and commercial deployment in the first quarter of 2027.
The chip announcement is important because developing large AI models requires enormous amounts of computing power.
China’s AI Chip Strategy
Alibaba’s chip development also reflects China’s broader push to strengthen domestic semiconductor capabilities.
U.S. restrictions on advanced AI chips have made access to cutting-edge computing hardware a major issue for Chinese technology companies.
Alibaba’s development of proprietary AI accelerators gives the company another potential source of computing capacity while reducing reliance on external suppliers.
Alibaba is not alone in this effort. Chinese companies including Huawei, Baidu and Cambricon are also developing domestic AI computing technologies.
The result is an increasingly competitive Chinese AI hardware ecosystem.
Qwen Is Becoming Alibaba’s AI Foundation
Alibaba’s 10-trillion-parameter roadmap builds on the rapid expansion of its Qwen model family.
In August 2026, Alibaba introduced Qwen3.8-Max, which it described as its largest and most capable Qwen model at the time.
The model has 2.4 trillion total parameters, supports a context window of up to 1 million tokens and uses a Sparse Mixture-of-Experts architecture. Alibaba says it is designed for coding, research, real-world tasks and long-horizon execution.
The proposed 5-to-10-trillion-parameter successor therefore represents a major increase in the scale Alibaba is targeting.
Alibaba’s $53 Billion AI Investment
Alibaba’s AI strategy extends beyond research laboratories.
The company has committed more than $53 billion over three years to adding AI capabilities to its businesses, including its traditional e-commerce operations.
The investment reflects a broader transformation of Alibaba from an e-commerce and cloud company into a major AI infrastructure provider.
Its AI ambitions include cloud computing, enterprise AI, autonomous agents, chips, foundation models and AI-powered consumer services.
Alibaba Cloud Targets 20 GW of Data-Center Capacity
Large AI models require enormous computing infrastructure.
Alibaba therefore plans to significantly expand its cloud infrastructure. The company has set a target for its globally operated data-center capacity to exceed 20 gigawatts by 2032.
Alibaba says demand for AI computing is already exceptionally strong and that shortages throughout the AI data-center supply chain are limiting how quickly computing capacity can be expanded.
This creates a direct connection between the company’s AI models, chips and cloud strategy.
More powerful models require more computing capacity, while more computing capacity can support more customers and AI applications.
How Alibaba’s Strategy Changes the Global AI Race
Alibaba’s announcement adds another major dimension to the global AI competition.
The race is no longer limited to developing the most capable chatbot or foundation model.
Companies increasingly need access to:
- Advanced AI accelerators
- High-performance CPUs
- Large-scale data centers
- Networking infrastructure
- Efficient model architectures
- Massive training datasets
- AI software platforms
- Global cloud capacity
Alibaba is attempting to build several of these layers itself.
That could become strategically important as competition increases between U.S. and Chinese technology companies.
What Does This Mean for Nvidia?
Alibaba’s strategy could also have implications for global AI chip companies such as Nvidia.
Nvidia remains a major supplier of advanced AI computing infrastructure, while Alibaba is developing its own chips for its cloud and AI systems.
The significance is not necessarily that Alibaba will replace Nvidia globally. Rather, the development of domestic AI chips could increase competition and provide Chinese companies with additional computing options.
For the global semiconductor industry, the trend could encourage more technology companies and governments to invest in specialized AI processors.
What Does a 10 Trillion-Parameter Model Mean for AI Users?
For ordinary users, the parameter number itself may not be noticeable.
What matters is whether larger models can deliver better performance in areas such as:
- Complex reasoning
- Software development
- Scientific research
- Multimodal understanding
- Autonomous agents
- Long-term task execution
- Business automation
Alibaba says its next-generation model is being designed for increasingly complex and longer-horizon tasks.
If successful, such systems could support AI agents capable of completing multiple stages of work rather than simply answering individual questions.
Key Facts About Alibaba’s AI Plans
| Feature | Details |
|---|---|
| Company | Alibaba |
| AI family | Qwen |
| Planned model size | 5–10 trillion parameters |
| Current flagship | Qwen3.8-Max |
| Current flagship scale | 2.4 trillion parameters |
| New AI chip | Zhenwu V900 |
| V900 performance | About 3× previous generation |
| V900 mass production | Expected Q1 2027 |
| Data-center target | More than 20 GW by 2032 |
| AI investment | More than $53 billion over three years |
Frequently Asked Questions
What is Alibaba’s 10 trillion-parameter AI model?
It is a planned next-generation Qwen model that Alibaba says will be trained at a scale of between 5 trillion and 10 trillion parameters. The company has not yet announced a final product name or release date.
Is Alibaba’s 10 trillion-parameter model already available?
No. Alibaba has announced the development and training roadmap, but the 5-to-10-trillion-parameter model is not yet publicly available as a finished product.
How large is Alibaba’s current Qwen model?
Alibaba’s Qwen3.8-Max has 2.4 trillion total parameters. Its Sparse Mixture-of-Experts architecture activates approximately 95 billion parameters at a time.
What is the Zhenwu V900?
The Zhenwu V900 is Alibaba’s next-generation AI accelerator developed by its T-Head semiconductor division. Alibaba says it delivers three times the performance of the previous Zhenwu M890 generation.
When will the Zhenwu V900 be available?
Alibaba expects the V900 to enter mass production and commercial release during the first quarter of 2027.
Why is Alibaba building its own AI chips?
Developing proprietary AI chips gives Alibaba greater control over the computing infrastructure used for its models and cloud services while supporting China’s broader domestic semiconductor ecosystem.
The Bigger Picture
Alibaba’s 10-trillion-parameter AI roadmap represents a broader shift in the global artificial intelligence industry.
The competition is increasingly becoming a race to build complete AI ecosystems, rather than individual models.
Alibaba is combining Qwen models, proprietary AI chips and Alibaba Cloud infrastructure into a vertically integrated strategy designed to support large-scale AI development.
Whether a 10-trillion-parameter model ultimately delivers a major leap in AI capability remains to be demonstrated. Parameter count alone does not determine performance, and the model’s architecture, training methods, data, computing efficiency and real-world results will ultimately matter.
What is already clear is that Alibaba intends to compete at the infrastructure level of the global AI industry.
With a planned model reaching up to 10 trillion parameters, a new AI accelerator and a target of more than 20 GW of data-center capacity by 2032, Alibaba is positioning itself for a much larger role in the next phase of artificial intelligence.
The global AI race is increasingly becoming a race over models, chips and computing infrastructure—and Alibaba is attempting to compete across all three.
About the Author
Anam Younas
Editor of Daily Press Release
I write about technology, AI, business, finance, and global news, bringing readers clear insights into the latest trends and developments.
For more informative content Follow the Daily Press Release channel on WhatsApp: https://whatsapp.com/channel/0029Vb8gRyoIHphMqA6T2w0U
