Alibaba has announced a series of updates to its full-stack artificial intelligence strategy, covering foundation models, AI chips, agentic cloud infrastructure and AI-powered mobile platforms, as the company seeks to expand the use of AI across industries.
The announcements were made at the Apsara Conference, Alibaba Cloud’s annual technology event. The updates include progress on the next-generation Qwen model family, new multimodal AI models, proprietary computing chips from T-Head, an agentic cloud architecture and Qwen Intelligence, an AI agent platform designed for smartphones.
“The theme of this year’s Apsara Conference is ‘Intelligence Goes Beyond’. Over the past few years, AI has continuously expanded our imagination of technological capabilities. AI possesses vast potential for development–it can be deployed and scaled in real-world scenarios, boosting productivity across thousands of industries. This is the true meaning of ‘Intelligence Goes Beyond’: guiding AI from technological breakthroughs toward value creation.”
– Joe Tsai, Chairman, Alibaba Group
“Today, the total volume of Machine Thinking is less than 3% of all Human Thinking. If that volume eventually scales to 1,000x human capacity, the simple math tells us: Machine Thinking still has an enormous growth runway. As machines are becoming the primary force behind Thinking, turning intelligence into a commodity supplied at scale, the truly groundbreaking products of the Machine Intelligence era have not yet arrived. With this in mind, our target is that by 2032, the global data center capacity operated by Alibaba Cloud will surpass 20GW, fueling the industry’s exponentially rising demand for AI.”
– Eddie Wu, CEO, Alibaba Group
Alibaba said its next-generation Qwen 4 model is currently in training, with Qwen 4.5 and Qwen 5 planned to scale to between 5 trillion and 10 trillion parameters. The company also highlighted progress in Recursive Self-Improvement, with Qwen3.8-Max completing 33 automated improvement cycles over more than a month and increasing its Artificial Analysis score from 40 to 45.
In a chip design experiment, Alibaba said the model conducted more than 60 hours of self-improvement across the design lifecycle and made more than 10,000 electronic design automation tool calls. The resulting production-grade chip bus modules reduced chip area by 42% without compromising performance, according to the company.
Alibaba also introduced updates across its multimodal model portfolio. Qwen3.8-LiveTranslate is designed for simultaneous interpretation, while Qwen-Audio-3.1-TTS-Next can generate cinematic soundscapes combining dialogue and ambient sounds from text scripts. The company also announced updates to its speech recognition, text-to-speech and real-time interaction models, alongside Qwen-Image 3.1 for creative design and e-commerce applications.
The company has also launched Qwen Intelligence, a full-stack agentic solution for smartphones that gives device manufacturers access to a Qwen-powered agent platform capable of handling complex tasks across applications.
On the hardware side, Alibaba’s chip design unit T-Head unveiled the Zhenwu V900 AI processor for training and inference. Alibaba said the processor delivers three times the performance of the Zhenwu M890, with 216 GB of GPU memory and 1,200 GB/s of inter-chip bandwidth. It supports FP8 and FP4 data precisions and is scheduled for mass production and commercial release in the first quarter of 2027.
T-Head also unveiled an upgraded supernode server integrating the Zhenwu V900 processor, ICN Switch, Panmai SmartNIC and Zhenyue SSD controller chip. The system is designed to support supernode clusters of up to 500,000 cards. Alibaba also announced the roadmap for its next-generation Yitian 720 and Yitian 730 CPUs, which are scheduled for launch in 2027.
Alibaba Cloud meanwhile introduced upgrades around its agentic cloud strategy, structured around three areas: model, harness and context. The architecture comprises AI Native Cloud for large-scale model training and inference, Agent Native Cloud for enterprise AI agent deployment and management, and Context Engine for real-time data and long-term memory.
The AI Native Cloud updates include enhancements to Alibaba Cloud’s Platform for AI, a new Cloud Parallel File Storage system designed for AI workloads, and HPN 8.0 Pro, an AI networking architecture capable of supporting 100 petabits of bandwidth and more than 130,000 network ports operating at 800G speeds.
Under the Agent Native Cloud, Alibaba launched AgentCore, an enterprise platform for building, operating and managing AI agents throughout their lifecycle. The company also introduced Agent Security Center for security and compliance management of enterprise AI agent applications.
The Context Engine includes Agent Context, which provides AI agents with real-time context and long-term memory by connecting enterprise documents, business systems, chat records and multimodal data. Alibaba said the service can reduce token usage by up to 67% in knowledge-intensive applications such as customer service, AI coding and data analytics.
Alibaba also upgraded OpenLake into a unified multimodal data lakehouse supporting structured, semi-structured, unstructured, vector and streaming data. The company said the platform can reduce total costs by 38% and query response times by 40% compared with traditional architectures.
