Alibaba Cloud has fired the latest Salvo with the advantages of artificial intelligence (AI) with numerous new AI models, agent development platforms and cloud infrastructure upgrades to become a full-stack AI service provider.
The announcement, released at the annual APSARA conference in Hangzhou this week, includes an investment of RMB38 billion ($53.4 billion) in AI and cloud infrastructure over the next three years, bringing the release of the most powerful large-scale language model (LLM) with one Trillion parameter.
The company's ambition is to create an ecosystem in which AI models act as a new type of operating system and incorporate into a wide range of devices backed by cloud platforms optimized for all stages of AI development and deployment.
Eddie Wu, Chairman and CEO of Alibaba Cloud Intelligence, said:
The key to the announcement is LLMS' new QWEN3 family. Alibaba Cloud claimed that the performance of the flagship QWEN3-MAX model is comparable to top closed source models of major benchmarks, such as the SWE Bench, which evaluates LLM's ability to solve real software engineering problems.
The company also enhanced multimodal capabilities with QWEN3-VL, a visual language model that can act as a visual agent, manipulate computer interfaces, and generate code directly from visual design.
Another addition is QWEN3-OMNI. It handles text, images, audio and video inputs to provide real-time streaming response response, making it ideal for hands-free interfaces on smart devices and vehicles.
Recognizing that the model alone is not enough, Alibaba Cloud has made a big move to simplify the creation and deployment of AI agents for businesses.
Model Studio, an AI development platform, includes an agent development kit that enables professional developers to create powerful AI agents in high code environments. Business users and people with limited programming skills can use the low-code agent development platform to build lightweight agents.
For the more complex enterprise needs, the company has deployed major updates to AgentBay, a multimodal cloud operational environment, and Lingyang Agentone, a one-stop platform designed to allow organizations to build and integrate production-enabled agents across marketing, customer service and operations.
To enhance these advanced AI workloads, Alibaba Cloud enhances cloud infrastructure, including new vector bucket capabilities for object storage services that integrate raw and vector data management, simplifying and reducing the cost of building search and augmented augmented generation (RAG) applications.
The PolardB database is also enhanced with Compute Express Link (CXL) technology, reducing latency by 72%, improving memory scalability, and is suitable for combining data and AI workloads.
In networking, Alibaba's new HPN 8.0 architecture offers 800gbps of throughput, doubles its previous ability to support ultra-large scale model training. Additionally, the company's cloud threat detection and response services now use AI agents powered by QWEN to automate security operations and improve incident response.
The latest announcement shows that it is clearly intended to compete actively from Alibaba Cloud with global hyperscalers such as Amazon Web Services, Microsoft Azure, and Google Cloud.
Alibaba Cloud currently operates 91 availability zones in 29 cloud regions around the world. We hope to open first data centres in Brazil, France and the Netherlands next year, with additional facilities continuing in Mexico, Japan, Korea, Malaysia and Dubai. We also established regional service centres in Indonesia and Germany, providing multilingual customer support 24/7.
