Alibaba Group unveiled a detailed strategic roadmap to monetise artificial intelligence at its annual Apsara Conference, held from Tuesday to Wednesday. The Chinese tech giant outlined a focused approach to capitalising on AI technologies, emphasising pragmatic execution amid a significant increase in capital expenditure.
Alibaba CEO Eddie Wu Yongming highlighted three main pillars driving the company’s transition into what he described as the “machine intelligence” era: advanced AI models, proprietary chips, and cloud infrastructure. Central to the presentation was the launch of the Zhenwu V900, branded as China’s most powerful AI processor, alongside plans for the Yinian 720 and 730 server CPUs scheduled for release in 2027. The firm also introduced the Panjiu supernode server, designed to scale clusters up to 500,000 chips, while hinting at future systems capable of handling models with 5 trillion to 10 trillion parameters. Upgrades to multimodal AI models, which process and generate data beyond text, were also announced.
To support these AI initiatives, Alibaba Cloud set an ambitious goal to operate over 20 gigawatts (GW) of global data centre capacity by 2032. Analysts noted that this expansion would be a key enabler for the company’s broader AI and cloud ambitions. Cathy Chan, an analyst at CCB International, described Alibaba’s approach as a shift from last year’s speculative enthusiasm toward a more results-oriented strategy focused on infrastructure efficiency. Alibaba plans to treat AI tokens as utilities, utilising in-house chip technology and cloud architecture to lower inference costs, while releasing open-source model weights to keep developers engaged within its ecosystem.
At the application level, Alibaba introduced various AI-powered products and services, including Qwen Intelligence—a solution aimed at smartphone manufacturers to create agents capable of operating across multiple apps. Other offerings unveiled at the conference included Model-as-a-Service platforms such as Bailian and Qwencloud, the productivity tool QwenWork, and Accio Work, which targets cross-border commerce. The company also debuted new hardware like the AI-integrated QwenBook laptop, AI glasses, earbuds, and a digital note-taking device.
The aggressive infrastructure buildout has prompted analysts to revise their estimates for Alibaba’s capital expenditures significantly upward. Alicia Yap, head of Pan-Asia internet research at Citi, projected expenditure in fiscal years 2027, 2028, and 2029 to reach 258 billion yuan (approximately HK$301.4 billion), 283 billion yuan, and 282 billion yuan respectively. Estimates from Citi Securities suggest that achieving a data centre capacity of 20GW could enable around US$170 billion in annual external cloud revenue, supporting Alibaba’s goal to surpass US$100 billion in combined external cloud and AI revenue by 2030.
Industry observers view Alibaba’s integration of AI models, proprietary chips, and cloud services as bolstering the company’s prospects for sustainable growth in the expanding AI market. This strategy aligns with broader trends in China’s AI infrastructure sector, where Morgan Stanley forecasts spending of approximately 8.5 trillion yuan on AI-related infrastructure between 2026 and 2030, with data centre power capacity expected to triple from 26GW in 2023 to 81GW by 2030.