TMTPOST -- The American depository receipts (ADRs) of Alibaba Group surged as much as 10.5% and finished 8.2% higher. Shares hit their highest close since October 2021 after Alibaba leadership showed their ambitious artificial intelligence (AI) bets.

Credit：Alibaba

Alibaba would increase spending on AI and cloud infrastructure beyond the RMB380 billion ($53 billion) it had committed in February, to be deployed over the next three years, said the CEO Eddie Wu at the annual Apsara Conference in Hangzhou on Wednesday. Wu stopped short of the new spending target, while emphasizing that growing demand for AI services and the rapid evolution of the technology warranted a more aggressive approach. 

“The AI industry has developed much faster than we expected, and the industry’s demand for AI infrastructure has also far exceeded our anticipations,” Wu said. "We are actively proceeding with the 380 billion (yuan) investment in AI infrastructure, and plan to add more."

Wu suggested the investment in Alibaba’s cloud computing power will boost exponentially.He disclosed a target that the annual energy consumption of global datacenters under Alibaba Cloud in 2032 will be 10 times as large as the consumption in 2022. 

Wu said global spending on AI computing capacity is likely to reach $4 trillion within five years, with consolidation leaving only “five or six” super cloud computing platforms worldwide. By pointing to Qwen — Alibaba’s open-source large language model that has become the most widely adopted globally — Wu signaled the company’s intention to be among those survivors. Alibaba Cloud positioned  itself as a “leading full-stack AI services provider”,offering everything from raw computing power to advanced models, according to Wu.

During the event, Alibaba announced a parternership with Nvidia Corporation on physical AI. Alibaba Cloud will integrate Nvidia’s Physical AI software stack, providing enterprise users with a comprehensive platform service encompassing data preprocessing, simulation data generation, model training and evaluation, reinforcement learning for robotics, and simulation testing. The With Nvidia’s AI development tools, Alibaba Cloud’s Platform for AI (PAI) can significantly shorten the development cycle for applications such as embodied intelligence and autonomous driving.

Alibaba unveiled upgrades across its AI ecosystem, including its largest and most capable large language model (LLM) to data, Qwen3-Max. The model is trained on 1 trillion parametrs and its  instruct version was said surpasses GPT-5-Chat. Other launches included the Qwen3-VL vision-language model, improvements to its multimodal Wan and Fun model series, and new tools such as Bailian, a cloud-based studio designed to support AI agent development.

Alibaba Cloud also unveiled plans to open its first data centers in Brazil, France, and the Netherlands in the coming year, and plans to build more of the same in Mexico, South Korea, Japan, Malaysia, and Dubai for next year. It will also set up new regional service centers in Indonesia and Germany to support customers in those areas.

Wall Street analysts saw Alibaba can keep the stock rally from its AI boost.CFRA analyst Angelo Zino reiterated a buy call for Alibaba’s U.S. stock in a note Wednesday. Alibaba's AI spending could restrain free-cash-flow, but "we think it will help further accelerate cloud and total revenue growth," Zino wrote.

“Additional demand is emerging from the education and healthcare sectors, as well as from companies developing tools that leverage Alibaba’s open-source model for artificial intelligence training,” Morningstar analyst Chelsey Tam wrote in a note. “We believe the shares are undervalued, as the market has yet to fully reflect Alibaba’s AI cloud potential and the current management’s ability to enhance competitiveness.”