TLDRs;
Contents
- Alibaba unveils a new AI inference chip offered through its cloud platform rather than direct sales.
- The chip builds on 2019’s Hanguang 800, strengthening Alibaba’s domestic AI capabilities amid U.S. restrictions.
- Heavy AI infrastructure spending strained Alibaba’s cash flow but fueled growth in cloud and AI services.
- China’s push for chip independence is accelerating, with Alibaba, Baidu, and Cambricon leading the charge.
Alibaba Group is deepening its push into artificial intelligence with the development of a new inference chip designed to power advanced AI applications.
The chip, built under Alibaba’s semiconductor arm T-head, signals the Chinese tech giant’s latest attempt to strengthen domestic chip capabilities and reduce reliance on U.S. technology amid rising geopolitical tensions.
The chip will not be sold directly to customers but instead offered through Alibaba Cloud’s computing services, a model that allows enterprises to rent processing power rather than purchase hardware. This approach reflects a broader industry trend in China, where companies are pivoting toward integrated chip-as-a-service models.
New chip builds on Hanguang legacy
Alibaba first entered the inference chip space in 2019 with the release of its Hanguang 800, which was praised for accelerating image recognition and search capabilities.
The new chip aims to advance those capabilities further, focusing on inference, the stage where AI models apply learned knowledge to perform real-world tasks.
While companies like Nvidia have dominated global AI hardware sales, U.S. export restrictions have limited access to cutting-edge chips in China. Alibaba’s latest release is part of a larger domestic movement to ensure supply chain resilience and maintain competitiveness in AI deployment.
Chip-as-a-service model gains traction
Unlike traditional chipmakers who sell hardware directly, Alibaba has chosen to retain control of its chip technology while monetizing it through cloud rentals.
This model has already proven profitable: Alibaba Cloud reported 26% year-over-year revenue growth, while AI-related products posted triple-digit growth for eight consecutive quarters.
Industry observers note that this strategy not only ensures technological control but also integrates hardware directly into Alibaba’s cloud ecosystem, creating a recurring revenue stream that may prove more sustainable than one-time chip sales.
AI spending impacts Alibaba’s cash flow
The expansion of Alibaba’s AI infrastructure has come at a steep financial cost. In its second-quarter 2025 results, the company reported revenue of $34.6 billion, up 2% year-on-year.
However, adjusted EBITA fell 14% due to heavy spending, particularly on Taobao’s instant commerce initiatives and cloud infrastructure.
Operating cash flow dropped nearly 40%, and free cash flow swung into negative territory at $2.6 billion. Executives attributed the decline to investments in cloud capacity and AI development, including the buildout of domestic chip technology. Despite short-term financial strain, Alibaba emphasized that these investments are necessary to capture long-term market share in AI and cloud services.
China accelerates semiconductor independence
Alibaba’s chip development comes at a critical moment for China’s semiconductor sector. Other domestic players, such as Baidu with its Kunlunxin chips and Cambricon with its AI processors, are also reporting record orders and profit growth. Baidu’s Baige 5.0 platform, powered by homegrown chips, has already secured contracts worth more than $139 million.
These advances underscore the resilience of China’s technology industry in the face of U.S. export bans, which recently blocked Nvidia’s H20 chips from entering the Chinese market. By accelerating domestic R&D, Chinese firms are not only filling the gap but also achieving measurable performance gains.
