China's Alibaba Tests New AI Chip Amid Nvidia Restrictions.

Alibaba is testing a new AI chip built by a Chinese manufacturer, aimed at handling a wider range of AI inference work. This comes as Beijing pressures Chinese tech companies to reduce reliance on Nvidia's H20 chip, following a partial resumption of sales last month. Alibaba's cloud revenue jumped 26% in the June quarter.

Alibaba (BABA) is testing a new AI chip built by a Chinese manufacturer, aiming to handle a wider range of AI inference work. This development comes as Beijing pressures Chinese tech companies to reduce reliance on Nvidia's H20 chip, following a partial resumption of sales last month. Alibaba's cloud revenue jumped 26% in the June quarter, underscoring the company's growing importance in the AI cloud market [1].

The new chip, designed to be more versatile than Alibaba's earlier AI processor, is built by a Chinese manufacturer. It is intended to handle a broader range of AI inference tasks, according to the report [1]. This move aligns with China's broader push for semiconductor self-sufficiency, as companies like Huawei, Tencent-backed Enflame, and Cambricon also advance homegrown solutions [5]. Alibaba's new chip is already interoperable with Nvidia's PyTorch and TensorFlow platforms, easing the transition for developers [4].

Alibaba's strategic pivot toward AI inference chips and its $53 billion investment in AI and cloud infrastructure signal a profound shift in the U.S.-China tech rivalry [2]. By developing a domestically produced inference chip, Alibaba aims to mitigate reliance on foreign semiconductors like Nvidia’s H100 and Blackwell series, which are restricted in China due to U.S. export controls [1]. This move not only strengthens Alibaba Cloud’s market position but also accelerates the adoption of its open-sourced Qwen 3 model [1].

Nvidia faces vulnerabilities in this evolving landscape. U.S. export restrictions have limited its access to China’s $2.7 billion AI cloud market in 2024, while its H20 chip, the most powerful version permitted for sale, has drawn scrutiny over security risks [5]. Chinese officials reportedly recommend halting its use, creating a vacuum Alibaba and others are poised to fill [3]. Additionally, Nvidia’s supply chain is exposed to geopolitical tensions, as its reliance on TSMC for advanced manufacturing contrasts with Alibaba’s shift to domestic production [4].

The implications for global AI supply chains are significant. China’s $98 billion AI investment in 2025—$56 billion of it government-funded—highlights a national strategy to build scalable, high-performance infrastructure [2]. This includes Huawei’s Ascend service, which reduces model training times, and Baidu’s Qianfan platform, which cuts training costs by 90% [4]. For investors, the rise of Chinese chipmakers like Cambricon and Huawei presents opportunities in alternative semiconductor solutions, while Alibaba Cloud’s expansion into Singapore and Japan underscores the potential for cross-border AI infrastructure growth [1].

In conclusion, Alibaba’s inference-focused chip and $53 billion investment mark a strategic reorientation in the U.S.-China tech rivalry. By leveraging cloud-first AI infrastructure and interoperable hardware, Alibaba not only challenges Nvidia’s dominance but also accelerates China’s path to semiconductor self-reliance. For investors, the key opportunities lie in AI cloud infrastructure, alternative chipmakers, and the broader ecosystem of open-source collaboration driving efficiency and innovation in China’s AI sector.

References:
[1] https://finance.yahoo.com/news/alibaba-tests-ai-chip-china-144131404.html
[2] https://www.ainvest.com/news/alibaba-ai-chip-strategy-impact-global-ai-semiconductor-landscape-2508/
[3] https://www.roic.ai/news/nvidia-shares-slip-on-report-alibaba-develops-new-ai-chip
[4] https://www.eweek.com/news/alibaba-ai-chip/
[5] https://www.cnbc.com/2024/09/17/chinese-companies-aiming-to-compete-with-nvidia-on-ai-chips.html