Key Points
- Alibaba’s new chip focuses on AI inferencing, complementing its existing reliance on semiconductors from vendors like Nvidia.
- The e-commerce behemoth has earmarked a staggering 380 billion Chinese yuan ($53.1 billion) for AI investments over the next three years.
- Spurred by geopolitical headwinds, Chinese tech entities are intensifying efforts to develop indigenous chip solutions.
Sources tell CNBC that Alibaba is quietly forging ahead with a new semiconductor tailored for artificial intelligence applications, a move highlighting the growing ambition of China’s tech giants to develop homegrown chip capabilities. The chip, designed specifically for AI inferencing – the execution phase of AI applications – differs from the training phase, which relies on massive datasets to build AI models, according to a source familiar with the project who requested anonymity. The Wall Street Journal initially reported on the development of the unreleased chip.
This push for domestic chip production comes amid heightened U.S.-China trade tensions, which have raised concerns about access to American technology. U.S. chipmaker Nvidia, for example, faced restrictions earlier this year on exports of its H20 product to China, a system engineered with reduced performance to comply with U.S. export controls. While Nvidia has since received the green light to resume some shipments, conditional on paying 15% of revenue to the U.S. government, the company revealed on its recent earnings call that H20 shipments to China have yet to commence.
CNBC’s source indicates that Alibaba will strategically balance its use of in-house chips with those from external vendors like Nvidia. Nvidia’s offerings are versatile, supporting both AI training and inferencing workloads. News of Alibaba’s chip development, coupled with robust earnings, boosted Alibaba’s shares by as much as 12% while Nvidia saw shares dip slightly in early trading, extending recent losses.
Differentiating its strategy from Nvidia, Alibaba will not commercialize its chip directly. Instead, it intends to offer computing power to its customers, leveraging these new chips as part of its cloud infrastructure. Alibaba declined to comment when reached by CNBC.
AI Ambitions Take Flight
Alibaba’s semiconductor design arm, T-head, previously unveiled the Hanguang 800 inference chip in 2019. These chips find application within data center servers, accelerating AI processing demands.
Alibaba’s aggressive investment into AI is already paying dividends. In its latest earnings report, the company highlighted a remarkable 26% year-over-year revenue surge in its cloud division, fueled in part by AI-related products. The company reported that these AI product revenues have experienced “triple-digit year-over-year growth for the eighth consecutive quarter,” signaling the strong demand for AI solutions.
Alibaba joins a growing cohort of Chinese tech firms, including Baidu, in developing proprietary semiconductors. While these companies continue to utilize Nvidia chips where possible, they are increasingly exploring alternatives from domestic players like Huawei, which has emerged as a significant force in China’s AI chip market. Nvidia CEO Jensen Huang has cautioned the U.S. government against restricting chip sales to China, arguing that it risks enabling Chinese firms to fill the void.
Adding fuel to this trend, Cambricon, an AI chip designer based in China, recently reported an astounding 4,000% increase in revenue and record profits for the first half of the year, further underscoring the rapid growth and potential of the Chinese semiconductor industry.
Original article, Author: Tobias. If you wish to reprint this article, please indicate the source:https://aicnbc.com/8292.html
