Huan Shi Cai Jing
-HUANSHICAIJING-
Editor's Note: "Hanguang Jiji makes NVIDIA nervous." This is how the Malaysian TechWire Asia website evaluated the Chinese high-tech company Hanguang Jiji, which has seen a significant rise in stock price recently. On the opening day, Hanguang Jiji's stock price fell 4% to 1524 yuan per share. Just yesterday, Hanguang Jiji's stock price rose sharply by 15.73%, closing at 1587.91 yuan per share, surpassing Guizhou Maotai, which had long held the position of "King of A-shares" (closing price 1446.1 yuan per share), and for the first time, its closing price ranked first among A-shares. This is not the first time that Hanguang Jiji has become the "King of A-shares". On the 27th, during the trading session of the A-share market, Hanguang Jiji's stock price once reached 1464.98 yuan per share, briefly surpassing Guizhou Maotai. Hanguang Jiji's performance has made the topic of whether Chinese chip manufacturers are capable of challenging NVIDIA a focus of domestic and international public opinion.
Over 100% increase this month
As of the close of the A-share market on Thursday, Hanguang Jiji's total market value reached 664.3 billion yuan. It is worth noting that Hanguang Jiji's stock price increased by more than 100% this month, and the cumulative increase since 2023 has exceeded 2800%.
On the evening of the 28th, Hanguang Jiji issued a risk warning announcement regarding stock trading, stating that the stock price may be detached from the current fundamentals, reminding investors to invest rationally and pay attention to investment risks.
On the 28th, Hanguang Jiji's stock closed at 1587.91 yuan per share (Visual China)
Public information shows that Hanguang Jiji was established in 2016, and its predecessor was a 10-person academic team set up by the Institute of Computing, Chinese Academy of Sciences in 2008 to explore the intersection of processor architecture and AI. The company focuses on the research and development of artificial intelligence chips and technological innovation.
As the "first stock of domestic AI chips," Hanguang Jiji experienced a long period of losses after its listing, and it finally achieved quarterly profit in the fourth quarter of 2024.
Previously, the company attributed its losses to high R&D investment, with annual R&D investment often being 1 to 2 times the revenue of the same period, which conforms to the characteristics of high investment and long cycle in the AI chip industry.
Data from Hanguang Jiji's semi-annual report released on the 26th showed that the company achieved operating income of 2.881 billion yuan in the first half of the year, an increase of 4347.82% compared to the same period last year; the net profit attributable to shareholders reached 1.038 billion yuan, achieving a turnaround from loss to profit, while it was a net loss of 530 million yuan in the same period last year; the net cash flow from operating activities reached 911 million yuan, while it was -631 million yuan in the same period last year.
In the financial report, Hanguang Jiji stated that the current revenue growth was mainly due to the company's continuous expansion of the market and active assistance in the implementation of artificial intelligence applications. In the first half of the year, Hanguang Jiji still maintained a high level of R&D investment, increasing by 2% to 456 million yuan, but because the revenue growth far exceeded the R&D expenditure, the proportion of R&D investment to revenue decreased to 15.85%.
Goldman Sachs recently released a research report, expressing further confidence in Hanguang Jiji's performance, and raised the company's target stock price to 1835 yuan per share. According to the latest target stock price released by Goldman Sachs, Hanguang Jiji's total market value will approach 770 billion yuan.
Goldman Sachs said that Hanguang Jiji continues to innovate in AI chip technology, the demand for domestic AI chips in China is constantly increasing, and the capital expenditure on cloud computing in China is increasing, and the continued promotion of diversified chip platforms, as well as the continuous increase in R&D investment of Hanguang Jiji, these factors are the important reasons for the adjustment of the target stock price.
The South China Morning Post called Hanguang Jiji "the small NVIDIA". It is reported that the performance of Hanguang Jiji's latest Siyuan 690 is close to NVIDIA's H100, and its performance will be better than NVIDIA's China-specific H20.
Ray, the research director of Futurum Group, a market research company, said that the rapid rise in Hanguang Jiji's stock price not only reflects the company's current growth, but also highlights the huge expectations of investors for its potential.
The changes in regulatory and geopolitical environments, combined with the growth of AI demand in the domestic market and the optimization of domestic hardware by DeepSeek, are all driving the improvement of optimistic sentiment.
"Policy + Demand" Dual Drive
Hanguang Jiji's rapid development is inseparable from the rising heat of the domestic chip industry. Liu Chunsheng, associate professor at the School of International Economics and Trade, Central University of Finance and Economics, who was interviewed by Global Times, said that the Chinese artificial intelligence chip industry is in a stage of rapid development.
Since the United States initiated the trade war against China, China has long faced technical constraints in the chip field, and is currently actively promoting localization. The Chinese AI chip industry is in an explosive period driven by "policy + demand".
A report by Deloitte China titled "Technology Trends 2025" shows that global AI chip sales exceeded 57 billion US dollars in 2024, and the market size of the next generation of AI chips is expected to exceed 150 billion US dollars this year, reaching 400 billion US dollars by 2027.
At the same time, the development of the domestic AI chip market has significantly accelerated. Bernstein's "2025 China Chip Industry Report" shows that the sales of domestic AI chips increased from 6 billion US dollars last year to 16 billion US dollars, with market share rising from 29% to 42%, with a growth rate of 112%, almost three times that of foreign chips.
Bloomberg said that Hanguang Jiji's excellent performance shows that as AI R&D accelerates, more and more Chinese companies, from small startups to large enterprises like Alibaba, are adopting domestic chips instead of NVIDIA products. In the context of tense geopolitical relations and supply chain disruptions, the sentiment of domestic chip manufacturers has been boosted.
Ma Cheng, head of Shenzhen Juzhe Investment Management Co., Ltd., told Bloomberg that in the face of U.S. restrictions on China's AI industry, government support for leading domestic enterprises is crucial for promoting industrial growth and promoting chip localization. Hanguang Jiji's growth is definitely not temporary. Government support has also boosted the overall market.
In the recent stock market boom in China, chip stocks led the way, highlighting investors' increasingly optimistic outlook on China's AI development prospects.
"Whether it's the chip sector or other sectors rising, the underlying power is the upward trend of the entire Chinese stock market," said Song Hongbing, a world financial historian and author of the "Currency War" series, who was interviewed by Global Times on the 28th. He said that the A-share market has returned to a ten-year high, and the overall market atmosphere has changed. Regarding this round of stock market rise, he believes one of the main reasons is that the country has taken a series of measures to inject confidence into the market.
According to the data from Securities Times · Databao, there are currently 24 stocks in the A-share market that have invested in AI chips. Based on the semi-annual reports already disclosed, according to the semi-annual report, performance快报 and the lower limit of performance forecasts, about two-thirds of the AI chip stocks with published performance have achieved a year-on-year increase in net profit.
This year's Spring Festival, the domestic AI large model DeepSeek gained widespread attention for its "low cost and high performance." Last week, Hangzhou Deep Seek Company released the DeepSeek-V3.1 version, which uses the UE8M0 FP8 parameter precision format.
In the top comment under the Weibo post, Deep Seek Company stated that UE8M0 FP8 is "designed for the next-generation domestic chips about to be released," stimulating expectations for a new round of development of domestic AI chips.
TrendForce's Jutong Consulting estimates that the proportion of AI server markets in China that purchase chips from NVIDIA and AMD (Advanced Micro Devices) will decrease from around 63% in 2024 to around 42% in 2025, and Chinese chip suppliers will be almost equal to the proportion of imported chips under the support of local AI chip policies.
Xiang Ligation, chairman of the Zhongguancun Information Consumption Alliance, who was interviewed by Global Times on the 28th, said, "In the field of artificial intelligence chips, China is continuously promoting the construction of intelligent computing centers, and there is a clear demand for inference chips in the market. The overall demand for chips in China remains strong. If this market is long-term dominated by international companies such as NVIDIA, domestic companies will find it difficult to gain space."
Currently, several GPU (Graphics Processing Unit) R&D companies have emerged in China, including Hanguang Jiji, Huawei, and Suiyuan. Some companies have already launched actual products and formed preliminary layouts.
Liu Chunsheng said that NVIDIA chips are currently facing security concerns, such as the unresolved vulnerabilities and backdoor risks of the H20 chip. Domestic chips have a natural advantage in terms of security and controllability, making them more trustworthy for domestic enterprises.
Xiang Ligation said that the domestic market is vast, but whether companies can seize opportunities depends on their own strength—whether they can provide stronger chips, ensure stable supply, and meet the needs of industrial development.
"China seems to be getting less afraid of being 'strangled' by the United States"
The UK Financial Times reported on Wednesday that Chinese chip manufacturers plan to double their AI chip production next year. SMIC also plans to double its 7nm capacity next year.
At the same time, Chinese companies are also accelerating the development of next-generation AI chips that are compatible with DeepSeek standards. In this context, Chinese chip design companies such as Hanguang Jiji, Muxi, and Biren are expected to get more capacity allocation from SMIC.
A senior executive of a Chinese chip manufacturer said that if successful in developing and optimizing Chinese chips, and training and running their own models within the continuously developing Chinese ecosystem, looking back on this transformation process in the future, it will be found to be a more significant "DeepSeek moment", and this unprecedented ecosystem collaboration helps to make up for the shortcomings of Chinese enterprises in hardware.
The Financial Times believes that the previous testing of DeepSeek on a small batch of Chinese domestic chips indicates that it is completely feasible to use Chinese chips for model training.
As more and more domestic chips are adapted to the DeepSeek standard and their performance continues to improve, and as Chinese AI software and hardware companies strengthen cooperation, China may challenge NVIDIA's global leading position.
With the rapid rise of domestic AI chips in China, the competitive situation in the relevant fields between China and the United States has also changed. "Deutsche Welle" website analyzed that in the field of AI chips, China seems to be increasingly unafraid of being "strangled" by the United States.
In recent years, the progress of Chinese domestic chips has been great, and many domestic GPUs have even exceeded the H20 in hardware, although the software ecosystem is not as complete as NVIDIA. Currently, companies such as Huawei are already working hard to strengthen the construction of the software ecosystem.
Ren Zhengfei, founder of Huawei, spoke about the current state of China's chip industry in June this year, saying that single-chip technology in China is still one generation behind the United States, but through "mathematics compensating for physics, non-Moore compensation for Moore, using group computing to compensate for single chips, the results can also reach practical conditions."
He also said, "There's no need to worry about the chip issue. Through methods such as stacking and clustering, the computational results can be equivalent to the most advanced level... Software cannot strangle the neck."
However, Xiang Ligation reminded that after years of technical accumulation, domestic companies have certain foundations in chip design, but they still face challenges in the manufacturing process. Domestic chips need to achieve process breakthroughs from 7nm to 5nm, 3nm, and build an independent and controllable manufacturing system.
Additionally, converting chips into commercial products requires establishing a complete system integration capability and software ecosystem. Companies like Huawei have already launched open platforms, but more industry chain enterprises need to participate collaboratively, gradually building a mature ecosystem.
Liu Chunsheng said, "The profitability of Hanguang Jiji proves that domestic chips can complete a commercial loop in specific scenarios, but in the long term, they still face five major shortcomings: advanced process is constrained, EDA (Electronic Design Automation) toolchain is incomplete, shortage of high-end talent, weak software ecosystem, and low internationalization."
Xiang Ligation believes that the key problem facing the current Chinese chip industry is the breakthrough in manufacturing capabilities and the improvement of the supply chain. Only through coordinated development along the entire chain can domestic chips gradually mature.
In this process, although Hanguang Jiji has certain technical foundations, it still needs long-term technological accumulation and industrial integration to reach the industry status of NVIDIA.
Sources | Global Times
Review | Yang Shasha
Edit | Yang Shuyu
Check | Ma Mengyang
Click to turn pages | Turn pages
Original: https://www.toutiao.com/article/7543885647765963274/
Statement: This article represents the views of the author and readers are welcome to express their opinions by clicking on the [Top/Down] buttons below.