Nandini Roy Choudhury, writer
In brief
- China is experiencing a significant shift in its AI landscape, driven by the adoption of open-source models, particularly highlighted by DeepSeek’s R1 model, which challenges the dominance of American technology and promotes faster innovation and broader adoption of AI solutions.
- Major Chinese tech companies, including Baidu, Alibaba, and Tencent, are moving towards open-source strategies, with Baidu planning to release its AI models for free and other firms following suit, indicating a broader corporate strategy shift away from proprietary licensing.
- The rise of open-source AI in China is seen as a response to the competitive pressure from American firms like OpenAI, with analysts suggesting that this trend could lead to faster product innovation and a more accessible AI ecosystem, reminiscent of the “Android moment” in the mobile industry.
Detailed news
Some people believe that China is experiencing a “Android moment” for the artificial intelligence industry as a result of its adoption of open-source AI models, which market watchers and industry insiders claim is increasing AI adoption and innovation in the country.
DeepSeek, an artificial intelligence firm, has been at the forefront of the open-source trend. The R1 model that DeepSeek released earlier this year posed a threat to the supremacy of American technology and raised issues about the tremendous expenditure that Big Tech has been doing on large language models and data centers.
The most significant impact that DeepSeek has had, according to some analysts, has been in accelerating the adoption of open-source artificial intelligence models. This is despite the fact that R1 caused a stir in the industry due to its performance and claims of cheaper prices.
“DeepSeek’s success demonstrates that open-source strategies can lead to faster innovation and broad adoption,” stated Wei Sun, chief analyst of artificial intelligence at Counterpoint Research. He also mentioned that a big number of companies have followed the concept.
She went on to say that “now, we see that R1 is actively reshaping China’s AI landscape,” and that “large companies like Baidu are moving to open source their own LLMs as a strategic response.”
Baidu made available to individual users the most recent version of its artificial intelligence model, Ernie 4.5, as well as a new reasoning model, Ernie X1, on March 16. Both of these models are free to use. Beginning at the end of June, Baidu intends to make the Ernie 4.5 model series available as open-source software.
Experts believe that Baidu’s ambitions to open-source software are indicative of a larger shift in China’s corporate strategy, which is moving away from a strategy that is centered on proprietary licensing.
“Baidu has always been very supportive of its proprietary business model and has been vocal against open-source,” Lian Jye Su, chief analyst with technology research and advisory group Omdia, said in a previous interview with CNBC. “However, disruptors such as DeepSeek have demonstrated that open-source models can be just as competitive and reliable as proprietary ones.”
Comparing open-source and proprietary systems
In general, the term “open-source” refers to software in which the source code is made freely accessible on the internet for the purpose of modification and redistribution.
Prior to the development of DeepSeek, there were already AI models that referred to themselves as open-source. In the United States, two prominent examples of such models are Meta’s Llama and Google’s Gemma. On the other hand, there are specialists who believe that these models are not truly open source because their licenses restrict certain uses and modifications, and their training data sets are not available to the public.
DeepSeek’s R1 is distributed under a license known as the MIT License, which Counterpoint’s Sun defines as one of the most permissive and commonly used open-source licenses. This license allows for unrestricted usage, modification, and distribution, including for commercial reasons.
Over the course of the previous month, the DeepSeek team even organized a “Open-Source Week,” during which they disclosed additional technical information regarding the development of their R1 model.
There is no cost associated with DeepSeek’s model; however, the start-up company does charge for its Application Programming Interface (API), which enables the incorporation of artificial intelligence models and their capabilities into the applications of other businesses. The API fees that it charges, on the other hand, are billed as being significantly lower than those that OpenAI and Anthropic are now offering.
Another way that OpenAI and Anthropic earn cash is by charging individual users and businesses for access to certain models that they have already developed. Due to the fact that their datasets and techniques are not available to the general public, these models are referred to as “closed-source.”
The opening of China
In addition to Baidu, other Chinese tech titans such as Alibaba Group and Tencent have been making their artificial intelligence (AI) solutions available for free and are also making more models open source.
As an illustration, Alibaba Cloud announced a month ago that it will be open-sourcing its artificial intelligence models for the development of videos. More recently, Tencent is alleged to have released five new open-source models earlier this month, each of which has the capability to translate text and images into three-dimensional visualizations.
The tendency is also being advanced by players of a smaller scale. A Chinese artificial intelligence company called ManusAI, which recently introduced an AI agent that asserts to be more effective than OpenAI’s Deep Research, has announced that it will move toward open source platforms.
Ji Yichao, one of the co-founders of the company, stated in a video demonstrating the product that “this would not be possible without the amazing open-source community,” which is why the company is dedicated to giving back. “ManusAI functions as a multi-agent system that is powered by several different models, and as a result, we are going to open source some of these models later on in this year,” he added.
One of the most successful artificial intelligence (AI) firms in the country, Zhipu AI, reported on WeChat earlier this month that the year 2025 would be “the year of open source.”
According to Ray Wang, the senior analyst and founder of Constellation Research, who spoke with CNBC, businesses have been pushed to take these actions as a result of the rise of DeepSeek.
Given that DeepSeek is offered at no cost, it is simply not conceivable for any other Chinese competitors to charge for the same service. In order to remain competitive, they are need to transition to open-source business models, as Wang explained.
Kai-Fu Lee, an AI scholar and entrepreneur, is also of the opinion that this dynamic will have an effect on OpenAI. In a recent post on social media, he mentioned that it would be difficult for the firm to justify its pricing when the competition is “free and formidable.”
The most significant discovery made by DeepSeek is that open-source software has emerged victorious, as stated by Lee, whose Chinese startup 01.AI has developed an LLM platform for businesses that are interested in use DeepSeek.
Competition between United States and China
OpenAI, which was the company that sparked the AI craze when it published its ChatGPT bot in November 2022, has not indicated that it intends to move away from its proprietary business model. The organization, which began in 2015 as a charitable organization, is currently transitioning toward a structure that is for profit.
According to Sun, OpenAI and DeepSeek are representative of two very distinct ends of the artificial intelligence landscape. She also discusses the possibility that the industry will continue to be divided between open-source players who develop off of one another and closed-source corporations that have been put under pressure to retain high-cost cutting-edge models.
The large amounts of money that firms like OpenAI have raised have been called into doubt as a result of the open-source movement. Thirteen billion dollars have been invested by Microsoft in the company. According to a confirmation made by CNBC at the end of January, the company is now in discussions to raise up to $40 billion in a capital round, which would bring its valuation to as high as $340 billion.
It was revealed by CNBC in September that the corporation anticipates losses of approximately $5 billion, with revenue estimated to be $3.7 billion. The Chief Financial Officer of OpenAI, Sarah Friar, has also stated that increasing the company’s revenue to eleven billion dollars is “definitely within the realm of possibility” for this year.
On the other side, according to Wang of Constellation Research, Chinese businesses have opted to go the open-source route in order to compete with the more proprietary approach taken by American enterprises. It was also stated by him that “they are hoping for faster adoption than the closed models of the United States.”
While appearing on the program “Street Signs Asia” on CNBC on Wednesday, Tim Wang, managing partner of Monolith Management, a hedge fund that focuses on technology, stated that models from businesses such as DeepSeek have been “great enablers and multipliers in China,” showcasing how things can be accomplished with fewer resources.
According to Wang, open-source approaches have resulted in a reduction in expenses, which has opened the door to product innovation. Product innovation is something that Wang believes Chinese companies have historically been particularly excellent at.
He refers to this development as the “Android moment,” which is a reference to the time when Google’s Android company made the source code for its operating system openly available. This encouraged creativity and development inside the ecosystem of non-Apple related applications.
According to Wang, “We used to think that China was 12 to 24 months behind [the United States] in artificial intelligence, but now we think that it’s probably three to six months behind.”
However, other academics have dismissed the notion that open-source artificial intelligence has to be viewed through the prism of competition between the United States and China. Indeed, a number of businesses in the United States have successfully implemented and benefited from DeepSeek’s R1.
I do not believe that the so-called DeepSeek moment is about whether or not China has superior artificial intelligence to the United States or vice versa. During the CONVERGE conference that took place in Singapore earlier this month, Alibaba Group Chairperson Joe Tsai stated that the power of open-source software is the most important thing.
Tsai went on to say that open-source models make the power of artificial intelligence accessible to everyone, from sole proprietors to major organizations. This will result in an increase in the number of apps that utilize AI, as well as greater development and innovation.
Source : CNBC news
