Chinese LLMs lead the reconstruction of AI development paradigm

Editor's note: Belunn Se, a special commentator on current affairs for CGTN, is a senior industry observer based in Shenzhen, China. The article reflects the author's opinions and not necessarily the views of CGTN.

The Artificial Intelligence (AI) Action Summit, held in Paris from February 10 to 11, has brought together leaders and tech executives from over 100 countries to discuss the promotion and safe application of artificial intelligence. Among the key topics was the rise of Chinese large language models (LLMs), particularly DeepSeek, which is reshaping the global AI landscape. This shift marks a pivotal moment in AI development, as Chinese models challenge traditional paradigms and introduce innovative approaches that could redefine the future of the industry.

A new era in AI development

DeepSeek has captured global attention since its release. Its success lies in its innovative "low-cost algorithms + open source" model, which breaks away from the traditional "high-investment computing power + closed source" approach. This shift represents a watershed moment in AI development, demonstrating that sophisticated AI capabilities can be achieved through algorithmic innovation rather than sheer computational power.

Historically, AI advancements were driven by scaling laws, which emphasized the need for massive computing resources. However, DeepSeek's strong reasoning capabilities and performance, achieved through smart optimization rather than raw computing power, challenge this assumption. This breakthrough not only reduces costs but also opens up new possibilities for AI applications, making the technology more accessible and sustainable.

Transforming the global AI landscape

With technology restrictions and computing infrastructure upgrades reshaping the AI landscape, China is making strides through its independent open-source ecosystem and extensive applications. DeepSeek's rise signifies a shift away from dependence on high-end chips, presenting new opportunities for AI advancement globally.

On January 21, U.S. President Donald Trump announced the Stargate AI infrastructure plan. Backed by OpenAI, SoftBank, and Oracle, the initiative involves a $500 billion investment in next-generation AI infrastructure, including advanced data centers and energy facilities. This move marks the beginning of a new AI technology arms race.

The central debate in AI development now revolves around the relevance of the Scaling Law. If computing power and algorithmic bottlenecks are overcome, a fundamental paradigm shift in AI development may be imminent.

The rise of DeepSeek and similar Chinese models is reshaping the global AI competitive landscape. Traditionally, AI development was dominated by high-end chips and closed-source technologies, creating significant barriers to entry. However, DeepSeek's open-source approach and low computing costs are democratizing AI development, enabling more countries and companies to participate in the AI revolution.

This shift is particularly significant in light of ongoing geopolitical tensions and technology blockades. For instance, on January 13, the U.S. government imposed stricter regulations on AI chip exports, restricting China's access to advanced chips below 16nm. This policy, however, has inadvertently spurred domestic technological breakthroughs. On February 1, Huawei Ascend became the first Chinese AI chip to support DeepSeek models. Within a week, over 15 domestic AI chip firms announced compatibility with DeepSeek, accelerating China's AI computing independence.

DeepSeek optimizes compatibility with domestic chip instruction sets, supports local operating systems and integrates with hardware acceleration. By collaborating closely with domestic chip manufacturers, DeepSeek enhances AI performance while promoting technological autonomy. This shift strengthens China's AI industrial chain and ensures greater information security.

The power of open source

One of DeepSeek's most notable contributions to the AI industry is its commitment to open source. By open-sourcing its models, DeepSeek has lowered technical barriers, fostered collaboration among developers and accelerated technological iteration. This approach not only promotes innovation but also helps build an open and shared technology ecosystem.

Open source is becoming a defining trend for Chinese AI companies, positioning them as leaders in the global open-source ecosystem. In contrast, many U.S. tech giants continue to keep their core technologies proprietary. OpenAI CEO Sam Altman recently acknowledged this shift, admitting that his company "has been on the wrong side of history" regarding open source.

For the industry, DeepSeek's open-source strategy has far-reaching implications. It facilitates the integration of domestic chips and AI frameworks, enhances supply chain autonomy, and provides cost-effective AI solutions for small and medium-sized enterprises. This, in turn, drives overall technological progress and strengthens the competitiveness of the AI industry.

Redefining AI governance and ethics

As AI technology advances, it brings new ethical and security challenges, including issues related to data quality, algorithmic fairness and model transparency. These challenges underscore the need for robust global AI governance frameworks.

China is actively contributing to this effort by promoting multilateral cooperation, technology sharing and policy coordination. Under the frameworks of international organizations such as the United Nations, China has proposed the "Global AI Governance Initiative," emphasizing core values such as security, people-centered development and fairness. Additionally, China is working to establish internationally recognized AI technical standards and hosting platforms like the World Artificial Intelligence Conference to facilitate knowledge exchange and capacity building.

However, differing approaches to AI governance between Western countries and China present both challenges and opportunities. While Western nations often prioritize stringent regulations, China advocates for a balance between competition and cooperation. By seeking common ground, countries can work together to address global AI challenges and ensure that the benefits of AI are shared equitably.

The future of AI: Collaboration and innovation

The reconstruction of the AI development paradigm represents more than just a technical shift – it signals a fundamental change in how AI technology will be developed, distributed and governed globally. DeepSeek's success, combining algorithmic efficiency with open collaboration, suggests a future where AI development becomes more accessible, sustainable and inclusive.

As the global AI competition intensifies, collaboration will be key to addressing shared challenges. Developed nations should support developing countries in building their AI capabilities, ensuring that the benefits of AI are not concentrated in a few regions. At the same time, countries must work together to establish transparent, fair and secure AI governance frameworks that mitigate risks and promote ethical practices.

The rise of Chinese LLMs like DeepSeek marks a turning point in the AI industry. By challenging traditional paradigms and embracing open collaboration, these models are paving the way for a more inclusive and sustainable future for AI. As the world navigates this transformative era, global cooperation and innovation will be essential to ensuring that AI truly benefits humanity.

(If you want to contribute and have specific expertise, please contact us at opinions@cgtn.com. Follow @thouse_opinions on X, formerly Twitter, to discover the latest commentaries in the CGTN Opinion Section.)