DeepSeek Surpassed: AI Competition Shifts - News Directory 3

DeepSeek Surpassed: AI Competition Shifts

February 24, 2025 Catherine Williams Business
News Context
At a glance
  • The race for dominance in the field of generative AI is intensifying, with new players and technological advancements rapidly reshaping the landscape.
  • Shinichi Aoki: The battle for the leading camp of generative AI (artificial intelligence) is changing rapidly.
  • "Although it is an important advance for AI, it is not a revolution." Ben Gotzell, the authoritative AI research, commented on DeepSeek's technology.
Original source: zh.cn.nikkei.com

The Evolving Landscape of Generative AI: A Battle for Supremacy

Table of Contents

The US xAI led by Musk announced on February 17 that it will start offering “Grok 3”.

The race for dominance in the field of generative AI is intensifying, with new players and technological advancements rapidly reshaping the landscape. Chinese startup DeepSeek initially stunned the world, but American companies have quickly regained their footing, leveraging their vast resources and talent pools. As methods and ideas become more transparent, new competitors are emerging, driving a more intense competition and accelerating technological progress.

DeepSeek’s Impact and Limitations

Shinichi Aoki: The battle for the leading camp of generative AI (artificial intelligence) is changing rapidly. Chinese emerging company DeepSeek shocked the world, but was soon overtaken by American companies. There has not changed the situation where companies with more funds, talents and resources have the advantage. On the other hand, as ideas and methods are revealed, new competitors are expected to increase. The competition for technology development will be more intense, and technological progress will also accelerate.

“Although it is an important advance for AI, it is not a revolution.” Ben Gotzell, the authoritative AI research, commented on DeepSeek’s technology. DeepSeek released the model “V3” that processes language and operates dialogue applications in December 2024, and began to provide “R1” with powerful inference functions such as programming and mathematics in January 2025.

R1 is believed to develop performance that matches the “o1” officially released by OpenAI in December 2024 at low cost. Although there are differences in cost estimates, it is epoch-making to use graphics processors (GPUs) to the limits and to improve performance by using AI to explore repeatedly while improving performance. Many researchers and technicians highly praised it. The “distillation” method that uses data output from existing generative AI to create new models also brings efficient development.

Ben Gotzell

So far, the more learning data and computing resources, the higher the performance of generative AI. It shows that even if resources are lacking, high-performance AI can be developed, which has made great contributions. Nevertheless, Gotzel points out that “the destructive paradigm shift has not been reached.”

DeepSeek's AI model

The Role of Computing Resources

In the development of cutting-edge AI, abundant computing resources remain a critical factor. For instance, the new AI model “Grok3,” led by entrepreneur Elon Musk, began offering its services on February 17, showcasing superior capabilities in mathematics, science, and programming. This advancement is largely due to a significant expansion of the data center, which allows the model to learn with more than ten times the computing power of its previous iteration. Computational power is essential for reasoning and solving complex problems.

The upcoming “GPT-5” model from OpenAI, expected to be released within a few months, is anticipated to have both extensive knowledge capabilities and high inference capabilities. The combined AI is expected to be massive, requiring a large number of the highest-performance GPUs.

American Dominance and Future Prospects

American technology companies are expected to study DeepSeek’s approach and leverage their rich computing resources for the next phase of development. In terms of talent and capital, the United States’ advantages remain unchanged. To drive the development of “General Artificial Intelligence (AGI)” beyond human intelligence, significant investment is essential.

If distillation methods are used, it is technically possible to catch up with cutting-edge generative AI. While U.S. technology companies may take measures to prevent DeepSeek’s distillation, distillation remains a crucial topic in AI research. Even if DeepSeek stagnates, the trend of emerging companies and scientists developing low-cost, high-performance generative AI independently will not stop.

AI technology is likely to make significant progress in the future. Revolutionary breakthroughs will emerge, and if a path to become smarter with less learning data, similar to human learning, is found, the competitive environment will change significantly.

Recreating DeepSeek’s Technology

DeepSeek has open-sourced most of its creativity and technology, inviting researchers and technicians worldwide to verify and build upon their work. A team at the University of California, Berkeley, announced that they have improved reinforcement learning and reproduced the core technology of DeepSeek for just $30. While third-party verification is necessary, Jiayi Pan, who leads the research, expressed his expectation that “if generative AI can be developed at low cost, the technology will develop further.”

The competition for international development of generative AI is becoming increasingly fierce (Reuters)
The competition for international development of generative AI is becoming increasingly fierce (Reuters).

The Future of AI Development

In the past, even innovative ideas struggled to materialize without capital. In the future, low-cost, high-performance AI that builds on DeepSeek’s methods may continue to develop. Many experts draw parallels to the “Cambrian explosion” on Earth around 500 million years ago, where diverse organisms rapidly evolved. The development of AI technology is expected to accelerate, with innovations in performance and technology.

The development competition for generative AI has entered a new phase. While striving to achieve the acceleration of the most cutting-edge huge AI, efforts to develop small, high-performance systems are also advancing, further strengthening polarization. In this fierce competition for survival, the next generation’s preferred technology will emerge, but predicting its origin is challenging.

As the AI landscape evolves, it is clear that the battle for supremacy will continue to intensify. Companies like DeepSeek and Grok3 are paving the way for future innovations, and the U.S. remains a dominant force in this technological arms race. The future of AI is bright, and the possibilities are endless.

The Evolving Landscape of Generative AI: A Battle for Supremacy

Q&A style Article

What is the current state of the generative AI landscape in 2025?

Answer:

The generative AI landscape in 2025 is marked by intense competition and rapid technological advancements. New players, such as the Chinese startup DeepSeek, have initially made notable impacts but have been quickly overtaken by American companies, which leverage their vast resources and talent pools. As ideas and methods become more transparent, more competitors are emerging, leading too more intense competition and accelerated technological progress.The launch of Grok 3 by xAI, led by Elon Musk, and the anticipated release of GPT-5 from OpenAI are examples of advancements shaping the field.

What role does computing power play in the growth of cutting-edge AI?

Answer:

Computing power is crucial in the development of cutting-edge AI. Models like Grok 3,which showcase superior capabilities in mathematics,science,and programming,owe their advancements to significant expansions in data centers,allowing the models to learn with exponentially more computing power than previous iterations. Computational power is essential for AI’s ability to reason and solve complex problems. Similarly, the upcoming GPT-5 by OpenAI is expected to require a massive number of high-performance GPUs to support its extensive knowledge capabilities and high inference capacities.

How has the competition between DeepSeek and American companies unfolded?

Answer:

DeepSeek initially made a considerable splash in the AI world with their release of the V3 and R1 models, demonstrating remarkable progress at lower costs compared to equivalents from established American companies. However,American tech giants have quickly regained dominance by leveraging their larger resources,talent,and capital.Despite the advantages, DeepSeek’s innovative approach has stimulated competition and highlighted the potential of distillation techniques in AI development, prompting further research and investment from U.S. companies.

What is the significance of DeepSeek’s distillation method in AI research?

Answer:

DeepSeek’s distillation method uses data output from existing generative AI to create new models, marking an efficient development approach.This technique has been praised for its ability to match the performance of high-end models at a lower cost and is seen as a crucial area in AI research. Even if U.S. companies are taking measures to prevent DeepSeek’s techniques from being directly adopted, the trend of developing low-cost, high-performance AI will likely continue.

how are emerging companies and scientists contributing to the AI landscape?

Answer:

Emerging companies and scientists are driving the AI landscape forward by focusing on developing low-cost, high-performance generational AI independently. Researchers like those at the University of California,Berkeley,have successfully reproduced DeepSeek’s core technologies at low costs,highlighting how ideas and techniques can be democratized and leveraged widely. This encourages a broader, more diverse innovation pool, accelerating AI advancement.

What are the future prospects for AI development?

Answer:

The future of AI development appears promising, with expectations of significant breakthroughs, especially if methods are found to create smarter, yet less data-reliant models. These developments would parallel how humans learn, possibly shifting the competitive surroundings dramatically. Drawing parallels to the “Cambrian explosion,” experts anticipate a period of rapid diversification and innovation in AI technologies. Both large and small-scale AI systems are expected to evolve, and while the next preferred technology remains unpredictable, the evolution will likely favor those who can rapidly adapt and innovate.

How are American technology companies positioned for the future of generative AI?

Answer:

American technology companies, backed by superior funding, talent, and computational resources, remain at the forefront of the generative AI race. They are expected to analyze and incorporate aspects of DeepSeek’s approaches to further their development efforts,particularly towards achieving General Artificial Intelligence (AGI),which exceeds human intelligence. Significant investments and innovations are required to maintain dominance and lead future AI advancements.

What can we learn from DeepSeek’s approach to technology development?

Answer:

DeepSeek’s approach, emphasizing cost-effective innovation and the open-sourcing of technology, offers valuable lessons in democratizing AI development. By making their technology widely accessible, they encourage widespread advancement and experimentation. this methodology promotes broader scientific collaboration and spurs accelerated progress by allowing independent verification and advancement globally.

How does the evolution of generative AI relate to past technological revolutions?

Answer:

The rapid evolution of generative AI mirrors past technological revolutions, drawing comparisons to the “Cambrian explosion” where diverse life forms rapidly evolved. In AI development, companies like DeepSeek and Grok3 are leading the way, setting the stage for a similar explosion in AI capabilities and innovations.This environment encourages rapid development, diversity in approaches, and significant advancements across the field.

By considering these questions and answers, we capture a comprehensive understanding of generative AI’s current state and future trajectory, considering its intense competition, significant technological advancements, and the role of resource-rich entities in shaping the landscape.

references

The Evolving Landscape of Generative AI: A Battle for Supremacy

Q&A Style Article

What is the current state of the generative AI landscape in 2025?

Answer:

The generative AI landscape in 2025 is marked by intense competition and rapid advancements. DeepSeek, a Chinese startup, initially caused a major stir but has since been overtaken by american companies with larger resources and talent pools. Notable advancements shaping the field include the launch of “Grok 3” by xAI, led by Elon Musk, and the anticipated release of “GPT-5” by openai. As AI methods and techniques become more obvious, the emergence of new competitors accelerates technological progress and intensifies the competition.

What role does computing power play in the growth of cutting-edge AI?

Answer:

Computing power is absolutely critical to the advancement of cutting-edge AI. For example, “Grok 3” showcased superior capabilities in mathematics, science, and programming by utilizing critically important data center expansions, allowing more than tenfold the computing power of its predecessor. Similarly, it’s anticipated that “GPT-5” from OpenAI will require a massive deployment of high-performance GPUs to support its vast knowledge and inference capabilities, underscoring the importance of computational resources for advanced AI development.

How has the competition between DeepSeek and American companies unfolded?

Answer:

DeepSeek gained attention by releasing the “V3” and “R1” models, which offered performance competitive with those of established American companies but at a lower cost. Despite the initial impact,American tech giants have quickly reclaimed their position of dominance by leveraging superior resources,talent,and funding. DeepSeek’s innovative distillation methods have spurred further research and investment from U.S. firms, highlighting the competitive and collaborative dynamics within the AI industry.

What is the meaning of DeepSeek’s distillation method in AI research?

Answer:

DeepSeek’s distillation method efficiently uses data from existing AI models to create new ones, allowing high-end model performance at reduced costs. This technique has received widespread praise and is considered pivotal in AI research. Even with U.S.companies taking measures to integrate or counteract this method, its role in developing cost-effective AI models is crucial and continues to influence research directions.

How are emerging companies and scientists contributing to the AI landscape?

Answer:

Emerging companies and scientists play crucial roles in driving AI innovation by focusing on low-cost, high-performance generative AI models. An example is a team at the University of California, Berkeley, which replicated deepseek’s core technology for only $30, demonstrating the potential for democratizing AI advancements. This approach broadens the innovation landscape and accelerates progress by making AI research and development more accessible.

What are the future prospects for AI development?

Answer:

AI development is expected to see significant breakthroughs, especially if the industry can develop smarter, less data-dependent models, akin to human learning. Experts draw parallels to the “Cambrian explosion” for the potential rapid diversification and innovation in AI technologies. The focus is on adapting both large and small-scale AI systems, with the prevailing developments leveraging foundations laid by methods such as those pioneered by DeepSeek.

How are American technology companies positioned for the future of generative AI?

Answer:

American technology companies, with their abundant funding, talent, and computational resources, are poised to maintain leadership in generative AI development. They are expected to integrate and capitalize on approaches like those of DeepSeek, advancing towards developing General Artificial Intelligence (AGI) that surpasses human capabilities. Continued investment and innovation will be essential for these companies to sustain and extend their dominant position.

What can we learn from DeepSeek’s approach to technology development?

Answer:

DeepSeek’s approach underscores the importance of cost-effective innovation and open-sourcing technology. By making their technologies widely accessible, DeepSeek fosters broader scientific collaboration and accelerates technological advancement worldwide. This strategy exemplifies how clarity and accessibility can democratize AI development and spur rapid innovation.

How does the evolution of generative AI relate to past technological revolutions?

Answer:

The rapid evolution of generative AI is comparable to past technological revolutions, akin to the “Cambrian explosion” in biological history, where diverse forms of life rapidly evolved. In AI, pioneers like DeepSeek and Grok3 are setting the stage for a similar spike in AI capabilities and advancements, facilitating rapid innovation, diversification of approaches, and significant technological breakthroughs.

By considering these questions and answers, we gain a comprehensive understanding of generative AI’s present dynamics and future trajectory, encompassing competition, technological advancements, and the significant role of well-resourced entities in shaping the AI landscape. This evergreen perspective highlights the continuous evolution and potential of AI technology.

Keep reading