# AI: A Catalyst for a New Era of Tech Competition

Artificial Intelligence (AI) has become an indispensable part of daily life. Since the release of OpenAI’s ChatGPT-3.5 in November 2023, the AI market has entered a significant growth phase. In 2024, the United States announced the $500 billion ‘Stargate’ AI infrastructure project, signaling its intent to lead in AI innovation. However, China is not lagging, ramping up its efforts in the AI technology race.

Shortly after the Stargate announcement, Liang Wenfeng, an engineer from Hangzhou, China, launched the AI model DeepSeek R1, causing a stir in the global market. DeepSeek R1 quickly climbed to the top of app store charts, boasting performance on par with existing big tech AI models but developed at 95% lower costs. This news led to a 20% drop in NVIDIA’s shares, wiping out billions in market capitalization. Although OpenAI rebutted the claims, DeepSeek’s innovative approach has garnered significant attention in the AI industry.

# U.S. vs. China: Intensifying AI Supremacy Battle

China’s commitment to AI technology development is long-standing. In 2017, the Chinese Communist Party unveiled the ‘New Generation AI Development Plan,’ aiming to become a leading AI nation by 2030. By 2024, it had rolled out the ‘AI Capacity-Building Action Plan for Good and for All,’ focusing on narrowing the global AI gap and strengthening AI capabilities.

The Chinese government has been driving AI development in collaboration with private enterprises. Since 2017, it has supported 15 major AI companies, including Alibaba, Baidu, Tencent, and Huawei, through the ‘National AI Team.’ Startups like DeepSeek are also fostering under this strategic AI development agenda.

Conversely, the U.S. is integrating AI innovation with the cryptocurrency and digital assets market. With the appointment of ‘Crypto Czar’ David Sacks, the fusion of AI and blockchain technology is accelerating. The Stargate project underscores America’s efforts to dominate AI infrastructure. This AI supremacy battle has evolved beyond corporate competition into a strategic ‘AI Race’ between the U.S. and China.

# DeepSeek R1: Innovative Technology Drawing Industry Attention

DeepSeek R1 stands out due to its unique learning techniques. According to research papers, DeepSeek uses two core technologies to maintain high performance while reducing costs:

– **Iterative Training:** The AI model reviews its outputs, learns from errors, and improves performance, cutting computational costs compared to traditional large-scale data training.

– **Mixture of Experts (MoE) Architecture:** The model divides into several ‘expert’ groups, activating specific groups based on queries. This method reduces unnecessary computation, maximizing energy efficiency while maintaining high performance.

These techniques are seen as revolutionary, significantly lowering the immense computational resources and cost burdens typically associated with traditional AI models.

# DeepSeek’s Cost Controversy: Is the 95% Reduction Real?

The claim that DeepSeek R1 was developed at 95% lower costs compared to existing AI models is contentious. While DeepSeek V3’s training costs were approximately $5.5 million, the exact development cost for R1 remains undisclosed. OpenAI has suggested that DeepSeek might have utilized the ‘Knowledge Distillation’ technique, learning from ChatGPT’s data, which could violate OpenAI’s service agreements.

Knowledge distillation allows a smaller model to improve by learning from the output data of a larger model, maintaining high performance while reducing costs. If DeepSeek leveraged ChatGPT’s data, it would breach OpenAI’s terms of service.

Moreover, in the heated global AI competition, there’s a possibility that DeepSeek and the Chinese government may have underreported the actual training costs. By emphasizing lower costs, they could strategically unsettle Western AI market investors and diminish competitors’ market value.

The crucial aspect of this AI competition is not just cost reduction but the acceleration toward global AI supremacy.

# The Battle for Computational Resources: Essential for AI Advancement

As AI technology advances, the demand for massive computational resources (such as high-performance GPUs) escalates. The era of Artificial General Intelligence (AGI) or AI Superintelligence will amplify the importance of computing resources.

In this context, decentralized computing networks are gaining attention. Projects like io.net (IO) offer a GPU network spanning over 130 countries, providing AI developers with fast and affordable computational resources. Io.net claims it can train AI models at 90% lower costs compared to traditional cloud services. Efficiently securing and utilizing computational resources will be a critical factor in the AI technology race.

The emergence of DeepSeek R1 indicates that the AI supremacy contest is expanding into a strategic national rivalry. As AI technology evolves, key issues will center around securing efficient computational resources, reducing costs, and optimizing AI model performance.

With the U.S. and China striving for AI industry leadership through projects like Stargate and the National AI Team, decentralized computing infrastructures might become a significant variable. Whether the AI market will progress through collaboration and innovation or polarize amidst competitive pressures remains uncertain. However, one thing is clear: AI is poised to become one of the most transformative technologies for humanity. The nation and companies that dominate AI will shape future industry landscapes, influencing lives based on current strategic choices.