China’s DeepSeek: A Turning Point in Global Tech Competition
Abigail Darwish and Aryamehr F | 12 February 2025
Summary
The launching of a Chinese AI-powered chatbot, DeepSeek, has considerable implications for AI competition, offering an effective yet low-cost alternative to US models.
Its open-source and autonomous learning approach enhances innovation but also introduces security vulnerabilities and ethical concerns.
US efforts to limit China’s AI growth remained ineffective, as DeepSeek’s success with limited resources challenges US dominance in the tech industry.
The recent launch of the Chinese Artificial Intelligence (AI) firm DeepSeek has had a significant impact on both global markets and broader technological competition. On 27 January, financial markets experienced a USD 1 trillion downturn following the release of DeepSeek’s advanced AI model, R1. Its remarkably low cost—just a fraction of United States (US) rivals’ prices—combined with its advanced capabilities comparable with OpenAI’s ChatGPT, triggered this market reaction. Its creation also carries far-reaching implications for the global tech landscape, highlighting China’s growing dominance in AI research and development whilst also driving the commercialisation and widespread adoption of generative AI.
Comparison and Implications
A key defining feature of DeepSeek’s approach is its open-source model, which facilitates innovation by easing the identification and resolution of issues by an external community. However, this openness also renders it more vulnerable to cyberattacks and data breaches, as illustrated by the recent DeepSeek data leak, which exposed over a million lines of log data–including chat histories, secret keys, and other sensitive information.
Moreover, DeepSeek’s AI operates with a more autonomous learning framework, compared to OpenAI’s models, which require significant human supervision and reinforcement learning from human feedback to refine performance. This allows the model to improve based on its iterative processes, reducing dependency on human intervention. Whilst this approach accelerates self-learning and scalability, it introduces higher risks from increased susceptibility to adversarial attacks or biases that may emerge without direct human oversight. This autonomy within DeepSeek is a key component that can feed into quantum computing, robotics and even military technology.
Paradoxically, the ongoing ‘Chip War’, wherein the US has attempted to limit China’s AI sector by imposing export controls on US semiconductors since 2022, may have inadvertently facilitated DeepSeek’s emergence. Indeed, without access to leading American chip technology, the Chinese startup was forced to optimise the performance of less sophisticated NVIDIA H800 chips compared to the more powerful H100 chips used in OpenAI’s GPT models. However, despite the gap in capabilities, DeepSeek’s R1 has demonstrated superior performance. This further underscores the efficiency and innovation of DeepSeek’s AI development, which could set new industry benchmarks. US officials are also currently investigating the prospect of DeepSeek having bypassed American export restrictions through third-party purchases of semiconductors in Singapore, highlighting further that export controls have been ineffective.
Although the US maintains a lead over emerging technology, China's rapid advancements challenge the US’ standing and raises uncertainty over the future of global tech leadership. President Donald Trump has claimed that DeepSeek is a “wake-up call” for US companies, emphasising “competing to win.” Others have echoed this sentiment, highlighting that “complacency” over US leadership, both within Washington and amongst its allies, jeopardises the country’s long-term standing and competitive edge.
The US’ emphasis on quality over quantity in technological research and development allowed American firms to justify premium pricing and dominate global markets at the expense of accessibility of such products. However, DeepSeek’s breakthrough challenges this long-standing perception, suggesting that the US may have overestimated the absolute necessity of high-cost, high-quality resources whilst underestimating the role of ingenuity and innovation. DeepSeek’s success demonstrates that strategic innovation and adaptability can, in some circumstances, outweigh sheer resource investment, potentially reshaping how the global tech industry measures progress and competitiveness.
Solen Feyissa/Unsplash
Forecast
Short-term
It is very likely that US tech firms will increase investment in research, efficiency, and accessibility to remain competitive. Meta CEO, Mark Zuckerberg, has since said that DeepSeek will ‘benefit’ the company and the future of its AI capabilities.
It is likely that strategies in the ‘Chip War’ will need to be adapted as export controls are ineffective in reducing the capabilities of other countries
Long-term
It is very likely that other non-Western countries will develop their own models of Artificial General Intelligence (AGI).
It is possible that China could eclipse the US’ technological advantage should the current trajectory of development continue.