IN January 2025, the Chinese startup DeepSeek, based in Hangzhou and founded by Liang Wenfeng, released its first AI Assistant, DeepSeek-R1.
This release not only created a stir in the tech domain but also had strategic implications.
DeepSeek-R1 was not merely another addition to existing Generative AI (GenAI) offerings—it directly challenged established players such as US-based OpenAI, Google, and Microsoft, thereby confronting American dominance in the realm of GenAI.
The release of OpenAI’s first GPT model in September 2022, followed by the participation of tech giants like Google and Microsoft, laid the foundation for US supremacy in this domain.
At the heart of the GenAI revolution are Large Language Models (LLMs).
These models are built using massive datasets and require tremendous computing power.
It is estimated that around 10,000 NVIDIA Graphic Processing Units (GPUs) were used to develop the first version of ChatGPT.
These GPUs perform complex tasks involving neural networks and large-scale data processing.
The latest version, ChatGPT-4, was trained using more than 20,000 NVIDIA GPUs.
Similarly, AI models from Google and Microsoft also required vast processing power and financial resources, giving a major boost to the computing hardware industry.
NVIDIA has emerged as a primary beneficiary, thanks to its efficient hardware architecture aligned with GenAI’s neural network technologies.
In contrast to the enormous expenditure by US companies, DeepSeek introduced a more economical AI Assistant by using fewer computing resources and optimizing data processing through improved algorithms.
Remarkably, only around 2,000 GPUs were used for the development of DeepSeek-R1.
It is noteworthy that the R1 model was launched on January 20, 2025—the day of President Trump’s second inauguration.
DeepSeek has quickly earned recognition as a frontier AI research lab in China, prompting a reassessment of global AI dynamics.
By January 27, DeepSeek’s iPhone app had overtaken OpenAI’s ChatGPT as the most downloaded free app on Apple’s US App Store.
Consequently, stock prices of some US tech companies dropped sharply.
Notably, NVIDIA lost over $600 billion in market valuation in a single day, signalling a shift in investor confidence towards Chinese tech firms.
Initially, the GenAI landscape was dominated by the US, largely due to its leadership in deep learning—a subfield of artificial intelligence.
These advancements were driven by superior computing power and access to massive datasets.
A pivotal moment came in 2017 when Google researchers introduced the transformer architecture, based on an attention mechanism.
This architecture, designed for sequence-to-sequence tasks such as translating text between languages, consists of two parts: an encoder that processes the input (e.g., a French sentence) and a decoder that generates the output (e.g., an English sentence) autoregressively.
GenAI is also referred to as “democratic AI” because of its accessibility and utility for ordinary users.
Unlike traditional AI, which was confined to research labs, GenAI empowers individuals to express creativity and produce synthetic content.
The next frontier—Agentic AI—is set to revolutionize professional workflows.
In a recent interview, Bill Gates predicted that in the next ten years, many professions—especially in teaching, healthcare and hospitality—will be significantly transformed or even overtaken by AI agents, which offer greater efficiency and cost-effectiveness.
China’s entry into GenAI aligns with international relations theory on great powers competition, which holds that dominant states inevitably compete to expand influence and elevate their status in the global order.
Following the Cold War, a unipolar world order prevailed for nearly two decades, led by the US.
This order is now being challenged by the peaceful rise of China, which seeks to assert itself in various spheres—not just militarily but also in economic, digital and climate domains.
Despite US-led chip export controls aimed at curtailing China’s AI capabilities, Beijing has made significant strides.
In February 2025, during a meeting between President Xi Jinping and leading Chinese tech executives—including DeepSeek CEO Liang Wenfeng—Huawei founder Ren Zhengfei declared that China is now ready to achieve self-sufficiency in the semiconductor value chain.
This development suggests that great powers competition in AI is set to intensify further.
Progress in the AI domain represents a human milestone—training machines for societal benefit and sustainability.
It is envisaged that this ongoing competition among great powers will ultimately benefit humanity, as breakthrough innovations like DeepSeek continue to emerge.
—The writer is contributing columnist.
