Great Powers Competition in AI Domain 

In January, 2025, the Chinese startup Deepseek based in Hangzhou, China and founded by Liang Wenfeng, released its first AI assistant Deepseek-R1.  This release not only created a stir in tech domain but impacted strategically. The AI assistant was not merely an addition to existing Generative AI (GenAI) artifacts but in fact it challenged the established players like US based OpenAI, Google and Microsoft etc. Thus confronting the US prowess in the realm of GenAI.

The release of first GPT model in September 2022 by OpenAIand subsequently participation of other tech giants like the Google, the Microsoft etc. In GenAI development paved the way for US dominance in this domain.

Building block of GenAI revolution is Large Language Models. These models are built on enormously large data. Thus development of GenAI assistants requires enormous computing. It is perceived that around 10,000 NVIDIA Graphic Processing Units (GPUs) were used for first version of ChatGPT.  These GPUs are mainly responsible to perform complex tasks related to neural networks on huge amount of data.

The latest version of OpenAI’s ChatGPT-4 was trained using more than 20,000 NIDIA GPUs. Similarly AI models from Google and Microsoft were created by involving huge amount of processing power which in turn requires enormous exchequer. Thus giving a new rise to computing hardware industry. NVIDIA proved to be main beneficiary, due to their efficient hardware which is perfectly aligned with GenAI core technologies of neural networks

Contrary to this enormous spending by US companies, the Deepseek came forward with an economized version of AI assistant while using less number of computing resources and efficient data processing with improved algorithms. It is estimated that only 2000 GPUs were used for the development of DeepSeek-R1. It is interesting to note that R1 model was released on January 20, 2025—the day of President Trump’s second inauguration.

DeepSeek has cemented its reputation as the top frontier AI research lab in China and caused a reassessment of assumptions about the landscape of global AI competition.

By January 27, DeepSeek’s iPhone app had overtaken OpenAI’sChatGPT as the most downloaded free app on Apple’s U.S. App Store. The stock prices of some U.S. tech companies were tumbled, including the AI chip designer Nvidia, which lost more than $600 billions off its valuation in a single day. Resultantly, investor interest in Chinese tech companies has grown significantly.

Initially Generative AI realm was dominated by US. The reason for this this dominance was advancements in deep learning, mostly made by US companies. The deep learning is a sub domain of Artificial intelligence. These advancements are result of enormous computing and capabilities to process huge data.

In a significant achievement, during 2017, the scientists at Google presented their seminal work on transformer architecture known as attention mechanism. It’s a sequence-to-sequence model capable of converting sequences from one domain into sequences in another domain. For example, translating French sentences to English sentences. The original transformer architecture consists of two parts: an encoder and a decoder. The encoder converts the input text (e.g., a French sentence) into a representation, which is then passed to the decoder. The decoder uses this representation to generate the output text (e.g., an English translation) autoregressively. 

GenAI is also known as democratic AI due to its acceptance and usage by common man. Unlike traditional AI which was mostly used in research labs, GenAI leverages a common person with creativity and generates synthetic data. Further advancements in GenAI are transforming it into more impactful Agentic AI which is bound to influence and disrupt traditional working methods. Bill Gates, the founder of Microsoft, in his recent interview predicted that within next ten years many professions including teaching, healthcare and hospitality would be taken over by AI agents. These AI agents would be more efficient and economically cost effective.

Entry of China in the domain of GenAI is in accordance with international relations theory of great powers competition. Scholars of international relations suggest that great powers tend to enter into a competition to maximize their influence and elevate their position at international order. 

Post-cold war a unipolar world order prevailed for two decades, which ended with the peaceful rise of China as global competitor to challenge US led unipolar world order. Prevalent world order is not merely a competition in military domain but great powers are competing in economic, digital and climate change domains to project their influence and maximize their gains.

China faced Chip export control restrictions by successive US regimes. These restrictions were aimed to curtail its capability to compete in emerging domain of Artificial intelligence. However China has progressed remarkably. At a meeting between Chinese Community Party (CCP) Chairman Xi Jinping and Chinese technology executives (including DeepSeek CEO Liang Wenfeng) in February this year, Huawei founder Ren Zhengfeitold Xi that now they are ready to achieve self-sufficiency in semiconductor value chain. This means great powers competition in AI domain will likely to accentuate.

Progress in AI domain is a human pinnacle to train the machine for own benefit and sustenance. It is envisaged that the great powers competition in AI domain will positively impact human society as artifacts like Deepseek will continue to emerge.