 "DeepSeek: Has China won 'AI war' against the US or just the first battle?")
DeepSeek, described by some as the “biggest dark horse” in the open-source large language model (LLM) space, is being hailed within China as Beijing's secret weapon in the artificial intelligence (AI) war with the United States (US). But does it ensure Beijing’s victory, or is this just the opening battle? Let’s explore what experts have to say.
The Hangzhou-based company has shocked Wall Street and Silicon Valley by creating AI models at a fraction of the cost incurred by OpenAI and Meta Platforms. This prompted US President Donald Trump to describe the achievement as a “wake-up call” for the US tech industry.
On January 20, DeepSeek announced the public release of its open-source R1 reasoning model, which it claims matches the capabilities of OpenAI’s o1 model. DeepSeek has integrated the R1 model into its AI chatbot, launched on January 10, which has since risen to prominence, claiming the top spot on Apple’s US App Store and surpassing ChatGPT. The chatbot also became the most downloaded free app on Apple’s China App Store as of Monday. These milestones come on the heels of DeepSeek’s December 26 public release of its V3 model. This model, developed over two months with a budget of $5.58 million, utilised significantly fewer computing resources and reportedly outperformed larger tech firms’ LLMs in benchmark tests.
'China will eventually win the AI war with the US'
For their part, Chinese tech leaders and commentators have enthusiastically praised DeepSeek’s disruptive potential, as reported by the South China Morning Post (SCMP).
According to the SCMP, Zhou Hongyi, co-founder, chairman, and chief executive of Chinese cybersecurity firm Qihoo 360, commended DeepSeek for having “upended the world” in a recent video posted on his Weibo account. This remark followed the start-up’s release of two advanced AI models, which were developed at a lower cost and with fewer computing resources compared to what major tech firms typically require for LLM development.
Also Read
In another widely circulated Weibo post, Feng Ji, founder and chief executive of Black Myth: Wukong developer Game Science, highlighted DeepSeek’s potential to reshape China’s “national fate” amid the ongoing tech rivalry with the US.
Referring to the global attention garnered by DeepSeek, Zhou stated in his social media post: “We should have confidence that China will eventually win the AI war with the US.”
DeepSeek’s ability to develop powerful models at a fraction of the cost incurred by larger tech companies highlights the significant strides made by Chinese AI firms, even in the face of US sanctions that restrict access to advanced semiconductors crucial for training LLMs, according to the SCMP report.
LLMs are the foundational technology behind generative AI platforms like OpenAI’s ChatGPT.
DeepSeek's breakthrough fuels market disruption and sparks US concerns
"China is the only market that pursues LLM efficiency owing to chip constraints," Jefferies equity analyst Edison Lee reportedly wrote in a research note on Monday. He highlighted that the Trump administration is likely to recognise that imposing further restrictions could "force China to innovate faster".
DeepSeek’s AI model breakthrough over the weekend triggered significant turbulence in the stock market on Monday, raising doubts about the US' technological dominance. Shares of major US AI and semiconductor companies plummeted amid fears that Chinese firms might surpass them in the high-stakes race for supremacy.
For example, American semiconductor giant Nvidia saw its stock price plunge by as much as 25 per cent on Monday before closing at $118.58, marking a 16.8 per cent drop and wiping out nearly $600 billion in market value. Similarly, in Europe, tech stocks led market losses, with shares of chip equipment maker ASML Holding falling by more than 8 per cent.
“Re-evaluating computing power needs could cause 2026 AI capex to fall or not grow,” Lee reportedly added in the Jefferies research note.
DeepSeek, which originated as a spin-off from Chinese hedge-fund manager High-Flyer Quant, recently made its R1 reasoning model free to use, a stark contrast to OpenAI’s o1 models, which cost $200 monthly for unlimited access.
'The race for AI Supremacy is over, but...'
US entrepreneur and AI expert Gary Marcus reportedly declared in an essay published on the online platform Substack on Monday: “The race for ‘AI Supremacy’ is over, at least for now, and the US didn’t win.”
Marcus, co-author of Rebooting AI: Building Artificial Intelligence We Can Trust, highlighted that DeepSeek’s development of an AI model “that required only roughly 1/50th of the training costs of previous models” instantly elevated the start-up “into the big leagues with American companies like OpenAI, Google, and Anthropic, both in terms of performance and innovation.”
However, Marcus noted, “None of this means, however, that China won the AI race or even took the lead. American companies will incorporate these new results and continue to produce new results of their own.”
Still, he credited China with “some style points” for achieving so much “without hundreds of thousands of Nvidia H100s,” the advanced graphics processing units that the US has restricted from being exported to Chinese firms.