KUALA LUMPUR, Jan 28 — DeepSeek, hailed as the “biggest dark horse” in the open-source large language model (LLM) space, is being described as China’s secret weapon in the artificial intelligence (AI) battle against the United States, according to the South China Morning Post (SCMP).
The Hangzhou-based start-up has made waves for developing AI models at a fraction of the cost incurred by tech giants such as OpenAI and Meta Platforms. This prompted former US President Donald Trump to call the breakthrough a “wake-up call” for America’s tech sector, describing it as a “positive” development.
At home, DeepSeek’s disruptive potential has been celebrated by Chinese tech leaders, including Zhou Hongyi, chairman of cybersecurity firm Qihoo 360.
Accordinf to the SCMP report Zhou declared in a Weibo video that DeepSeek has “upended the world” following the release of two powerful AI models that were created with fewer computing resources than typically required.
Feng Ji, founder of game developer Game Science, echoed the sentiment, suggesting in a Weibo post that DeepSeek’s advances could influence China’s “national fate” amid its ongoing tech competition with the US.
DeepSeek’s R1 reasoning model, released on January 20, has been compared to OpenAI’s o1 model, which excels in solving complex scientific and mathematical problems.
The company’s AI chatbot, integrated with R1 and launched earlier this month, quickly climbed to the top of Apple’s US App Store, surpassing ChatGPT in popularity, SCMP reported.
DeepSeek’s earlier model, V3, unveiled in December, was reportedly trained in two months at a cost of US$5.58 million (RM25.8 million), a fraction of the resources used by its larger rivals, according to SCMP.
The start-up’s cost-efficient approach underscores the progress of Chinese AI firms despite US sanctions that restrict access to advanced semiconductors. Qihoo 360’s Zhou asserted in his social media post that China is well-positioned to “win the AI war with the US” thanks to DeepSeek’s innovations.
Jefferies equity analyst Edison Lee highlighted in a research note that China’s constraints on chip supplies have led the market to prioritise efficiency in LLM development.
He noted that while DeepSeek’s breakthroughs are impressive, they have not yet translated into monetisation of its AI innovations.
The buzz around DeepSeek’s achievements has shaken global markets, with US tech giants seeing significant stock drops. Nvidia, for instance, lost nearly US$600 billion in market value as its shares plunged by 16.8 per cent on Monday, SCMP reported.
AI expert Gary Marcus, writing on Substack, said DeepSeek’s achievements placed it in the same league as OpenAI and Google in terms of performance and innovation.
“None of this means that China won the AI race or even took the lead,” Marcus wrote.
DeepSeek, spun off from Chinese hedge fund High-Flyer Quant, recently made its R1 model free to use, contrasting with OpenAI’s US$200 monthly fee for its o1 models.
0 Comments