Chinese start-up DeepSeek is threatening to upend the global order in technology with the release of its most recent AI models, which it claims are on par with or better than industry-leading models in the US for a fraction of the price, reports Breaking News.
After stating in a paper last month that the training of DeepSeek-V3 cost less than $6 million (€5.7 million) in computational power from Nvidia H800 processors, the start-up has gained interest in international AI circles.
With the help of DeepSeek-V3, DeepSeek’s AI Assistant has surpassed competitor ChatGPT to claim the top spot among free apps in the US App Store, reports Breaking News.
Shares of many major tech firms, notably Nvidia, have suffered as a result, and questions have been raised about the rationale behind several US tech companies’ choice to commit billions of dollars in AI investment.
Here are some facts about the firm that is revolutionising the AI industry globally, reports Breaking News.
Chinese IT companies hurried to develop their own AI-powered chatbots once OpenAI’s ChatGPT was released in late 2022 (see below).
However, the disparity in AI skills between US and Chinese companies caused significant dissatisfaction in China with the release of the first Chinese ChatGPT counterpart, created by search engine behemoth Baidu, reports Breaking News.
This story has completely changed due to DeepSeek’s models’ high quality and affordability. According to the Chinese start-up, the two models, DeepSeek-V3 and DeepSeek-R1, which have received high accolades from Silicon Valley leaders and engineers at US tech companies, are comparable to OpenAI and Meta’s most sophisticated models.
They’re also less expensive to use. According to a post on DeepSeek’s official WeChat account, the DeepSeek-R1, which was published last week, is 20–50 times less expensive to operate than the OpenAI o1 model, depending on the assignment, reports Breaking News.
However, several people have openly voiced doubts about DeepSeek’s success.
He stated that 50,000 Nvidia H100 chips would not be made public as doing so would be against Washington’s export restrictions, which forbid the sale of such cutting-edge AI processors to Chinese businesses. A request for comment on the accusation was not immediately answered by DeepSeek, reports Breaking News.
In a research note released on Monday, Bernstein analysts pointed out that DeepSeek’s V3 model training expenses were unknown overall but significantly more than the $5.58 million (€5.3 million) the business claimed was spent on processing power.
According to Chinese corporate documents, Liang Wenfeng, a co-founder of the quantitative hedge fund High-Flyer, is the majority shareholder of DeepSeek, a start-up located in Hangzhou, reports Breaking News.
In March 2023, Liang’s fund said on its official WeChat account that it was “starting again” and focussing its resources on establishing a “new and independent research group, to explore the essence of AGI” [Artificial General Intelligence], moving beyond trading. Later in the year, DeepSeek was developed.
ChatGPT creators AGI is defined by OpenAI as autonomous systems that outperform humans in the majority of economically significant tasks, reports Breaking News.
Tell us your thoughts in the Facebook post and share this with your friends.
