On Jan 27, BEIJING (Reuters) – The AI sector sees a startling new star, Chinese AI startup DeepSeek, with the latest AI model technology, which the company claims is as good as, or perhaps even better than, the leading competitors in the U.S. but makes use of only a small fraction of the cost. This kind of improvement of progress comes to be a player on the global scene of technology which is reshaping the world.
A month ago, DeepSeek declared through a research paper that the flagship model, DeepSeek-V3, was trained employing less than $6 million worth of Nvidia H800 chips. Difficult to believe but this is very true and the money spent for research and development is very little compared to U.S. hi-tech businesses.
DeepSeek has reached another outstanding milestone, its AI Assistant, which features DeepSeek-V3, has outstripped the ChatGPT to become the most top-rated free app in Apple’s App Store in the U.S. The public may even wonder how it is possible to spend such a tremendous amount of money on a project like this which seems to create more disruptions than it is supposed to give. Nvidia stock has also been hit by DeepSeek’s success.
DeepSeek’s creative method to business model is getting the multinational AI attention and is causing the AI industry to collapse. Below is the information you should know about the startup that is radically transforming the AI industry.
Why DeepSeek Is Shaking Things Up in the AI World
In late 2022, the release of ChatGPT by OpenAI started an unprecedented race among China’s tech companies in developing their chatbots independently. Yet the first reported Chinese equivalent, released by Baidu, did not impress people in China and was way behind U.S.-developed models.
DeepSeek broke this narrative. Two AI models, DeepSeek-V3 and DeepSeek-R1, already gained wide acclamation among the executives in Silicon Valley and US tech engineers; according to DeepSeek, both these models match or even outperform advanced AI systems developed by OpenAI and Meta at a fraction of the cost.
DeepSeek-R1, released just last week, is reported to be 20 to 50 times cheaper to use than OpenAI’s comparable models, depending on the task, as stated in a post on the company’s official WeChat account.
Notwithstanding the accolades, skepticism has accompanied DeepSeek’s rapid rise. Alexandr Wang, chief executive officer of Scale AI, estimated to CNBC in an interview that DeepSeek could have as many as 50,000 Nvidia H100 chips, which would be illegal under U.S. export controls barring the sale of advanced AI chips to Chinese companies. DeepSeek has not responded to the allegations so far.
Analysts at Bernstein further questioned DeepSeek’s claims, as despite reporting $5.58 million worth of computing power for the training of the V3 model, the actual costs of training may well have been manyfold higher. The costs for training R1, similarly lauded, remain unreported.
DeepSeek has become a global disruptor in AI, accomplishing the feat of effectively interlacing quality, efficiency, and affordability while challenging norms and competitors across the globe.
Who Is Behind Deepseek?
DeepSeek, a quantitative hedge fund, with HQ in Hangzhou led by Liang Wenfeng co-founder of-High-Flyer (DeepSeek is an anagram of High-Flyer) The founder is the principal or controlling stockholder of the startup based on Chinese company registers
High-Flyer stated in March 2023 report that it is moving away from tradings to work on AGI-highly autonomous systems that can do all of the world items humans do better-than-and arguably for cheaper than-humans (OpenAI intended the initial AGI here). DeepSeek was born later in that vision of a year.
We don’t know the size of High-flying’s DeepSeek stake, though: This hedge fund is very, very close to the startup—both share a building and one of them has patents for clusters of chips that can be used as the engine-training AI models. It also states, in July 2022 that its AI division had assets and operated a cluster of 10000 up-to-date-nvidia-a100-chips
The rapid rise of DeepSeek in the AI space underlines substantial backing and expertise it draws from its connections with High-Flyer, placing it as a key player in the global AI race.
How Does Beijing View DeepSeek?
DeepSeek’s feat is now known to the highest-ranking officials of the country China. On the day DeepSeek-R1 was launched i.e. on January 20, the company’s co-founder Liang Wenfeng was one of the attendees of the representative of the Chinese Premier, Li Qiang who held a closed-door symposium as per the information given by Xinhua news agency.
A case in point is that Liang’s invitation to such a high-level event may imply that DeepSeek’s accomplishments are in line with Beijing’s strategic goals for the purpose of the US export restriction as well as to achieve the self-sufficiency in the core sectors in the AI sector.
Of great importance, in the symposium Mr. Baidu also Robin Li last year, a message was there that Artificial Intelligence (AI) leaders are the ones that are most suitable to China’s future technology policies that they will shape. With the AI changes resulting from DeepSeek, China can thus play a major part in the race for AI global leadership.