Deepseek: Chinese AI Startup Deepcik has played the world of technology. American technical giants are in tremendous trouble and India is also in serious concern. The world has reversed in just 48 hours. The IT sector is showing shock in the stock markets and there has been a loss of $ 1 trillion in tech shares. China has shown by playing masterstroke that the US is no longer the only leader of AI, while India has been aware of the reality that it is not even involved in the race for AI tools.
All this was caused by a small startup research AI lab in China, which Silicon Valley and Washington D.C. Has shook it. A year ago, the “Deepcic”, launched by Chinese entrepreneur Liang Wenfeng, has suddenly become a strong competitor of OpenAI with chat GPT but has become a leading AI lab and AI is challenging American dominance in Super Power and Technology. The Dipcic R1 model features that it is competing with GPT 4 of Open AI, Lama of Meta and Gemini of Google while it has been developed at a very low cost.
The surprising thing is how it made such a big jump in AI despite the US banning the supply of high power AI chips to China.
Year -long startup
The new AI model has been developed by DipC and the startup came into existence a year ago. The company has said that it has spent only $ 5.6 million to power its base AI model, while US companies spend hundreds of million dollars on their AI techniques. This means that the Deepcic is capable of creating cheap models on AI chips with relatively low power.
What is Deepcik?
Chinese hedge fund manager Liang Wenfeng established a deep company at the end of 2023. Like other AI Startups in the world, Deepciak released several AI models last year, which caught some attention to the industry. Its R1 model suddenly surfaced at the end of last year. It was launched only last week for the public but it surprised when the company revealed a very low cost of operating this model. It is an open-source model, which means that other companies can test and manufacture the model to improve it.
The Deepcic app has gained a great lead on the App Store Chart and has been downloaded about 2 million times, leaving behind the chat GPT.
Who is Liang Wenfeng
– The first thing is that Liang Wenfeng is not an engineer but comes from the world of finance.- 40-year-old Venfeng co-folded Quantitative Hedge Fund Company “High Flire” in 2015 after graduation from Jhejiang University and AI Projects Research and development preferred in advancing.- In 2021, Venfeng began buying thousands of NVidia chips as part of a side AI project. Later, the Biden administration started controlling the supply of state -of -the -art AI chips to China. But no one would have wondered why Venfeng graphics processors were depositing. – Although Deepciak was initially a side project, Venfeng’s vision was to establish Dipcik as a domestic leader in AI that could compete with China’s largest tech companies as well as American tech veterans.- Interestingly It is that to form a team of AI researchers, Venfeng recruited top young talents from local Chinese universities and paid them equal to top Chinese technology companies such as Ticketkock Company BiteDance. Due to this strategy, Venfeng could attract the best talents from China. Even though the youth did not have years of experience of creating AI, they definitely had the technical skills required to develop AI from the beginning.
the road so far
– The first AI model of Dipcik was “Deepcic Kodar” which was released in November 2023 as an open-source model designed for coding.- After this, Deepcic LLM came with aimed to compete with other big language models. .- DeepC-V2 was released in May 2024. Due to its strong performance and low cost, there was an uproar in the Chinese AI market. Price war broke out and major Chinese technical giants like Bittens, Tensent, Bidu and Alibaba had to reduce their AI model prices.-Deepcick-V2 came after Deepcike-Koder-V2, which came a very advanced with 236 billion parameters Is model. It was designed for complex coding. It is a highly economical model.-The latest models of the company, Deepcich-V3 and Dipcic-R1 have established Dipcik as a leading AI research laboratory in China. The Dipcic-V3 model requires very little resources than the AI model of the US based technical giants.-The specialty of the DeepC-R1 released in January 2025 is that it has been focused on the Reasoning Task. It challenges the GPT-4 model of OpenAI with its advanced capabilities.
Very cheap and economical
Wenfeng’s one -year -old company said that its latest AI model, R1, spent only $ 5.6 million on computing power for its base model, while American companies spend hundreds of million or billions of dollars on their AI techniques. Not only this, the R1AA AI model has been developed by the company with only a team of 200 people.
Know that it is not only a lot of cost in developing the Artificial Intelligence Model, but it requires a lot of energy to operate it. This is the reason why major American tech companies are spending billions of dollars on developing AI models. The truth is that there is no guarantee that these tech companies will be able to make up for their investment in AI development.
Now, DipC has shown that it is possible to develop AI capabilities at a very low cost and less powerful chips like American tech companies. In the case of deepcick, the company trained its latest model on Nvidia H800 chips, which are much less powerful than NVDia’s Blackwell Chips. Since the techniques of deepcick require much less computing power for training, its cost is reduced.
Open source model
This model of DeepC is open-source. Which means that other companies, especially small developers, can manufacture over this model and make it better without paying the license fee.
Instead of developing their own models, companies can modify and execute the model of DeepCic at a very low cost. As a result, it can promote adoption of AI on a large scale. Using the open-source model, the deepcick strategy can provide access to AI devices for small businesses.
