China’s most famous AI company begins first round of financing. The para-state fund “Big Fund” is ready to invest, as the value of the company continues to grow.
The race for artificial intelligence is becoming increasingly tight between China and the United States. AI can without exaggeration be defined the modern equivalent of the “space race” of the 1960s.
It follows that the governments of the two superpowers do everything in their power to create the best possible conditions for their respective companies tech. It is with this interpretation that the exclusive news of the Financial Times.
DeepSeek’s leap in quality
According to the British newspaper, China’s main investment fund in the semiconductor industry, known as “Big Fund” and supported by the state, is in talks to lead the first round of financing Of DeepSeekthe company that owns the chatbot of the same name that revolutionized LLM model training last year.
The AI company, by far the most famous among the Chinese ones, was back in the spotlight at the end of April, when it previewed the new model V4.
A model that according to what was declared by the company it would be trained exclusively on memory chips produced by China’s Huaweicutting out the American for the first time Nvidia.
The objective of the financing would be to achieve a company valuation of as much as $45 billion. A goal which, however, could soon be revised upwards, just think that just a month ago the valuation was 10 billion, then increased to 20 and finally a 45.
For the company, ownership of the entrepreneur Liang WengfengIn short, it seems the time has come to make a leap in quality. The Chinese government, through the Big Fund, seems to agree: China must also have its own “chatGPT”.
Until now, the company had in fact only been financed thanks to money from High-Flyer Quantitative Investment, also owned by Wengfeng.
A business model that is no longer sustainable for what it is third most used chatbot in China (QuestMobile data), with more than 130 million monthly users, as well the most used Chinese AI software outside national borders.
The reasons for the change
AI is racing, even in China, and limited access to financing risks leaving DeepSeek behind, which has to deal with rivals monsters as ByteDance (owner of TikTok and the chatbot Dubai), the e-commerce giant Alibaba (owner of the chatbot Qwen) and countless other companies.
The competition is fierce. Between the end of last year and the beginning of 2026, well 5 top DeepSeek researchers have left the company for more remunerative positions in other national companies.
Added to this are the increasing development costs of the new V4 model. Costs that can no longer be supported with High-Flyer forces alone. As model iterations accelerate, DeepSeek faces a bottleneck in terms of computing power.
According to some rumors published in the Chinese media, the company is also using the latest chips Blackwell Of Nvidiaas well as those of Huawei, to train its new generation models.
In other words, DeepSeek is conforming to the typical model of large AI companieshow OpenAI And Claude: Exorbitant cost growth and the need for multibillion-dollar funding to continue cutting-edge research.
The differences between the Chinese and American approaches
DeepSeek is easily comparable to OpenAI and Claude: all three companies were created exclusively for AI, unlike Google, Microsoft, ByteDance or Alibaba, which developed chatbots as a complement to their existing ecosystems.
Yet DeepSeek’s approach is fundamentally different. American companies aim to create a “universal supervisor” of AI, a closed and proprietary omniscient assistant, fully usable only by subscription.
The Chinese company, however, embodies Beijing’s strategy of prioritizing “industrial diffusion”: its models are open source (i.e. it can be downloaded and modified by users) and designed to be integrated everywhere, from factories to logistics.
AND a more distributed, less flashy (and expensive), but perhaps more pervasive AI.
