The domestic large model sector is experiencing an unprecedented capital storm. According to the latest news, the AI company DeepSeek, known as the "price killer," is launching its first major round of fundraising, with a target amount of 50 billion yuan. If this plan is ultimately realized, it will set a new record for the highest single-round funding in Chinese large model companies.
The most remarkable detail of this financing is that the lead investor is not a traditional venture capital firm or an internet giant, but rather the founder of DeepSeek, Liang Wenfeng. It is reported that Liang Wenfeng's personal investment can reach up to 20 billion yuan, accounting for 40% of the total funding. This move not only demonstrates the founder's extreme confidence in his own technology, but also caused DeepSeek's valuation to surge four times within three weeks, rising from 10 billion USD at the beginning of April to the current 50 billion USD (about 350 billion yuan).
DeepSeek has long been labeled as "no funding, no commercialization, no roadshow" idealism, relying on the background of quantitative fund incubation to focus on technical development. However, as the AI competition enters the deep water zone in 2026, this pure laboratory model has begun to actively transform into a heavy asset and high turnover company form. The exponential growth of computing power demand, the competition for top talent salaries, and the transformation of leading models into stable enterprise products are all practical logic behind this huge funding.
At the same time as the financing accelerates, DeepSeek's technology iteration has also entered a "fast mode." The newly launched V4 series model has achieved support for 1M ultra-long context, significantly enhancing the ability to handle complex tasks and long code repositories. The highly anticipated V4.1 version is also scheduled for June, when it will provide more comprehensive industry protocol support and have the ability to process images and audio simultaneously in a multimodal way.
For DeepSeek, this large amount of capital injection means that its release schedule will be closer to the industry's regular pace, and the competition dimension will evolve from a single model index competition into a comprehensive game covering computing power reserves, commercial delivery, and talent incentives. From a low-key idealistic laboratory to a domestic large model giant valued at 350 billion yuan, DeepSeek is learning to take a more mature commercialized posture, continuing to take the initiative in this global top-tier AI competition.
