DeepSeek’s founder has rejected proposals from investors who wanted to profit from the company. He said he wanted to maintain the company’s original ethos, which brought him global fame.
DeepSeek has been overwhelmed due to the large amount of traffic. Hence, the chatbot has frequent hiccups and service outages. At the same time, global authorities have been restricting the usage of DeepSeek, citing national security concerns. In particular, the US has considered banning DeepSeep. Other internet companies use its free code to drive their business operations.
Despite the chatbot’s success, Liang Wenfeng, the founder, told his associates that he was not in a hurry to secure investment. According to WSJ’s sources, he fears that investors would try to influence decision-making at the company. Another major concern for Liang Wengfeng is that of government-linked investments. He believes that global adoption would become difficult if there is a connection between DeepSeek’s AI models and government officials.
Currently, Liang Wenfeng is at a point that’s common among all tech entrepreneurs when their passion project goes viral. Early this year, DeepSeek disrupted the AI market with its low-cost and free-to-use open-source AI models. These models were said to rival products from mega firms like OpenAI’s ChatGPT and Anthropic’s Claude. What set it apart was the fact that it was trained without expensive AI chips.
The success of DeepSeek was no less than a patriotic win for China itself. The startup’s virality sparked patriotism among Chinese leaders, as Xi Jinping himself invited Liang for a talk. China’s state-owned bank also offered a low-interest loan to the startup, according to sources familiar with the talks. At the same time, other Chinese giants have also been in talks with Liang. These include executives from Alibaba and Tencent, two companies looking at a potential cooperation.
Sources say that Liang does not want to change the core AI models of DeepSeek, which are currently available free of cost. In the last week of February, DeepSeek’s parent company published techniques that it used for the training of its models using downgraded Nvidia chips. These chips were specifically designed for China.
Now, the company is gearing up for its next reasoning model, which aims to solve complex problems. Sources say this model could be released as early as April.
DeepSeek’s vision goes back to 2015 when Liang founded High-Flyer
Liang started a hedge fund, High-Flyer, in 1975. However, its name sounded more like a tech company in its advertisements, and the company even tried to include ‘technologies’ in its name. However, regulators did not agree to that. The firm used to look for people excelling at math at contests and promised up to $270,000 per year for AI engineers. Liang and his team used the hedge fund’s profits to support their AI goals. Between 2015 and 2020, the quants under Liang were outperforming the market. Eventually, assets under the management of High-Flyer reached $14 billion before the fund’s performance started to fade.
Eventually, the hedge fund closed, and it announced a strategic shift towards AI, resulting in the birth of DeepSeek. Today, DeepSeek’s AI models are mostly free. However, it does offer paid services, too. The biggest problem that DeepSeek suffers from is handling its huge traffic.
Its servers frequently crash, and according to some people, they can only interact with the chatbot a few times per day. Chinese giants like Tencent are testing the startup to power its features, such as a search engine in Tencent’s payment and messaging app, WeChat. However, they do not need to pay DeepSeek. Users can either opt for a DeepSeek-powered chatbot or use Tencent’s own network, which is much more stable.
According to familiar sources, DeepSeek pitched itself to various venture capital funds back in 2023, including certain foreign firms. However, these funds were not too keen on investing in the startup as they couldn’t see a path to recover the funds.
Yet, fast-forward to 2025, investors have been expressing their interest in investing in the form even though there’s no clear plan to make money via DeepSeek. Liang has refused these offers as of yet as he thinks about the company’s strategy. DeepSeek is currently looking for ways to help large tech companies make commercial AI applications and receive a share of the benefits.
In 2023, Liang said, “We don’t do applications, we just do research and exploration,” and it looks like he is still sticking to this vision in 2025.
Cryptopolitan Academy: Tired of market swings? Learn how DeFi can help you build steady passive income. Register Now
