China's artificial intelligence startup DeepSeek has entered the final stage of its first external fundraising round since its founding. The round is expected to exceed 50 billion yuan, and the company has reportedly been valued at close to $60 billion.

Hong Kong's South China Morning Post reported on June 3 that DeepSeek's valuation in the fundraising process was set at just under $60 billion. That is nearly a six-fold jump from the roughly $10 billion valuation discussed in the market in April.

DeepSeek has operated without outside funding, backed by its parent company High-Flyer Quant. But analysis suggests that as AI competition has intensified sharply, the need has grown to fix an official valuation and secure capital.

Major Chinese big tech firms and institutional investors are said to have participated in the investment. Tencent is expected to invest about 10 billion yuan, while NetEase and JD.com are each expected to invest about 3 billion yuan, the report said. CATL, the world's largest electric vehicle battery maker, is also being discussed as investing about 5 billion yuan.

In the venture capital sector, IDG Capital, Monolith, Royal Valley Capital and Ssang Tech are said to be included among potential investors. Founder and CEO Liang Wenfeng (량원펑) is expected to invest about 20 billion yuan of his own money, the report said.

The deal has not been finally confirmed, but signs have also emerged that the investment structure has already been fleshed out to a considerable extent. Qichacha, a Chinese corporate registration information service, shows multiple special purpose vehicles linked to key investors were established one after another from mid-May. On May 19, in particular, Hangzhou Chengche Business Consulting was set up at the same address as DeepSeek's registered address, and Liang was confirmed as registered as the de facto controlling person.

The industry points to the more intense AI competitive landscape as the backdrop to DeepSeek's fundraising. DeepSeek has focused on developing artificial general intelligence without taking outside capital, but it has reportedly needed to set its valuation to respond to rivals' aggressive hiring and to clarify the basis for valuing employee equity stakes.

Its technological competitiveness is also drawing investor interest. DeepSeek was assessed as having demonstrated, through its open-source AI models V3 and R1 released last year, that it can develop high-performance AI models at lower cost than U.S. rivals such as OpenAI and Anthropic.

DeepSeek also claims its recently released V4 model is comparable to the latest undisclosed models from OpenAI and Anthropic on key performance indicators.

It is also pursuing aggressive price competition. DeepSeek permanently cut the price of its V4 Pro model by 75 percent last month. It lowered costs to $0.0036 per 1 million cached input tokens and $0.87 per 1 million output tokens, moving to expand market share.

The industry sees further reductions in DeepSeek's operating costs if full-scale supply of Huawei's next-generation AI chip platform, the Ascend 950PR Supernode, begins. Huawei recently said it fully applied DeepSeek's V4 model to the Ascend 950PR platform, and Chinese AI semiconductor companies Cambricon and MetaX are also participating in related optimisation work.

But service stability remains a task. DeepSeek has experienced intermittent service disruptions since releasing V4, and infrastructure burdens are also increasing with user growth.

The industry expects that if the large fundraising is completed, DeepSeek will invest heavily in not only AI model development but also data centre expansion, service stabilisation and securing key talent, further expanding its influence in China's AI market.