DeepSeek IPO: Chinese AI Startup Engages CITIC Securities for Shanghai Listing at $75 Billion Valuation

DeepSeek Advances IPO Plans with CITIC Securities

Chinese AI startup DeepSeek has engaged CITIC Securities to prepare for an initial public offering on Shanghai's tech-focused STAR Market, according to two sources familiar with the matter1

1

. The Hangzhou-based company aims to begin the DeepSeek IPO process this year, marking a significant step in the startup's capital-raising strategy2

2

. DeepSeek's engagement of CITIC Securities for domestic IPO signals the company is advancing in its listing plans, as mainland China-bound companies typically appoint securities firms to conduct pre-listing tutoring before submitting applications1

1

. However, the timing of the offering, the amount DeepSeek seeks to raise, and its target valuation have not been finalized4

4

. Neither DeepSeek nor CITIC immediately responded to requests for comment1

1

.

Capital Requirements Drive IPO Push

The IPO discussions come as DeepSeek seeks fresh capital to fund computing infrastructure, AI model development, and talent retention amid intensifying competition among leading Chinese and U.S. AI firms2

2

. DeepSeek is currently pursuing a funding round that would value the company at 500 billion yuan, equivalent to approximately $75 billion1

1

. This represents a substantial increase from the $7.4 billion the startup raised in June at a post-money valuation of over $50 billion4

4

. In the June funding round, DeepSeek founder Liang Wenfeng personally committed 20 billion yuan, while Tencent and battery company CATL contributed 10 billion yuan and 5 billion yuan respectively, becoming the largest external shareholders1

1

. Founder Liang hopes the IPO proceeds will help DeepSeek offer stronger incentives to retain key staff and researchers, as the company has recently lost talent to better-funded rivals including ByteDance and Xiaomi4

4

.

DeepSeek's Technology and Market Position

DeepSeek rose to prominence after releasing powerful open-source models that appeared capable of competing with leading U.S. systems while using comparatively fewer computing resources3

3

. Founded in 2023 by entrepreneur and quantitative investor Liang Wenfeng, the company develops large language models capable of generating text, writing computer code, solving mathematical problems, and performing advanced reasoning tasks3

3

. The company attracted international attention in early 2025 following the launch of its DeepSeek-R1 reasoning model, which DeepSeek claimed could deliver performance comparable with OpenAI's o1 model on several reasoning benchmarks3

3

. The company released the model under an MIT license, allowing developers to use and modify its model weights relatively freely3

3

.

Global AI Firms Rush to Public Markets

The DeepSeek IPO push comes as rivals at home and abroad rush to public markets for funding amid rising costs of competing in AI, which requires large amounts of computing power, data-center capacity, and engineering talent2

2

. Chinese large language model developers Z.AI and MiniMax both went public in Hong Kong earlier this year, while Beijing-based startup Moonshot has filed confidentially for a Hong Kong IPO1

1

. Moonshot was valued at $50 billion in a funding round, below DeepSeek and Z.AI, which has a market capitalization of about $54 billion4

4

. However, these valuations remain far below those sought by U.S. rivals in the intense competition with U.S. and Chinese AI firms. Anthropic could be valued at up to $2 trillion in its IPO, some investors have said, while OpenAI could be valued at up to $1 trillion in a planned offering2

2

. The valuation gap reflects a sharp revenue disparity, underscoring Chinese AI developers' challenge in converting increasingly competitive technology into profit1

1

. Anthropic is planning to launch its offering in mid-October and complete the listing before the midterm elections in November4

4

. The surge in AI IPO activity highlights the enormous capital requirements for developing frontier AI models, including expensive semiconductor chips, data-center infrastructure, electricity, and highly paid technical employees3

3

.