Skip to content
World clockEU--:--UK--:--USA--:--CN--:--PLDEFRIT中文EN

portal about AI and technologyevents · analysis · interviews · technical background

Search
LIVE
›

DeepSeek accelerates its capital plans: IPO this year, second round targets 50 billion yuan

Reuters and The Information report that DeepSeek has hired CITIC Securities to prepare a STAR Market listing and is pushing a second funding round targeting 50 billion yuan. Annualized revenue is said to have passed 1 billion US dollars.

AI & modelsNewsGrace OkonkwoPublished: 24 September 20266 min readSources 3
DeepSeek accelerates its capital plans: IPO this year, second round targets 50 billion yuan

DeepSeek is moving faster on outside capital. Reuters reported on September 9 that the company has hired CITIC Securities to prepare a STAR Market listing and plans to start the IPO process within the year. How much it will raise and at what valuation are not yet set. DeepSeek wants the proceeds to expand computing infrastructure, spend more on model research and in-house chips, and strengthen incentives for core staff.

On funding, the report says DeepSeek is working on a new round at a target valuation of about 500 billion yuan. In June the company closed its first outside round at a post-money valuation above 50 billion US dollars, raising about 7.4 billion US dollars. Founder Liang Wenfeng put in about 20 billion yuan himself, the largest single contribution. Tencent contributed about 10 billion yuan and a CATL-affiliated company about 5 billion yuan. JD.com, NetEase, IDG Capital and the National Artificial Intelligence Industry Investment Fund also took part.

On September 24, The Information reported, citing two people familiar with the matter, that DeepSeek's annualized revenue has passed 1 billion US dollars and that a second round of 7.5 billion US dollars is set. A few months earlier annualized revenue was still under 500 million US dollars. The growth comes partly from a recent increase in API pricing and partly from continued market demand for large models. The company plans to close the round by the end of October.

Liang told investors that DeepSeek keeps most of its resources on new model research, the people said. More than 70% of computing power goes to model training, leaving under 30% for inference on existing models. He also said the price change did not cost the company customers and that demand remains strong.

The scramble for a share of the round has brought abuses. Caijing magazine reported that some middlemen are asking for 18% upfront and 35% of later gains. An investor putting in 100 million yuan would pay 18 million yuan upfront and hand over 35% of investment returns later. By comparison, a fund manager normally charges an annual management fee of about 2% of fund size.

DeepSeek vets who can join the round, the report adds. Some individuals face a minimum ticket of 100 million yuan, and the company requires look-through checks on the limited partners behind investing funds. The investment carries a five-year lock-up with no transfers. Apart from the National Artificial Intelligence Industry Investment Fund, most outside investors get no voting rights and no board seat.

Comments 0

Sources

3
  1. 01曝 DeepSeek 已聘请中信证券筹备 IPO 事宜ZH
  2. 02曝 DeepSeek 确定 75 亿美元第二轮融资,年化营收突破 10 亿美元ZH
  3. 03DeepSeek 投资份额有多抢手?1 亿元入场先收 1800 万前端费用ZH

All figures and quotations in this text come from the sources listed below.

Content prepared by the editorial team with AI assistance.

Grace Okonkwo

Grace Okonkwo

AI, models and technology

Grace Okonkwo covers AI, models and technology for FLASH24, working from primary sources such as model cards, API documentation and benchmark papers rather than vendor summaries. She checks training data provenance, evaluation conditions and reported scores against the underlying datasets before any figure reaches print. She interviews researchers and engineers directly, tracks release calendars from major labs, and compares successive model versions on the same tests. Her own self-hosting, home-network and documentation-reading habits feed straight into that desk, since she tests tools on her own hardware first. She does not publish benchmark claims without a reproducible method.

Newsroom →

Comments

0
  1. No comments yet — be the first.

Write a comment

Comments are public. We do not publish abuse, spam or advertising.