DeepSeek Valued at $52bn After First External Funding Round

DeepSeek has raised about 50 billion yuan, or roughly $7.4bn (£5.8bn), in its first external funding round, giving the Chinese AI startup a post-money valuation of more than 350 billion yuan — approximately $52bn (£41bn) — according to a regulatory filing.

The previously unannounced deal was closed on 16 June, with the transaction finalised by 17 June. Anhui Korrun, a Chinese luggage manufacturer, received confirmation of its completed investment on 15 July and disclosed details through a stock exchange filing. The filing offers the first public evidence of DeepSeek’s valuation and investor group since the company began taking outside capital.

Anhui Korrun invested indirectly through a fund managed by Monolith Management. Its 2.9 billion yuan investment, for an indirect stake of 0.8265%, implies a company valuation of about 350.88 billion yuan, or $51.82bn.

Big Chinese names join DeepSeek’s first outside round

DeepSeek had funded itself through its founder Liang Wenfeng and High-Flyer, the quantitative hedge fund behind the AI lab, since it was founded in July 2023. This is the company’s first known external fundraising round.

The investor group includes Tencent, which contributed 10 billion yuan, and battery maker CATL, which invested 5 billion yuan. JD.com, NetEase and venture firm IDG Capital each contributed 3 billion yuan, according to the available details. Other backers include China’s national AI fund, Loyal Valley Capital, Monolith Management and Shixiang Capital. The planned group is expected to contain fewer than 10 investors.

Liang himself contributed 20 billion yuan. That substantial founder commitment is part of a deal structure designed to maintain his control over the company despite the scale of the outside investment.

Most investors were required to put capital into a limited partnership managed by Liang rather than invest directly in DeepSeek. They are generally locked in for five years and receive no voting rights, although they gain access to financial information and priority rights in later financing rounds.

China’s National Artificial Intelligence Industry Investment Fund is the key exception. It invested 1 billion yuan directly in DeepSeek, has voting rights and is not bound by the five-year lock-up. The arrangement makes the state-backed fund’s role distinct from that of the other financial backers.

Money is earmarked for infrastructure and staff

DeepSeek plans to use the capital partly for employee equity incentives, staff retention, technical breakthroughs and expansion. Its ambitions also require expensive physical infrastructure: the company is pursuing gigawatt-scale proprietary data centres, in-house AI inference chips and AI agent products.

That helps explain why a company which had previously avoided outside investors has moved to raise such a large sum. DeepSeek built its reputation on delivering strong AI results at comparatively low reported training costs, but operating and expanding frontier-scale AI services demands far more than a single model-training run.

The company became internationally prominent after publishing its V3 model in December 2024 and releasing its R1 reasoning model in January 2025. V3 had 671 billion parameters, while R1 was presented as an open-weights reasoning model with performance comparable to leading proprietary systems on several benchmarks.

DeepSeek estimated that V3 cost about $6m to train, far below the hundreds of millions or billions of dollars associated with some Western frontier-model efforts. It achieved its results while facing US export restrictions that limit Chinese companies’ access to the most advanced chips.

The R1 release also triggered a striking market reaction on 27 January 2025, when the S&P 500 shed more than $750bn and Nvidia alone lost $590bn in market value.

A larger round and an IPO may follow

The Financial Times reported on 14 July that DeepSeek had wrapped its first funding round close to the end of May, raising $7bn at a $52bn valuation. It also reported that the company had begun preliminary discussions with new investors about another round that could value DeepSeek at about $71bn before the investment.

Other reports say DeepSeek is seeking a 500 billion yuan, or $74bn, valuation in a further fundraising round as it prepares for a mainland Chinese stock market debut. The details of a new round have not been finalised.

DeepSeek has also begun early work towards a possible Shanghai STAR Market listing, with a potential filing as early as late 2026 and a target listing in 2027. The company is reportedly speaking with accounting firms and investment-bank advisers as it works to complete financial statements by the end of December.

Why UK users should pay attention

For ordinary users, this deal does not appear likely to change DeepSeek’s chatbot overnight. Its V3 and R1 models remain open-weight and free to use, a combination that has attracted millions of users globally and made the company a meaningful alternative for developers and businesses seeking lower AI costs.

DeepSeek has already sharpened that price competition: in April it launched V4-Pro with a 75% price reduction. That matters to UK developers paying for AI APIs, because lower-priced competition can force larger Western providers to respond.

However, the funding structure also brings governance into clearer focus. The state-backed AI fund’s direct voting rights and special exemptions will raise fresh questions for users deciding what information to put into an AI chatbot. DeepSeek’s funding is a major vote of confidence in its technology, but it also makes the company’s strategic importance to China more explicit.