DeepSeek raises new capital weeks after closing first funding round

Chinese AI lab DeepSeek is raising external capital again, just weeks after closing its first funding round at more than $7.4 billion. The company wants to acquire its own data centres and chips in order to continue its aggressive pricing strategy, an ambition that requires more money than the initial round generated.

DeepSeek was founded in July 2023 by Liang Wenfeng, who also runs Chinese hedge fund High-Flyer. Liang reportedly holds 84 percent of the company through two holding structures. Following the most recent funding round, his estimated net worth has risen to around $36 billion, making him the wealthiest AI model builder in the world.

The first external funding round, which closed in late May or early June 2026, raised more than 50 billion yuan, equivalent to just over $7 billion, at a valuation of approximately $52 billion. Founder Liang himself invested around $3 billion. Other participants included Tencent, battery giant CATL, JD.com, NetEase and state fund National AI Industry Investment Fund.

Low prices require own infrastructure

DeepSeek is known for its low API pricing, which sits well below that of Western competitors. That strategy attracts a large number of users and developers, but also places significant pressure on costs. To keep operations cheap as usage grows, the company no longer wants to depend on rented compute capacity from third parties.

Building its own data centres and acquiring its own chips costs considerably more than what the first round generated, which explains why DeepSeek is already approaching investors again so soon after closing that round. Specific figures for the new round have not yet been disclosed.

DeepSeek's models, including DeepSeek-R1, launched in January 2025, and the more recent V4-Pro and V4-Flash, are released as open-weight models under open-source licences such as the MIT licence. That makes them widely accessible, but also means DeepSeek must generate its revenue primarily through its own API service rather than through licence sales.

From 160 employees to doubling the team

DeepSeek currently has around 160 employees, a relatively small team for a lab that is attracting global attention. The company has announced plans to double its headcount. That growth aligns with its broader ambitions: own infrastructure, more models and a stronger position in the international AI landscape.

Its headquarters are in Hangzhou, the technology city in Zhejiang Province that is also home to companies such as Alibaba. The backing from parties including Tencent, JD.com and the state fund reflects how broad the interest in DeepSeek has become within the Chinese technology and investment community.

Relevance for the European and Dutch perspective

The speed with which DeepSeek is seeking new funding illustrates a pattern that European AI labs will also recognise: the costs of scaling AI infrastructure rise quickly, particularly when a company places low-barrier access at the centre of its strategy. Anyone offering cheap or free access to developers must be able to absorb the associated compute capacity themselves.

For Dutch and European founders, investors and policymakers who are investing in or thinking about AI infrastructure, the DeepSeek case shows that even a multi-billion-dollar round is not always sufficient to capitalise a fast-growing AI lab over the longer term. Those facing the same choices, between rented and owned infrastructure, between open and closed models, and between rapid scale and financial sustainability, have a concrete reference point in this example.