DeepSeek Raises $7.4B in First-Ever Funding Round

DeepSeek, the Chinese lab that shook global markets in early 2025 with a low-cost open-weight model, has finally taken outside money. The company closed its first-ever external funding round on June 16, 2026, raising more than $7.4 billion at a post-money valuation reported between $52 billion and $59 billion. That makes it China's highest-valued AI startup, and the deal arrives with terms unlike almost any other venture round in the industry.

Key takeaways

• DeepSeek raised over $7.4 billion (about 51 billion yuan) in its maiden funding round, closing June 16, 2026.
• The post-money valuation lands between roughly $52 billion and $59 billion, crowning DeepSeek China's most valuable AI startup.
• Tencent led with around 10 billion yuan; battery giant CATL invested about 5 billion yuan, with JD.com and NetEase adding roughly 3 billion yuan each.
• Most commercial investors received no voting rights and accepted a five-year lock-up, with funds routed through a limited partnership controlled by founder Liang Wenfeng.
• China's state-linked National AI Industry Investment Fund secured direct ownership, voting rights, and an exemption from the lock-up.

What happened

For most of its life, DeepSeek operated as a side project of the quant hedge fund High-Flyer, funded internally and famously frugal with compute. That changed this month. After weeks of reported talks, the lab confirmed a Series A worth more than $7.4 billion, a figure large enough to fund the kind of compute clusters that frontier training now demands.

The backer list reads like a who's who of Chinese tech. Tencent committed the largest single check at roughly 10 billion yuan, gaining a partner with deep WeChat distribution and cloud capacity. JD.com and NetEase each put in about 3 billion yuan. The most unusual name is CATL, the world's biggest battery maker, which invested around 5 billion yuan in what observers called an energy-plus-AI crossover, with CATL positioned to supply power and storage solutions for DeepSeek's data centers.

Here is the round broken out by investor and what each one brings beyond cash:

Investor	Approx. commitment	Strategic angle
Tencent	~10 billion yuan	WeChat distribution, cloud capacity
CATL	~5 billion yuan	Power and storage for data centers
JD.com	~3 billion yuan	E-commerce and logistics data
NetEase	~3 billion yuan	Gaming and consumer reach
National AI Industry Investment Fund	State stake	Direct ownership, voting rights, no lock-up

Why it matters

The structure is the real story. Outside commercial investors were required to place their money in a limited partnership managed by founder and CEO Liang Wenfeng, rather than buying equity directly in the operating company. Those investors hold no voting rights and face a five-year lock-up. The one exception is the state-linked National AI Industry Investment Fund, which received direct corporate ownership, voting power, and freedom from the lock-up.

In plain terms, capital flowed in but control did not. Liang retains tight command, and the Chinese state holds the only meaningful governance stake among new entrants. For a company whose efficient training methods became a reference point in debates over AI compute and efficiency, the financing says as much about Beijing's industrial priorities as it does about DeepSeek's roadmap.

The lock-up and lack of voting rights are not generosity from investors; they are the price of access. DeepSeek does not need the governance interference that comes with conventional venture equity, and Chinese tech giants want exposure to the country's standout AI lab even on lopsided terms. The structure also insulates the company from the kind of investor pressure that pushes US labs toward closed, monetizable models, which helps explain why DeepSeek can keep shipping open weights. It is a financing built to preserve a strategy, not to maximize returns for the people writing the checks.

For context, here is where the raise puts DeepSeek among well-known AI labs by reported valuation:

Lab	Reported valuation	Model strategy
DeepSeek	$52-59 billion	Open-weight
Anthropic	$100 billion-plus	Closed frontier
OpenAI	$300 billion-plus	Closed frontier
Mistral	Roughly $14 billion	Open and closed mix

DeepSeek is not the most valuable lab on the list, but it is the only one in the top tier built almost entirely around releasing model weights publicly, and the only one doing it under chip-export constraints.

The raise also lands amid a global memory and compute crunch that has reshaped budgets across the sector, the same pressure driving record results at suppliers covered in our report on the AI memory boom. DeepSeek's pitch to investors leaned on this: it intends to fund extensive compute clusters, accelerate pre-training of next-generation open-weight models, and scale domain-specific applications for enterprises.

What is next

The company has signaled three priorities for the new capital:

1. Compute scale-up. Build out large clusters to train successor models, the persistent bottleneck for any frontier lab operating under export limits.
2. Open-weight strategy. Continue releasing model weights publicly, a posture that has won developer goodwill and pressured closed competitors on price.
3. Enterprise applications. Push domain-specific tools into Tencent, JD, and NetEase ecosystems with access to billion-user platforms.

Watch for whether DeepSeek can convert capital into a model that closes the gap with US frontier systems, and whether the unusual governance terms become a template for other Chinese AI raises.

Frequently asked questions

How much did DeepSeek raise and at what valuation?

The company raised more than $7.4 billion (roughly 51 billion yuan) in a round that closed June 16, 2026, at a post-money valuation reported between $52 billion and $59 billion, its first-ever external funding.

Who invested in DeepSeek?

Tencent led with about 10 billion yuan. Battery maker CATL added roughly 5 billion yuan, while JD.com and NetEase each contributed about 3 billion yuan. The state-linked National AI Industry Investment Fund also participated.

Why are the funding terms considered unusual?

Most commercial investors put money into a limited partnership controlled by founder Liang Wenfeng instead of buying equity directly. They received no voting rights and accepted a five-year lock-up. Only the state fund got direct ownership and voting power.

What will DeepSeek do with the money?

The company says it will fund large compute clusters, accelerate training of next-generation open-weight models, and scale enterprise applications across its investors' platforms.

DeepSeek spent two years proving it could compete on a shoestring. With $7.4 billion in the bank, it no longer has to.