All posts
Trending Tech

DeepSeek raises $7.4B at $50B+ valuation with founder-control twist

Huma Shazia16 June 2026 at 1:27 pm5 min read
DeepSeek raises $7.4B at $50B+ valuation with founder-control twist

Key Takeaways

DeepSeek raises $7.4B at $50B+ valuation with founder-control twist
Source: Tech-Economic Times
  • DeepSeek raised $7.4 billion at a $50 billion+ valuation in its first external funding round
  • Investors must place capital in a limited partnership controlled by CEO Liang Wenfeng, not in DeepSeek itself
  • A five-year lock-up and zero voting rights apply to all investors except China's National AI Investment Fund

DeepSeek, the Chinese AI lab that rattled Silicon Valley earlier this year with its cost-efficient R1 and V3 models, has closed a $7.4 billion funding round at a valuation exceeding $50 billion. The structure is unlike anything seen in major tech fundraising: investors are locked in for five years, have no voting rights, and their money goes into a limited partnership managed by founder Liang Wenfeng rather than into DeepSeek itself.

The Information reported the details on Tuesday, citing people with direct knowledge of the deal. Reuters has not independently verified the figures. DeepSeek has not commented.

Why would investors accept these terms?

The answer lies in DeepSeek's sudden emergence as China's de facto national AI champion. After the V3 and R1 models drew praise in the U.S. for matching or exceeding GPT-4 class performance at a fraction of the training cost, the company became a symbol of what Chinese AI could achieve under hardware export restrictions. Investors appear willing to accept harsh terms for a seat at that table.

Liang himself is putting in 20 billion yuan (roughly $2.9 billion) of his own money, according to earlier Reuters reporting. That makes him the largest contributor by far. Tencent is reportedly considering 10 billion yuan, and battery giant CATL is weighing 5 billion yuan. If both commit, they would become the biggest external backers.

The only investor exempted from the lock-up and voting restrictions is China's National Artificial Intelligence Industry Investment Fund, a state-backed vehicle. It invested directly in DeepSeek and retains full rights. That carve-out signals where the real power in this deal sits.

The limited partnership model explained

In a typical venture round, investors buy equity in the company. They get shares, voting rights proportional to their stake, and the ability to sell when a liquidity event happens. DeepSeek's structure inverts most of that.

Instead of buying DeepSeek shares, investors place capital into a limited partnership that Liang controls as general partner. They become passive LPs with no say in operations, board seats, or strategic direction. The five-year lock-up means they cannot exit before 2030, regardless of what happens to the company or the broader market.

This is not unprecedented in private equity or hedge funds, but it is extremely rare for a tech startup raising at a $50 billion valuation. The structure effectively insulates Liang from investor pressure. He can pursue long-term research bets without quarterly check-ins or demands for near-term commercialization.

What this means for China's AI ecosystem

DeepSeek's funding is not just a company milestone. It is a policy statement. The structure ensures that no foreign capital can enter through conventional channels, and domestic investors who do participate are aligned with Beijing's industrial priorities, not their own exit timelines.

The structure is less of a traditional investment and more of an alignment of industrial policy, ensuring the lab remains entirely under the founder's strategic vision.

— Industry Analyst, AI Market Observer

The involvement of CATL is particularly telling. A battery company backing an AI lab suggests strategic supply chain thinking. CATL may be betting that DeepSeek's models will eventually power autonomous vehicles, robotics, or energy grid optimization, all areas where China is pushing for self-sufficiency.

How DeepSeek compares to OpenAI's funding

OpenAI has raised cumulatively larger sums, burning through $34 billion in 2025 alone and still posting cash losses ahead of its expected IPO. But OpenAI operates under a complex nonprofit-capped-profit structure that has drawn legal scrutiny and board drama. DeepSeek's model is simpler in one sense: Liang holds absolute control, full stop.

The difference in capital efficiency is starker. DeepSeek claimed its R1 model cost roughly $5.6 million to train. OpenAI's GPT-4 training cost estimates range from $60 million to over $100 million. Even if DeepSeek's figures are optimistic, the gap is wide enough to reframe the AI cost curve debate.

Also Read
OpenAI spent $34B in 2025, still burned cash before IPO

A direct comparison of how the leading U.S. lab's capital structure and burn rate differ from DeepSeek's approach

The risks no one is pricing in

A five-year lock-up assumes DeepSeek's trajectory only goes up. But AI model leadership is volatile. OpenAI, Google, Anthropic, and Meta are all shipping frontier models on 6-to-12 month cycles. A model that leads today can be commoditized in two years. Investors are betting that Liang's research instincts will stay ahead of that curve.

There is also geopolitical risk. If U.S.-China tech decoupling accelerates, DeepSeek may face restrictions on cloud partnerships, talent recruitment, or hardware access beyond current NVIDIA chip bans. The company's efficiency advantage could shrink if it cannot access next-generation accelerators.

Finally, the state carve-out is a double-edged sword. Beijing's AI fund retains full rights and direct equity. If state priorities shift, if DeepSeek's models are redirected toward military or surveillance applications, that could create reputational and regulatory friction internationally.

Also Read
Sarvam AI's $234M raise: big for India, small globally

For context on how AI funding rounds in other Asian markets compare

Frequently Asked Questions

How much did DeepSeek raise in its funding round?

DeepSeek raised over 50 billion yuan, approximately $7.4 billion, at a valuation exceeding $50 billion.

Why don't DeepSeek investors have voting rights?

Investors placed capital into a limited partnership controlled by CEO Liang Wenfeng, not into DeepSeek directly. This structure gives Liang sole decision-making authority.

How long are DeepSeek investors locked in?

All investors except China's National AI Investment Fund are subject to a five-year lock-up, meaning they cannot sell their stakes until 2030.

Who are the largest investors in DeepSeek?

Founder Liang Wenfeng is the largest at 20 billion yuan. Tencent is reportedly considering 10 billion yuan, and CATL is weighing 5 billion yuan.

Is DeepSeek backed by the Chinese government?

Yes. China's National Artificial Intelligence Industry Investment Fund invested directly and is the only investor with voting rights and no lock-up restrictions.

ℹ️

Logicity's Take

This round is less about raising capital and more about cementing control. Liang now has $7.4 billion in firepower with zero outside interference. The five-year lock-up is a bet that investors trust his judgment more than their own liquidity needs. If DeepSeek's cost-efficiency advantage holds, this structure could become a template for founder-led labs globally. If it does not, investors will have no exit and no recourse. The state carve-out is the quiet tell: Beijing is not just supporting DeepSeek, it is ring-fencing it.

ℹ️

Need Help Implementing This?

Logicity.in provides strategic analysis and implementation guidance for enterprises navigating AI adoption, vendor selection, and competitive intelligence. Contact us for briefings on how shifts in global AI funding affect your technology roadmap.

Source: Tech-Economic Times / ET

H

Huma Shazia

Senior AI & Tech Writer

Related Articles

Tesla's Remote Parking Feature: The Investigation That Didn't Quite Park Itself
Trending Tech·8 min

Tesla's Remote Parking Feature: The Investigation That Didn't Quite Park Itself

The US auto safety regulators have closed their investigation into Tesla's remote parking feature, but what does this mean for the future of autonomous driving? We dive into the details of the investigation and what it reveals about the technology. The National Highway Traffic Safety Administration found that crashes were rare and minor, but the investigation's closure doesn't necessarily mean the feature is completely safe.