Reporter: Jia Nan Song
The latest scoop on DeepSeek’s first external funding round keeps evolving.
On June 3, Cailian Press tapped into market chatter revealing that the heavily watched Chinese AI powerhouse, DeepSeek, is looking to raise roughly $7 billion in its debut external funding round. That could push its valuation up to a jaw-dropping $59 billion. Tencent and CATL are stepping up as the biggest outside backers, while NetEase and JD.com are also reportedly planning to join the round.
We reached out to Tencent, JD.com, and other key players for comment, but as of now, they’re staying quiet.
What’s really catching people’s attention is how fast things are moving. Back on May 6, reports surfaced that China’s National Integrated Circuit Industry Investment Fund was already in talks to lead this initial external round, with post-money valuation hovering around $45 billion. That means in less than a month, DeepSeek’s price tag has jumped by a cool $14 billion.
That said, Shanghai Securities News quoted insiders confirming the talks are real, but emphasized that the final valuation is still up in the air. Alongside the state-backed fund, several internet giants and other state-owned capital groups are also in the mix. Everyone’s playing it cool right now, and the final investor lineup and deal terms are still being hashed out.
Here’s what makes DeepSeek stand out: unlike most AI startups that hustle for venture capital from day one to keep their models running, DeepSeek took a completely different path. Since spinning out of quant trading giant High-Flyer Quant in July 2023, it stubbornly avoided outside equity funding for nearly three years. R&D, compute costs, and talent paychecks were entirely bankrolled by High-Flyer’s own operating profits. Founder Liang Wenfeng used his controlling stake in High-Flyer to build a development environment completely shielded from investor pressure. That’s exactly why it’s brushed off countless investment pitches from internet giants and industrial funds over the years.
Public records show Liang founded High-Flyer Quant back in 2015. At its peak, it managed over 70 billion RMB in assets, and by 2025, its flagship products delivered a solid 56.55% average return, ranking it among the top-tier quant funds in China. That steady cash flow has been quietly funding DeepSeek’s supercomputing clusters and large-scale model training. Everything from the early Yinghuo custom hardware to the end-to-end development of the V3 model came straight from the group’s internal R&D budget—a textbook case of organic, self-sustaining growth.
This isn’t the first time funding rumors have swirled. Back in February last year, whispers suggested DeepSeek was finally open to outside capital, with insiders quickly dismissing it as pure speculation. At the time, rumors even claimed Alibaba and state funds were circling, with Alibaba allegedly eyeing a 10% stake at a $10 billion valuation. Alibaba VP Yan Qiao swiftly shot that down.
Fast forward to April 22 this year, and the chatter resumed: Tencent and Alibaba were reportedly in talks to invest, with DeepSeek allegedly shopping for funding at a valuation north of $20 billion.
By May, the numbers got a serious upgrade. Reports pointed to a target raise of around 50 billion RMB (roughly $7.35 billion), pushing the post-money valuation past $51.5 billion. Liang Wenfeng reportedly planned to personally pitch in 20 billion RMB to secure the lion’s share of this round, ensuring he keeps tight control once the dust settles.
Then came the widely circulated version you’re seeing now: a 70 billion RMB (~$7B) raise at roughly a $45 billion valuation. Insiders claim negotiations are in the final stretch, with Tencent, IDG Capital, and Lisi Capital close to locking in their commitments. But as always in these deals, the exact amounts and final investor list could still shift.
On top of the financials, insiders note that Liang has made it crystal clear in investor meetings: DeepSeek will keep pushing open-source AI models while chasing the bigger prize—Artificial General Intelligence. His main focus? Tech breakthroughs, not quick monetization.
If this $7 billion round actually closes as planned, it’ll likely set a new benchmark for first-time funding among Chinese tech startups. Analysts are pointing out that DeepSeek’s valuation leap from the low $10B range to $59 billion isn’t just hype—it’s the market pricing in the real-world value of China’s homegrown large models. Their relentless open-source strategy is drastically cutting AI adoption costs for smaller businesses, which in turn is forcing industrial capital to scramble into foundational model development.
Just a few weeks ago on April 24, DeepSeek officially dropped its V4 model series and made it fully open-source. Built on a 1.6-trillion-parameter Mixture-of-Experts architecture, it natively handles million-token long-context inputs. In benchmarks, it’s going head-to-head with top-tier closed-source models overseas.
Right after the launch, they turned up the heat on May 22 with a permanent API price slash. V4-Pro rates were permanently cut to a quarter of the original price. Cache-hit input costs dropped from 1 RMB per million tokens to just 0.025 RMB, while cache-miss inputs and outputs fell from 12/24 RMB down to 3/6 RMB respectively.
The price war didn’t stop there. On June 2, Tencent Cloud quickly followed suit, slashing its DeepSeek-V4 API pricing across the board. Cache-hit categories saw cuts up to 97.5%, effectively matching their cloud rates with DeepSeek’s official self-operated API.
But let’s be real—the Chinese AI space is brutally competitive. ByteDance, Baidu, Alibaba, and other tech behemoths are throwing serious money, compute power, and ecosystem leverage into their in-house models. Meanwhile, top players like Zhipu AI and MiniMax aren’t slowing down either. To stay ahead, DeepSeek needs this fresh capital to scale up compute resources, keep iterating its models, and offer competitive packages to hold onto elite AI talent.
