Chinese AI Lab DeepSeek Secures Record $7.4 Billion Mega-Round

DeepSeek, the Hangzhou-based artificial intelligence laboratory that disrupted the global tech sector with its highly efficient open-source models, has officially closed its first external funding round. The company secured more than 50 billion yuan—approximately $7.4 billion—catapulting its valuation past the $50 billion mark and cementing its status as China's most valuable AI startup.[1][3]

The landmark financing deal, which was recently finalized in corporate registry filings, marks a significant shift for a company that had previously relied entirely on internal capital. Founded in 2023 by the quantitative hedge fund High-Flyer Quant, DeepSeek built its reputation as a pure research lab before releasing models that rivaled the performance of industry leaders like OpenAI and Meta.[1][4]

The structure of the mega-round is highly unusual for the venture capital industry, reflecting founder and CEO Liang Wenfeng's determination to maintain absolute control over the company's direction. According to financial disclosures, the vast majority of investors were required to channel their capital into a limited partnership managed directly by Liang, rather than purchasing equity in DeepSeek directly.[1][3]

Under these strict terms, participating investors—which reportedly include industrial heavyweights like Tencent, JD.com, and battery manufacturer CATL—receive no voting rights. Furthermore, they are subject to a stringent five-year lock-up period during which they cannot sell their shares, a mandate that underscores DeepSeek's focus on long-term research over short-term liquidity.[1][3]

There is, however, one notable exception to this restrictive framework. China's state-backed National Artificial Intelligence Industry Investment Fund was permitted to invest directly into the company. This direct investment grants the state fund voting rights and exempts it from the five-year lock-up, highlighting the strategic national importance Beijing places on DeepSeek's foundational models.[1][3]

DeepSeek's rise to global prominence accelerated in early 2025 and continued into 2026 with the release of its V4 model, currently the largest open-weights AI model available. The company achieved this by utilizing highly optimized training techniques that require significantly fewer high-end semiconductors, proving that top-tier AI performance can be achieved without relying entirely on massive clusters of the newest Nvidia chips.[3][4]

This cost-efficiency has allowed DeepSeek to aggressively undercut its American competitors on pricing. The company recently made a 75 percent discount on its V4 Pro model permanent, rendering it roughly 11 times cheaper for input processing and 35 times cheaper for output generation compared to OpenAI's flagship models.[3]

Despite the successful close of the $7.4 billion round, DeepSeek's immediate financial roadmap has hit a temporary speed bump. In late July, the company paused discussions for a second, $1.5 billion follow-on funding round that would have valued the firm at over $70 billion.[2][5]

The suspension reportedly stems from the unauthorized leak of internal comments made by Liang during the initial fundraising process. Unverified transcripts circulating on social media allegedly captured the CEO expressing frustration over the persistent gap in AI sophistication between the U.S. and China, as well as the industry's ongoing reliance on Nvidia hardware.[2][6]

Rather than proceed under the shadow of the leaked transcripts, DeepSeek opted to verbally inform prospective backers that it would delay signing new investment agreements. The company is expected to revamp its fundraising strategy and could resume the follow-on round later this year once the public scrutiny subsides.[5][6]

Looking ahead, the $7.4 billion capital injection will be deployed to aggressively expand DeepSeek's data center footprint and computing infrastructure. As the company scales its hardware capabilities to train the next generation of foundational models, it is also laying the groundwork for a highly anticipated initial public offering.[4][5]

Financial advisors are reportedly already in discussions with the firm regarding a domestic IPO in mainland China. With preparations underway, DeepSeek could file its prospectus as early as late 2026, setting the stage for what would likely be one of the most significant technology listings of the decade.[4][6]