DeepSeek is closing a new funding round that would value the Chinese AI lab at roughly $74 billion, the Wall Street Journal reported, capping a stretch of weeks in which the company's revenue has surged tenfold and its ambitions have expanded from open-weights disruption to a potential stock market listing as early as 2027.
The round, which multiple outlets report is nearing completion, would make DeepSeek one of the most valuable private AI companies in the world — and the most valuable to emerge from China's new wave of frontier labs. For more context on this story, see our ongoing artificial intelligence updates.
A $74 Billion Scorecard
The Wall Street Journal reported Thursday that DeepSeek is poised to reach a $74 billion valuation in its latest funding effort. PYMNTS reported the company is looking to raise around $7 billion, while MLQ.ai reported a target of roughly $7.4 billion with backing tied to High-Flyer, the quant fund founded by DeepSeek's creator.
The figures represent a striking recovery for a fundraising effort that appeared to stall earlier this summer, when DeepSeek suspended a previous round amid viral speculation and internal signals that founder Liang Wenfeng was in no hurry to take outside money at a discount. Since then, the company has released new models, ignited a price war on inference, and — crucially — begun to show the revenue to justify renewed investor interest.
Revenue Up 10x, Margins Up, Losses Too
The financial numbers leaking out of the round are unusual for a private AI lab, and they paint a picture of a business scaling faster than almost anyone expected.
The Information reported that DeepSeek's revenue reached approximately $70 million as of July — a tenfold jump compared with 2025. The Standard in Hong Kong reported the company took in 475 million yuan in its first seven months of the year, consistent with the $70 million figure. A report cited by Korean tech outlet Digital Today put DeepSeek's API gross margin at 82.9 percent, a level most AI startups can only dream of, and finance.biggo reported similar margin figures near 83 percent.
Growth has come at a cost. Finance outlet biggo reported that DeepSeek still posted a loss exceeding 700 million yuan, evidence that the company's aggressive model releases, compute buildout and low pricing continue to outpace even its surging revenue. DeepSeek did not respond to requests for comment on the figures, which were not publicly audited.
Eyes on a 2027 Listing
The most consequential detail may be the exit plan. The South China Morning Post reported, citing sources, that DeepSeek is targeting a stock market listing in 2027 as its pre-IPO funding nears completion.
CNBC reported that the fresh capital comes as Liang Wenfeng's quant trading empire navigates China's choppy IPO market, where regulators have tightened scrutiny of listings even as Beijing signals support for homegrown technology champions. A 2027 listing would place DeepSeek's debut after what is expected to be a crowded period of AI flotations — Anthropic has reportedly been laying groundwork for its own IPO, and OpenAI's valuation in private markets has reached extraordinary levels.
Even at $74 billion, DeepSeek would trail the towering private-market valuations of its American rivals, which press reports have placed in the hundreds of billions or higher. But the trajectory matters more than the level: no Chinese AI lab has ever closed the gap this quickly.
High-Flyer's Expanding Footprint
The round also shines light on the unusual structure behind DeepSeek. The lab was incubated by High-Flyer, Liang's quantitative hedge fund, which funded early GPU purchases years before DeepSeek became a household name.
Quartz reported this week that High-Flyer has been snapping up pre-IPO stakes in China's chip and robotics boom, positioning Liang's empire to profit from the broader industrial wave that AI is driving — not just from DeepSeek's own models. MLQ.ai reported that High-Flyer is backing China IPOs as part of the same strategy.
That dual identity — cutting-edge AI lab and quant fund portfolio company — has always made DeepSeek an outlier. It also explains the lab's famously patient approach to commercialization: DeepSeek answers to a founder with his own capital, not to a venture syndicate expecting a fast return.
Why It Matters
For the AI industry, the round is the clearest signal yet that China's frontier labs are attracting serious institutional capital despite export controls that restrict their access to the most advanced chips. DeepSeek built its reputation on efficiency — extracting frontier-adjacent performance from constrained hardware — and its revenue trajectory suggests that strategy is monetizing.
For US labs, the message is uncomfortable. The assumption that open-weight Chinese models would remain a cheap alternative to the frontier is being replaced by something more competitive: a funded, margin-generating, potentially public company with a tenfold revenue growth rate.
The round has not closed, and valuation figures in private markets are negotiable until the wire transfers clear. But if the Wall Street Journal's number holds, the question for 2027 will not be whether DeepSeek can go public. It will be how high the market lets it fly.
Follow AI Buzz Wire for continuing coverage of the global race in AI funding and frontier models.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
