Chinese AI lab DeepSeek is planning a fresh fundraising round at a valuation of roughly 500 billion yuan, or $74 billion, ahead of a potential initial public offering on Shanghai's STAR Market, two people with knowledge of the matter told Reuters. [1] The figure would mark another sharp jump for a company that raised $7.4 billion in June at a post-money valuation of about $52 billion, itself already a leap from where DeepSeek stood a year earlier. [1]
DeepSeek is looking to raise as much as 50 billion yuan in the new round, according to a third person briefed on the matter, and has started early deliberations on a STAR Market filing it hopes to complete this year. [1] The company's revenue has grown alongside its valuation — DeepSeek has reportedly reached roughly $500 million in annual recurring revenue — but the scale of the new raise reflects something more specific than growth: the rising cost of staying competitive at the frontier of Chinese AI. [1]
That cost pressure marks a reversal for a company whose entire public identity was built on the opposite claim. DeepSeek shook global markets last year by releasing models that appeared to rival leading U.S. systems at a fraction of the training and operating cost, and founder Liang Wenfeng had long resisted outside capital, bankrolling the company through his own quantitative hedge fund, High-Flyer. [1] Since June's round, DeepSeek has moved to double staff across departments including data centers and AI agents, and Reuters separately reported the company is developing its own AI inference chip, quietly hiring chip-design engineers for the effort. [1]
China's National AI Industry Investment Fund — the same state vehicle anchoring Moonshot AI's parallel fundraising push — has backed DeepSeek's financing alongside Tencent, CATL, NetEase, JD.com and several private investment firms. [1] That state participation underscores DeepSeek's strategic value to Beijing's push to build domestic AI champions independent of foreign technology, at the same moment DeepSeek faces intensified competition at home from ByteDance, Alibaba and well-funded rivals including Moonshot, Z.ai and MiniMax. [1]
Reuters' coverage treats the fundraising as a straightforward pre-IPO capital story: a fast-growing company raising money ahead of a listing, sourced carefully and hedged with the caveat that terms and timing may still change. [1] What that framing understates is the tension in DeepSeek's own narrative. The company that became a global phenomenon by proving frontier AI did not require Silicon Valley-scale capital is now raising back-to-back multibillion-dollar rounds to keep pace — a shift AI commentators argue is itself the more interesting story than the valuation number.
Whether DeepSeek's STAR Market ambitions survive contact with public markets, where investors will scrutinize a valuation built on state backing and reported revenue rather than audited results, remains the open question. For now, the $74 billion figure stands as both an investor vote of confidence and a quiet admission that the cheap-AI story DeepSeek told in 2025 no longer fully describes the company raising money in 2026.
-- DAVID CHEN, Beijing