For roughly two years, DeepSeek was the most-discussed AI company in the world that didn't have outside investors. Its V3 and R1 models rattled Silicon Valley. Its founder, Liang Wenfeng, ran the operation through his quantitative hedge fund, High-Flyer, without a venture cap table, without a roadshow, without a press-release funding round. That is about to change.
DeepSeek is preparing to raise about $7.4 billion in its first-ever funding round, a deal that could value the startup at up to $59 billion and rank among the largest private technology financings in China. The round is expected to close within the next couple of weeks, though financial terms could still change. Fewer than ten investors are expected to take part.
Liang is expected to contribute 20 billion yuan of his own money, accounting for roughly 40% of the total fundraising. Among outside investors, Tencent is reportedly considering a 10 billion yuan commitment, with battery giant CATL exploring a 5 billion yuan investment. If finalized, the two companies would become DeepSeek's largest external backers.
The startup is said to be in late-stage discussions with several additional investors, including China's national artificial intelligence fund, gaming company NetEase, e-commerce giant JD.com, and Hong Kong-based IDG Capital.
An Investor List That Includes a Battery Company
The composition of the round is unusual by any standard. There is no Sequoia tranche, no SoftBank vision fund, no sovereign wealth vehicle from the Gulf. What there is: the world's largest electric vehicle battery manufacturer and a gaming company. CATL's potential role points to a specific part of the AI boom — energy. The company is best known as the world's largest maker of electric vehicle batteries, but it has been moving into AI data centers by exploring power equipment and energy storage solutions. That push comes as AI workloads increase demand for large-scale, reliable power.
Tencent's involvement carries its own significance. The company has invested heavily in its Hunyuan AI model but remains behind domestic leaders such as ByteDance's Doubao and DeepSeek. A closer relationship with DeepSeek could help Tencent strengthen its position as competition intensifies across China's AI sector.
The investor logic, taken together, describes something closer to a state-adjacent industrial consortium than a typical venture deal. Tencent gets distribution leverage. CATL gets a seat at the table where energy infrastructure meets AI compute. China's national AI fund gets what national AI funds are for.
What $59 Billion Looks Like Next to $965 Billion
DeepSeek, which became globally known after the success of its V3 and R1 models, has until now largely avoided outside capital and relied on support from founder Liang Wenfeng and his quantitative hedge fund, High-Flyer. The company's entire identity was built around doing more with less. The company gained global attention after claiming that one of its flagship models was trained for about $6 million, a small fraction of the cost typically associated with leading AI systems. Its V3 and R1 models later impressed researchers and experts by delivering strong reasoning capabilities while using significantly less computing power than many expected.
Against that backdrop, raising $7.4 billion reads simultaneously as a capitulation to scale economics and a continued act of restraint. DeepSeek's implied valuation is a fraction of the figures now attached to its American rivals. Anthropic, which filed confidentially for an IPO on June 1 having just closed a $65 billion Series H, carries a post-money valuation of $965 billion. OpenAI raised $122 billion in March at an $852 billion valuation and is separately targeting a public listing.
Western export bans mean DeepSeek cannot access frontier American silicon. Without the ability to buy that hardware, they have no reason to match the multi-billion-dollar computing budgets of their U.S. rivals, said Alfredo Montufar-Helu, managing director at Ankura China Advisors in Beijing. That constraint, which was initially framed as a disadvantage, became the company's marketing story. The question now is whether the capital raise signals that the constraint is no longer sufficient, or that the story was always more complicated than a $6 million training run.
DeepSeek's V4 series became the company's first major AI family optimized to run on Huawei's Ascend AI accelerators instead of relying primarily on Nvidia hardware. That transition matters for what the raised capital might actually fund: not Nvidia clusters, but the domestic compute stack that Beijing has spent years trying to build.
What This Costs a Small Business to Care About
The practical answer is: nothing yet, and then potentially a lot. DeepSeek's V3 and R1 models are open-source. Small businesses and independent developers have been running them on commodity infrastructure since early 2025 at costs that made the U.S. lab pricing tiers look like a different industry. If this round funds faster model development and broader API distribution — which Tencent's cloud infrastructure would enable — the competitive pressure on OpenAI and Anthropic pricing stays in place, or intensifies.
The more specific risk to watch: DeepSeek has so far operated without enterprise SLAs, without uptime guarantees, without the compliance paper trail that procurement teams at larger companies require. A $7.4 billion war chest, run through Tencent's distribution, could close that gap. At that point, the question of which model a small business builds on gets considerably less obvious.
DeepSeek has not publicly disclosed any plans regarding a future initial public offering, and several parties involved have either declined to comment or have not responded to media inquiries. The company that became famous for saying nothing while building everything has not changed that particular habit.
