Here is the question institutional investors should be sitting with this morning: when the world’s largest EV battery maker and one of China’s most powerful internet conglomerates write the biggest checks into an AI model lab, what kind of asset are you actually buying?
Bloomberg reported Tuesday that DeepSeek is close to raising at least $12 billion in its latest funding round, having originally aimed for about $7.5 billion at a valuation of roughly $75 billion. People familiar with the matter say the total could approach $15 billion. Battery maker CATL and Tencent are contributing the largest shares. After the round closes, DeepSeek plans to restructure for an IPO in early 2027.
Why Wall Street Cares
The size alone would be notable. The composition is what changes the conversation. CATL does not run a venture fund. It builds batteries for electric vehicles and, increasingly, is positioning itself in “compute-energy integration” for AI data centers. Its DeepSeek stake is not a financial bet on model performance; it echoes China’s current efforts to achieve self-sufficiency in the entire AI chain, from models to power supply.
Tencent’s motive is different but equally strategic. It operates WeChat and competes directly with AI-native products. A large stake in the country’s leading model lab is a distribution and capability hedge, not a return-maximizing trade.
The investor selection policy makes this even harder to price through a conventional lens. Bloomberg has reported that DeepSeek has fielded strong interest from state-backed funds, investment arms of listed Chinese companies, and venture capital firms, but is vetting prospective investors closely, turning away private investment funds raised from individual investors and limiting the pool largely to government and corporate funds. That is a deliberate architecture, not a coincidence.
The Bull Case
The optimists argue that industrial and platform capital is smarter money than venture for this stage. CATL has skin in DeepSeek’s infrastructure success because its own energy storage business grows with every data center DeepSeek builds. Tencent needs DeepSeek to stay ahead of Alibaba’s Qwen and Baidu’s Ernie. Both investors have reasons to help DeepSeek win that go beyond share price appreciation. The strong investor interest also stems from DeepSeek’s V4-Flash model, which has been marketed as pushing the cost-performance frontier versus leading U.S. models, though many headline benchmark numbers are vendor-reported.
At investor meetings this year, DeepSeek founder Liang Wenfeng committed to keeping DeepSeek’s AI models open and said the company’s overarching aim is advancing toward artificial general intelligence, placing commercial returns below technological ambition. For holders of KWEB and FXI, that positioning is actually a feature: a well-capitalized open-weights Chinese model lab is a rising tide for the broader ecosystem.
The Bear Case
The concern is that what looks like a funding round is better understood as a state-industrial project with equity paperwork attached. From the first round, the structure was telling. The Information reported that China’s National Artificial Intelligence Industry Investment Fund invests directly into DeepSeek, is not subject to the lock-up, and enjoys voting rights at the company. Tencent’s larger check bought no governance influence. Foreign investors were effectively screened out.
That structure makes DeepSeek very difficult to value on conventional multiples. The all-domestic investor syndicate pairs platform giants with an industrial energy player and state AI capital, reducing the lab’s former neutrality and binding major Chinese platforms more tightly to a single model stack. Portfolio managers who try to price a 2027 DeepSeek IPO the way they would price an OpenAI listing are working with the wrong model.
What Investors Are Missing
The less-discussed consequence is what this does to the competitive read-through for U.S. hyperscalers and model companies. DeepSeek’s funding is accelerating its push toward domestic hardware independence. Reports have said the company plans to deploy at least 160,000 of Huawei’s Ascend accelerators at a massive data center under construction in Inner Mongolia, creating one of the largest publicly discussed clusters of Huawei AI chips. Every dollar of CATL and Tencent capital that funds that build-out is also a dollar funding a chip ecosystem that competes directly with Nvidia’s.
DeepSeek recently partnered with Huawei Technologies to release open-source programming tools optimized for Huawei’s Ascend AI chips, highlighting China’s push to reduce reliance on Nvidia’s AI ecosystem. That is the second-order story the headline numbers obscure.
Stocks to Watch
Tencent (TCEHY): Its lead position in DeepSeek gives it privileged access to the most competitive open-weights Chinese model. That strengthens its AI product layer across WeChat, games, and cloud. The DeepSeek stake is a strategic asset with optionality on the IPO.
CATL (CTATF): For CATL chairman Zeng Yuqun, the logic is that AI’s exponential growth comes with an insatiable appetite for electricity. CATL has been explicit about targeting data-center and “compute-energy” opportunities, but specific figures around how much it has deployed into AI data center infrastructure are not consistently verified in public reporting. DeepSeek is both customer and validation for that pivot.
KWEB / FXI: Both ETFs carry meaningful Tencent weight. A DeepSeek IPO in early 2027 at a valuation above $75 billion would be one of China’s biggest stock debuts in years, and the sentiment lift for Chinese tech broadly could be substantial even before the first trade.
Nvidia (NVDA): Watch it as a risk, not a beneficiary. A DeepSeek cluster running 160,000 Huawei Ascend chips would be a clear public proof point that China’s domestic AI stack can scale without U.S. silicon. Every successful iteration makes the export control argument harder to sustain as a durable Nvidia tailwind.
