DeepSeek Is Reportedly Raising $12 Billion From Tencent, CATL And State-Linked Funds

In this storyDeepSeek

DeepSeek is close to raising at least 80 billion yuan, about $12 billion, with Tencent and the battery manufacturer CATL among its largest backers, Bloomberg reported on 6 October, citing people familiar with the matter.

The round has exceeded its initial 50 billion yuan target, and signed term sheets suggest the final figure could approach 100 billion yuan. Bloomberg reports a target valuation of around 500 billion yuan, roughly $75 billion, and describes the money as pre-IPO capital for a planned listing in early 2027. Proceeds are directed at computing infrastructure, model development and hiring. The company is expected to restructure once the round closes.

Reuters said it could not independently verify the report, and DeepSeek, Tencent and CATL did not respond to requests for comment.

The round would be considerably larger than DeepSeek's first outside funding earlier this year, which raised about 50 billion yuan, roughly $7.4 billion, at a valuation reported between $52 billion and $59 billion. Founder Liang Wenfeng was the largest investor in that round, putting in about 20 billion yuan of his own money. Tencent added around 10 billion yuan and CATL about 5 billion, alongside JD.com, NetEase and China's national AI investment fund.

The Hangzhou lab is known for its open-weight R1 and V4 model families and for training approaches that achieve competitive performance at low cost. It began seeking fresh capital in July, a month after closing the first round.

Tencent, which operates WeChat and a large cloud business, is reported to be weaving AI through its product range. CATL, which makes batteries for electric vehicles and is developing solid-state cells, may be seeking a position in the data centre supply chain.

Timing, valuation and scale of the listing have not been determined.

A battery manufacturer putting billions into an AI lab is the detail worth stopping on.

CATL makes cells for electric vehicles. It has no obvious use for a language model and no history of investing in software. The explanation reported is that it is seeking a position in the data centre supply chain, and once stated that way it stops being strange.

Data centres need uninterrupted power. As grids tighten, that increasingly means storage on site rather than a connection and a diesel backup. A company that makes the world's batteries at scale has a direct commercial interest in how fast AI compute is built, and a shareholding in one of the largest buyers of that compute is a reasonable way to sit close to the demand.

It also illustrates how differently China's AI buildout is financed. DeepSeek's backers are Tencent, CATL, JD.com, NetEase and a national investment fund: platform companies, industrial manufacturers and the state. There is almost no venture capital in that list. In the United States the equivalent round would come from venture funds, sovereign wealth and the chip supplier itself, as with Nvidia's investments in OpenAI and others.

Corporate and state money buys strategic access rather than a financial return, which means it is more patient and less sensitive to whether the valuation is defensible. That is an advantage while building, and it is the reason the eventual listing price will be the more informative number.

The repricing speed deserves attention on its own.

In April, investors were reportedly discussing a $300 million round at a $10 billion valuation. In May, DeepSeek closed roughly $7.4 billion at $52 billion or more. By July it was seeking capital at 500 billion yuan. The current round is reported at that target, around $75 billion.

That is a sevenfold move in valuation across five months, in a market where investors are overwhelmingly corporate and state rather than financial. Western AI labs have repriced quickly too, but through venture rounds where a mark has to be defensible to limited partners who will eventually want cash back. Strategic and state capital answers to a different test, and valuations set that way say less about what the asset is worth than they appear to.

The listing plan is where the difference becomes concrete. DeepSeek is preparing to go public in early 2027, which would make it the first major AI lab to list anywhere. OpenAI ruled out a 2026 listing on safety grounds last month while reportedly seeking private capital at $1.5 trillion. Anthropic has been reported as marketing an offering from mid-October.

A Chinese lab listing domestically would create the first public market price for a frontier model developer, set by investors operating under different disclosure rules and a different investor base from the one that would price an American equivalent. Whatever that number is, everyone else will be measured against it.

For daily, sharp analysis of the biggest moves in the Indian business and startup ecosystem, follow StartupFox.