DeepSeek became famous in 2025 for doing the opposite of what everyone expected. While leading American labs spent fortunes to build frontier artificial intelligence (AI), the little-known Chinese company produced a model that matched many of their capabilities at a fraction of the apparent cost, operating for years without taking a dollar of outside money. Now, in a striking reversal, it is preparing to raise more money in one go than any Chinese AI company ever has.
What Was Reported
The details come from reporting by Bloomberg, citing people familiar with the matter, and were widely picked up by Chinese financial media on October 6, 2026.
According to those reports, Hangzhou-based DeepSeek is close to locking in a new funding round of at least 80 billion yuan, roughly $12 billion at exchange rates around 7.1 yuan per dollar. Under the signed term sheet — a document outlining the agreed deal terms — the final size could approach 100 billion yuan, well above the company's original target of about 50 billion. The two biggest committed investors are said to be Tencent, the internet and gaming giant, and CATL (Contemporary Amperex Technology Co., Limited), the world's largest maker of electric-vehicle batteries. The process was described as nearing completion, though the deal had not been finalized and terms could still change.
The round would set a Chinese record for a single AI investment and, the reports suggest, pave the way for a potential initial public offering (IPO) — a stock-market listing — as early as 2027. The company's target valuation was reported at around 500 billion yuan.
From 'Never Take Outside Money' to a Record Round
The turnaround is the interesting part of the story.
DeepSeek was founded in July 2023 by Liang Wenfeng and nurtured by High-Flyer, a prominent Chinese quantitative hedge fund he helped build. For its first years, the company had a firm rule: no outside financing, no dilution of ownership, no letting investors set its timetable. The fund's profits covered its research costs. That independence was central to its reputation for doing more with less.
That stance already shifted earlier in 2026. According to Chinese reporting, DeepSeek completed its first external round in June, raising about 51 billion yuan at a post-investment valuation of nearly 400 billion yuan — itself a domestic record — with founder Liang, Tencent, CATL, NetEase, JD.com, and a national AI investment fund among the participants. The new round would come only months later and dwarf it.
Why the Change of Heart?
Chinese reports point to three pressures that made self-funding harder to sustain.
First, the cost of staying at the frontier keeps climbing. The reported gap between the best American models and the strongest Chinese rival had narrowed to just a few percentage points, according to Stanford's 2026 AI Index, meaning further gains are harder and more expensive to win. Moving to the next generation of models — measured in trillions rather than hundreds of billions of parameters — demands exponential increases in computing power and data work.
Second, DeepSeek faced talent losses, with several core researchers reported to have left for large Chinese technology companies. Paying competitive salaries to keep engineers is itself expensive. Third, the money gap with American rivals widened visibly: OpenAI completed a roughly $122 billion round earlier in 2026. In that environment, even an unusually efficient lab needs capital at a scale no single fund's profits can comfortably cover.
What the Money May Buy
The reported plans give a sense of the direction.
Chinese media say DeepSeek intends to deploy at least 160,000 of Huawei's high-end AI accelerators in a large data center it plans in Inner Mongolia — potentially one of the largest known clusters of Huawei AI chips. The company has also open-sourced infrastructure components built for Huawei's AI computing platform. The strategy, in other words, is to pour the new capital into domestic computing power and the next flagship model, reducing reliance on the American chips that export controls have made hard to obtain.
The Honest Comparison With the United States
This is where the "record" needs context.
DeepSeek's round would be the largest in China, but the American leaders have raised even vaster sums. OpenAI completed a round of about $122 billion earlier in 2026 at a reported valuation of roughly $852 billion, with Amazon, Nvidia, and SoftBank among the backers. Rival Anthropic went further: in May 2026 it announced a $65 billion round that valued it at about $965 billion, briefly making it the world's most valuable private AI company, with Amazon, Google, and major chip and memory makers among its investors. Anthropic had earlier raised rounds of $30 billion and $13 billion within months.
Set those figures side by side and the picture is clear. A single American round can be several times DeepSeek's entire reported raise, and the valuations of the leading U.S. companies — close to a trillion dollars — stand well above DeepSeek's reported target of 500 billion yuan, roughly $70 billion. China's most efficient lab is raising serious money, but it is joining an American-led capital contest rather than overtaking one.
Two Different Funding Models
There is also a difference in where the money comes from.
The American rounds lean heavily on global technology giants, cloud providers, chip companies, and large international investors — OpenAI and Anthropic are bankrolled in part by the very companies (Microsoft, Amazon, Google) that distribute their products. DeepSeek's reported backers are overwhelmingly domestic Chinese industry leaders and state-linked funds. Tencent and CATL are strategic industrial investors rather than purely financial ones, and a national AI fund adds an explicit policy dimension. Where the U.S. race is driven by private global capital and cloud economics, China's is coordinated more around national champions, domestic chips, and a goal of technological self-reliance.
What to Keep in Mind
It is important to distinguish reported from completed. The Bloomberg figures came from unnamed sources and the deal was described as not yet finalized, so the exact amount and valuation could change. Currency conversions are approximate and shift with exchange rates. The size of prior rounds, valuations, and the Inner Mongolia chip cluster were reported by Chinese media rather than confirmed in audited filings. DeepSeek itself had not issued a formal public announcement. Finally, raising large sums is not the same as spending them well: whether the capital turns into stronger models, a large Huawei-based computing cluster that performs as hoped, and a sustainable business is the open question — a point Chinese financial commentators themselves made.
What to Take Away
DeepSeek's likely $12 billion round is a genuine milestone. It marks the moment China's most famously frugal, independent AI lab fully entered the capital-intensive big leagues, betting on domestic computing power to build its next generation and prepare for a possible 2027 listing.
But an honest ledger does not let the headline stand alone. The round is a Chinese record set against American rivals who have raised tens of billions more at valuations approaching a trillion dollars, funded by a different, more globally integrated investor base. DeepSeek once proved that clever engineering can narrow a money gap. Its new billions show that even the cleverest engineers, at the frontier, eventually need to pay for scale — and that in the U.S.–China AI contest, the financial race, like the technical one, is far from settled.