AI company DeepSeek is close to wrapping up its latest funding round. According to multiple media reports, the round has nearly locked in at least 80 billion yuan, roughly $12 billion; based on term sheets already signed, the final figure could approach 100 billion yuan, far above the company's original target of about 50 billion yuan. CATL and Tencent Holdings are leading the round.

The money is there to close a computing power gap. DeepSeek is building a large data center in Inner Mongolia and plans to deploy at least 160,000 Huawei Ascend AI accelerator chips, which accounts for the other half of why the round was oversubscribed.

Round Size: At Least 80 Billion Yuan, Nearing 100 Billion on Signed Terms

According to multiple media reports, DeepSeek's current round has nearly locked in at least 80 billion yuan, about $12 billion. Based on investment term sheets already signed, the final size could approach 100 billion yuan.

That is well beyond the roughly 50 billion yuan the company initially set out to raise. People familiar with the matter said the target was revised upward repeatedly after investor demand ran ahead of expectations.

The deal has not been finalized and terms could still change. As of publication, DeepSeek, Tencent and CATL had not responded to the reports.

Tencent and CATL Lead, Paving the Way for an Early 2027 IPO

The round is led by CATL and Tencent Holdings, and the process is nearing completion.

The funding will lay the groundwork for an initial public offering DeepSeek plans for early 2027, according to reporting.

Valuation Starts at 500 Billion Yuan and May Rise

People familiar with the matter said DeepSeek initially targeted a valuation of around 500 billion yuan for the round.

After the release of its latest model, investor demand exceeded expectations and the round size was marked up repeatedly, which could lift the final valuation as well. The deal has not been finalized, and specific terms may still change.

A 51 Billion Yuan First Round Two Months Ago Was Already an Industry Record

This is DeepSeek's second large raise this year.

In June, the company closed its first external funding round since its founding, totaling 51 billion yuan at a post-money valuation of nearly 400 billion yuan, then the largest single round in China's AI industry. Founder Liang Wenfeng led that round with roughly 20 billion yuan of his own money, Tencent put in about 10 billion yuan, entities tied to CATL about 5 billion yuan, and the National AI Industry Investment Fund also participated.

Behind the Oversubscription: Market Reaction to V4.1-Flash

The immediate driver of the oversubscribed round was the market response to DeepSeek's new model release.

Industry observers say DeepSeek's earlier models reset expectations for how its price-performance compares with similar products from Anthropic and OpenAI. The company followed up in September with V4.1-Flash, pitched on stronger capability, faster inference and higher throughput.

The continued inflow of capital is, in effect, a price put on the pace of its technical iteration.

The Other Side of the Money: 160,000 Ascend Chips in Inner Mongolia

The flip side of the funding is enormous spending on computing power.

DeepSeek is building a large data center in Inner Mongolia and plans to deploy at least 160,000 Huawei Ascend AI accelerator chips. It has also worked with Huawei on programming tools for Ascend chips and open-sourced infrastructure components including TileLang.

Industry sources told reporters that computing power demand for model training and inference is the fundamental reason the round ballooned so quickly, and that the rollout pace of domestic computing clusters will directly shape its future fundraising and the valuation it can deliver at listing.

Notably, Liang Wenfeng said publicly in 2024 that the company's problem had never been money but the embargo on high-end chips. Industry watchers read the shift in this round's logic as an inevitable choice under constrained access to high-end computing power: trading capital for time and ecosystem.

Industry Context: The Funding Race Has Reached the Eve of IPOs

Across the industry, funding rounds at leading large-model companies are scaling up fast.

In July, Moonshot AI closed a round of about 25 billion yuan, also far above its initial target of roughly 7 billion to 14 billion yuan. Zhipu raised a combined HK$70 billion across rounds in July and September. Moonshot AI's latest private round valued it at about $50 billion, and it plans to list in Hong Kong in the first quarter of 2027.

The scramble for Chinese large-model stakes in the private market has pushed the funding race to the eve of IPOs.