DeepSeek Nears $12B Funding Round with Tencent and CATL, Doubling Its Original Target

What Happened

Citing people familiar with the matter, Bloomberg reports DeepSeek's latest round is nearing completion at no less than 80 billion yuan (~$12 billion), with Tencent and CATL committing among the largest amounts. The company originally targeted about 50 billion yuan, but demand from investors far exceeded expectations after the V4.1-Flash launch, pushing the round ever higher — the final tally could approach 100 billion yuan. Reuters relayed Bloomberg's report but says it could not independently verify it; none of the three parties responded to requests for comment.

This is DeepSeek's largest funding round ever, and the biggest private AI raise in China this year. The same day, also per Bloomberg, Moonshot AI closed its final private round at a $50 billion valuation and set its sights on a Q1 2027 Hong Kong IPO targeting $5 billion. Per the sources, the money will go to compute infrastructure, model R&D and hiring.

Key Facts

1. ¥50B → ¥80B+: demand rose after V4.1-Flash: The original target was 50 billion yuan; after September's V4.1-Flash launch (faster inference, higher throughput, built to scale to larger models), investor demand far exceeded expectations — at least 80 billion yuan (~$12 billion), possibly approaching 100 billion yuan (~$15 billion).
2. Tencent + CATL: Among the largest backers: Tencent (cloud, traffic, distribution) and CATL (batteries, storage, the power supply chain).
3. Three tracks: new model, Ascend tools, STAR Market IPO: DeepSeek-V4.1-Flash shipped in September; the lab recently partnered with Huawei on programming tools for Ascend chips; last month it hired CITIC Securities to prepare a STAR Market IPO — timing, valuation and raise size all undecided.
4. No valuation disclosed: The report gives no valuation. Moonshot AI, reported the same day, is at $50B on (projected) $2B ARR — about 25x forward price-to-sales. DeepSeek's revenue comes mainly from API inference volume (the V4.1 family is among the most-called open models in the world). All figures come from people familiar with the matter — the company has never disclosed financials.

Context

DeepSeek was founded in 2023 by Liang Wenfeng (a High-Flyer / quant background). In early 2025, R1 sharply cut inference costs and rattled global AI stocks; after that the lab did not raise outside money. Per earlier reporting, state-affiliated funds have since taken stakes, and now the lab is raising about $12 billion. Tencent, Alibaba, Meituan, IDG and state funds are on both DeepSeek's and Moonshot's cap tables.