Deepseek is moving toward a much larger funding round than it first planned, with investor interest building around its latest AI model work and a possible IPO timeline.
A bigger round than originally planned
Deepseek is close to raising at least $12 billion in a new funding round, Bloomberg reports. The Chinese AI startup had originally aimed to raise about $7.5 billion at a valuation of roughly $75 billion.
People familiar with the matter say the total could approach $15 billion. That would make the round substantially larger than the company’s initial target, while keeping attention on how quickly Deepseek is trying to scale its capital base.
Battery maker CATL and Tencent are contributing the largest shares. Their role matters because the source describes them as the biggest contributors in the round, not merely as small participants.
CATL declined to comment. Tencent and Deepseek didn’t respond.
Why investors are paying attention
The source ties the strong investor interest to Deepseek’s V4-Flash model. The model has set new benchmarks for cost and performance against Anthropic and OpenAI.
That point is central to the funding story. Deepseek is not described as raising capital only because of broad enthusiasm around AI. The article connects the round directly to a model that has drawn attention for both cost and performance.
Deepseek released V4-Pro and V4-Flash as preview versions in April. Those releases now sit at the center of the company’s momentum, especially as investors assess whether its model work can support a larger valuation and a public-market path.
The company’s strategy also includes developing open models with freely available weights. Founder Liang Wenfeng wants to keep that approach, according to the source.
The IPO path depends on restructuring
After the round closes, Deepseek plans to restructure for an IPO in early 2027. The timing is not immediate, but the funding round appears to be part of the company’s preparation for that next phase.
Restructuring before an IPO can signal that a company is aligning its organization, capital structure, or operating setup for public-market expectations. The source does not provide those details, so the clearest takeaway is narrower: Deepseek plans to close the round first, then restructure with an early 2027 IPO in view.
The new round also follows another recent financing milestone. In June, Deepseek closed its first external funding round at about $7.4 billion with a valuation above $50 billion.
Taken together, the two rounds show a sharp increase in the scale of outside capital being discussed around Deepseek. The first external round came in June; the newer round is now described as being close to at least $12 billion, with the possibility that it could approach $15 billion.
Chips remain a key constraint
Deepseek is building a data center with at least 160,000 AI chips from Huawei. That infrastructure plan sits alongside the company’s model development and IPO preparation.
The chip story is complicated by Deepseek’s reliance on major suppliers. The company had paused the round after comments by Liang about the company’s reliance on Nvidia chips went viral.
Deepseek is also working on its own inference chip to reduce its dependence on Nvidia and Huawei. The source does not say when that chip will be ready or how it will perform, but it does state the strategic purpose: lowering dependence on those suppliers.
That makes chips one of the clearest operational themes in the article. Deepseek is using Huawei chips for a large data center, has faced attention over Nvidia reliance, and is developing its own inference chip as part of an effort to reduce dependence on Nvidia and Huawei.
What to watch next
The immediate question is whether the round closes at at least $12 billion, or whether it moves closer to the $15 billion level described by people familiar with the matter.
Several other points will shape how the story develops:
- whether CATL and Tencent remain the largest contributors when the round closes;
- how Deepseek’s restructuring proceeds before the planned IPO in early 2027;
- whether V4-Flash continues to support investor interest after setting new benchmarks for cost and performance against Anthropic and OpenAI;
- how the company balances Huawei chips, Nvidia reliance, and its own inference chip work;
- whether Deepseek continues developing open models with freely available weights.
For now, the funding round shows that Deepseek is trying to turn technical momentum into financial scale. Its next steps will be measured not only by the size of the raise, but also by how it handles infrastructure, chip dependence, and the restructuring needed before its planned IPO.