Chinese tech giants Tencent and Alibaba are eyeing stakes in DeepSeek, which is targeting a valuation exceeding $20 billion as it launches its debut fundraising round, The Information reported on Wednesday.
The amount has surged to twice its original target of roughly $10 billion in just a few days, as sources point to exceptionally strong investor demand for the company’s first-ever outside funding round.
The company originally set out to raise at least $300 million. Now both the valuation and the total fundraising appear to be moving targets. Alibaba’s stock ticked up about 2% on the news.
The reversal
DeepSeek built its reputation on independence from external capital. Founder Liang Wenfeng funded the company through his hedge fund High-Flyer, reportedly spending about $5.6 million to train its flagship model, far below the multi-billion-dollar budgets typical in Silicon Valley.
Its DeepSeek-V3 model, released in early 2025, delivers performance comparable to much larger systems while using significantly fewer computing resources. That efficiency has become central to DeepSeek’s identity. The company is said to have declined multiple investment offers from leading Chinese venture firms and technology groups, with Liang previously showing skepticism about investors seeking time-bound returns.
Against that backdrop, the decision to pursue external funding marks a clear strategic shift rather than a conventional early-stage raise.
Why now?
While training a model for a relatively modest sum is one challenge, scaling it for widespread use introduces a very different cost structure. DeepSeek’s spending is understood to have shifted from one-off training to ongoing inference, infrastructure, and product integration, which are more difficult to sustain without external capital.
The company has also reportedly faced researcher departures and delays in its next model, increasing pressure on resources. With a workforce estimated at fewer than 300 employees, even limited attrition can have an outsized impact.
The hardware problem
DeepSeek’s architecture is built in part on NVIDIA PTX, optimized for chips from NVIDIA that are now subject to U.S. export restrictions to China.
Chinese authorities have promoted domestic alternatives, particularly from Huawei. Transitioning away from NVIDIA-optimized systems could present technical and operational challenges, especially if the existing infrastructure requires significant adaptation.
Reports indicate the company has already spent time evaluating alternative chip ecosystems, potentially diverting resources from core research.
If the funding round reaches $20 billion or more, DeepSeek would rank among the most valuable AI startups globally, alongside peers such as Anthropic and xAI.
