What happened
DeepSeek raised $15 billion in a new funding round, nearly doubling its original target, according to The Wall Street Journal. The company was valued at approximately $74 billion before the round closed. Investors included Tencent, Contemporary Amperex Technology, Stony Creek Capital, Legend Capital, and Geely. DeepSeek is preparing for a Shanghai IPO in early 2027, potentially listing before OpenAI.
Chinese AI startup DeepSeek has completed a $15 billion funding round, nearly doubling its initial target due to strong investor demand, as reported by The Wall Street Journal. The company was valued at approximately $74 billion prior to the closing of this round. This financing is part of an accelerated timeline for a Shanghai IPO planned for early 2027, which could precede OpenAI's potential public listing.
The round included participation from existing investors Monolith Management and Shixiang, as well as new backers such as Tencent, battery manufacturer Contemporary Amperex Technology, semiconductor-focused private equity firm Stony Creek Capital, Lenovo-backed Legend Capital, and automaker Geely. DeepSeek founder Liang Wenfeng informed investors in September that the company’s annualized revenue run rate had reached $1 billion, a significant increase from the $400 million to $500 million reported in July.
This revenue growth is attributed largely to cloud-based API access to its models. DeepSeek reportedly maintains gross margins of 70% to 80% on its V4 flagship model, despite charging less than U.S. competitors. In August, the company introduced peak-hour pricing that increased rates for flagship models by roughly fourfold, though they remain cheaper than major rivals. However, annualized revenue does not confirm profitability, especially given ongoing investments in computing infrastructure.
To reduce dependence on foreign chips, DeepSeek is expanding domestic computing infrastructure. While it continues to use Nvidia processors for complex AI training, the company is collaborating with Huawei to shift more training workloads onto Chinese chips. This strategic shift aims to mitigate supply chain risks from U.S. suppliers, although the specific financial benefits of this transition remain unquantified in the source material.
Why it matters
This significant capital injection allows DeepSeek to expand domestic computing infrastructure and reduce reliance on foreign chips, specifically by shifting training workloads to Huawei. The funding supports a strategy to maintain high gross margins while offering lower prices than U.S. competitors, challenging the sustainability of high-cost AI infrastructure models in the global market.
The $15 billion raise provides DeepSeek with substantial capital to sustain its aggressive expansion and infrastructure build-out. This is critical as investors increasingly question whether AI companies can convert rising infrastructure spending into sustainable profits. DeepSeek’s ability to maintain high margins while undercutting U.S. competitors on price presents a significant competitive challenge to the prevailing high-cost AI business model.
The planned Shanghai IPO will subject DeepSeek to greater public scrutiny regarding its revenue, margins, and infrastructure spending. This listing will serve as a test case for whether a lower-cost AI strategy can generate sustainable profits and justify a valuation of approximately $74 billion in the public markets. The timing, potentially before OpenAI, highlights the intensifying race for public market validation in the AI sector.
The strategic pivot toward Huawei chips for training workloads is a major geopolitical and technical development. By reducing reliance on Nvidia and U.S. suppliers, DeepSeek is insulating itself from potential export controls and supply chain disruptions. This move could influence other Chinese AI companies to adopt similar domestic hardware strategies, reshaping the global AI hardware landscape.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
Which component of an AI application is the machine-learning model itself?
What to watch next
Monitor the progress of DeepSeek's shift from Nvidia to Huawei chips for training workloads. Watch for the official IPO filing and how public scrutiny affects its reported revenue run rate and infrastructure spending. Observe if the peak-hour pricing strategy continues to drive revenue growth without losing market share to cheaper or more established rivals.
Track the technical and financial outcomes of DeepSeek’s migration of training workloads from Nvidia to Huawei chips. Success in this transition would validate the feasibility of high-performance AI training on domestic Chinese hardware, potentially altering the global balance of AI infrastructure power.
Monitor the official IPO filing and subsequent public disclosures. These documents will provide verified data on DeepSeek’s profitability, cash flow, and specific infrastructure costs, offering a clearer picture of the sustainability of its business model compared to private estimates.
Observe market reactions to DeepSeek’s pricing strategies, particularly the impact of peak-hour pricing on enterprise adoption. If DeepSeek can maintain its market share while increasing revenue per user, it may force competitors to adjust their own pricing structures or value propositions.