Chinese artificial intelligence startup DeepSeek is close to completing its first external funding round, raising over 50 billion yuan (approximately $7.4 billion or £5.5 billion) at a valuation of just under $60 billion. That marks a dramatic six-fold increase from the company's $10 billion valuation just two months ago, according to people familiar with the matter.
The funding round has attracted a broad group of strategic and financial investors. Tencent Holdings is expected to invest around 10 billion yuan, with NetEase and JD.com each expected to contribute about 3 billion yuan. CATL, the world's largest electric vehicle battery manufacturer, is also expected to invest approximately 5 billion yuan. Reports indicate that venture capital firms including IDG Capital, Monolith, Loyal Valley Capital, and Shixiang Tech are participating as well.
DeepSeek's founder and chief executive, Liang Wenfeng, is expected to inject roughly 20 billion yuan of his own capital into the round. State-backed funds may also join, though final terms and the specific investment vehicles have not yet been finalized. The involvement of state-linked investors underscores the strategic importance of DeepSeek within China's broader push for technological self-reliance in artificial intelligence.
Why DeepSeek Is Now Seeking Outside Capital
For a long time, DeepSeek had no need or desire for external funding. The startup is backed by High-Flyer Quant, a highly profitable quantitative hedge fund also founded by Liang. That internal financial engine gave DeepSeek the freedom to pursue ambitious research without pressure from outside investors. However, local media reports indicate that DeepSeek recently decided to open its doors to external capital for a specific reason: to secure a well-defined market valuation that helps retain top talent.
In China's intensely competitive AI talent market, equity compensation is a major lever. A clear, credible valuation makes it easier to offer potential hires and existing employees meaningful stock-based incentives. Without an external round, the company's shares remained hard to price, complicating efforts to lure senior researchers and engineers away from rivals such as Alibaba's Qwen, Baidu's Ernie, and a growing list of well-funded startups.
By bringing in strategic investors like Tencent, JD.com, and CATL, DeepSeek is not only raising cash but also building alliances across technology, e-commerce, and manufacturing. These partnerships may prove valuable as DeepSeek looks to integrate its AI models into a wider range of products and services, from cloud computing to enterprise software to industrial applications.
A Disruptive Force in AI
DeepSeek burst into the global AI consciousness in early 2024 with the release of powerful open-weight models that were remarkably cheap to train and run. Its R1 model, introduced in January 2025, sent shockwaves through global markets, briefly rattling AI-focused tech stocks and prompting Silicon Valley executives to reconsider the assumptions underpinning massive AI infrastructure spending.
The core of DeepSeek's advantage lies in its engineering ingenuity. The company has made significant advances in mixture-of-experts architectures, multi-head latent attention, and reinforcement learning techniques that reduce computation costs while maintaining strong performance. These innovations allow DeepSeek to deliver competitive results at a fraction of the cost of leading American models, a fact that upended the prevailing logic that cutting-edge AI required ever-larger capital investments.
The company's latest V4 model, released in April, continues this trend. Shortly after its release, DeepSeek said that prices would drop significantly in the second half of the year, when Huawei's Ascend 950PR supernodes are expected to ship at scale. The Ascend 950PR is a high-performance server platform designed for large-scale AI training and inference, and its arrival could further reduce the cost of running DeepSeek-based applications in China.
China's AI Supply Chain Moves Toward Self-Reliance
Domestic chipmakers are paying close attention. Huawei, Cambricon, and MetaX, among others, have announced that they are optimizing their AI processors for DeepSeek's V4 model. This is part of a broader national effort to build a resilient, self-sufficient AI supply chain that does not depend on advanced semiconductors from Nvidia, which are subject to strict US export controls.
The optimization work is critical because many of the software stacks and tools used to train and deploy large language models were originally developed to run on Nvidia's CUDA platform. Chinese chipmakers have been forced to build their own software ecosystems, but model-specific optimizations can go a long way toward closing the performance gap.
DeepSeek has been eager to support this cause. By ensuring its models run efficiently on domestic chips, the company makes it easier for Chinese enterprises to adopt AI without purchasing Nvidia hardware. This aligns well with government policy and with commercial interests, as domestic hardware is often cheaper and more readily available.
The Strategic Importance of DeepSeek's Valuation
A $60 billion valuation puts DeepSeek in the rarefied air of the world's most valuable AI companies. For comparison, OpenAI has been valued at far higher figures, but DeepSeek has achieved a notable milestone without relying on Microsoft-style mega-deals or large-scale enterprise sales. The company is still in relative terms a research-driven startup with a strong focus on engineering excellence.
Tencent's participation is particularly telling. Tencent already operates its own large AI models, including the Hunyuan family, and has invested broadly across the Chinese technology landscape. By taking a stake in DeepSeek, Tencent gains exposure to an independent, high-performing AI lab while preserving the autonomy of its internal AI efforts. Tencent also has deep pockets to invest in AI infrastructure, and DeepSeek's models could ultimately be offered through Tencent Cloud, strengthening its competitive position against Alibaba Cloud and Huawei Cloud.
JD.com and NetEase bring complementary strengths. JD.com is the largest retailer in China and has a sprawling logistics network; AI models that improve warehouse automation, demand forecasting, and customer service could deliver enormous value. NetEase, which has strong positions in gaming, music, and education, sees the potential for generative AI to transform content creation and interactive computing.
CATL, meanwhile, represents an interesting bridge between the digital and physical worlds. As the world's biggest EV battery manufacturer, CATL is focused on manufacturing efficiency, supply chain optimization, and intelligent management of batteries across their lifetime. DeepSeek's models could be used to optimize factory operations or to predict battery performance, though a financial stake likely has more to do with CATL being a major Chinese industrial champion looking for exposure to frontier technology.
What Happens Next
The final documentation and closing of the round are still pending, and the exact roster of investors and the size of state-backed participation could change. Reports suggest that the round is being led by established venture capital firms with strong government ties, which would help ensure regulatory approval and smooth the path for any future public listing in Hong Kong or elsewhere.
For Liang Wenfeng, the round represents a deep personal commitment. By placing more than 20 billion yuan of his own wealth into the company, he is signaling conviction in DeepSeek's trajectory. Liang, who was born in the mid-1980s and earned a master's degree from Zhejiang University, started High-Flyer in 2015 and built it into one of China's largest quantitative funds. He then redirected substantial profits and talent toward what was initially a side project exploring artificial general intelligence. That project eventually became DeepSeek.
Liang has been described as an unconventional founder who prefers long-term research over short-term revenue. In interviews and rare public statements, he has emphasized the importance of original innovation over rapid product iteration. The new funding will give DeepSeek more resources to pursue that mission, while also creating market discipline through external shareholders.
DeepSeek's rise comes at a time when Chinese regulators are cautiously encouraging AI innovation while also imposing safeguards on large-scale AI systems. The company's open-weight distribution model, under which it releases model weights to developers around the world, has been both praised and scrutinized. Outside researchers have used or studied DeepSeek models extensively. The company has also faced security probes from governments, including Italy's privacy regulator and South Korea's data protection authority, over how it handles user data.
Despite these challenges, DeepSeek's underlying technology remains in high demand. Its models have been adopted by developers, independent engineers, and even rival AI labs that use the weights to fine-tune specialized systems. The release of V4 has reinforced DeepSeek's standing as the most respected Chinese developer of open-weight language models.
The company is also pushing forward on multimodal capabilities, reasoning, and agentic workflows, areas that are central to the next wave of AI commercial applications. With new capital and a clear valuation, DeepSeek can afford to pay competitive salaries, fund ambitious research clusters, and sustain the steady cadence of model releases that has made it a global talking point.
In a broader sense, DeepSeek's successful funding round demonstrates the continuing vigor of China's AI ecosystem even as geopolitical tensions limit access to cutting-edge chips. The company has proven that constraints can sometimes breed creativity. DeepSeek's training efficiency innovations are now widely studied by researchers in Europe and the United States, making it an exporter of ideas, not just a consumer of technology.
As the company closes this first external round, the focus will shift to execution: how it deploys the capital, whether it can sustain the technical momentum, and whether its strategic investors can open new distribution channels. For now, the numbers alone, a $60 billion valuation, more than $7 billion raised, and participation from some of China's most powerful firms, signal that DeepSeek has crossed a major threshold. It is no longer a research collective funded by a hedge fund; it has become a major independent player with global ambitions.
Source: Silicon UK News