Chinese artificial intelligence start-up DeepSeek is finalising its first external funding round of more than 50 billion yuan ($7.4bn, £5.5bn) at a valuation of just under $60bn, according to a report by the South China Morning Post. The valuation represents a more than six-fold jump from the $10bn valuation the company commanded just two months ago, underscoring the intense investor appetite for frontier AI players in China.
The funding round is expected to bring together some of the biggest names in Chinese technology and industry. Tencent Holdings is reportedly set to invest 10bn yuan, while NetEase and JD.com are expected to contribute about 3bn yuan each. CATL, the world's largest electric vehicle battery manufacturer, is expected to pump in roughly 5bn yuan. Venture capital firms including IDG Capital, Monolith, Loyal Valley Capital and Shixiang Tech are also said to be participating. DeepSeek's founder and chief executive, Liang Wenfeng, is expected to inject about 20bn yuan of his own capital into the round.
Surging valuation and strategic shift
DeepSeek, backed by Liang Wenfeng's highly profitable quantitative fund High-Flyer Quant, has long been an outlier in China's AI ecosystem because it initially resisted bringing in external capital. The startup relied on its own resources to fund research and development, a strategy that allowed it to maintain independence and focus on long-term breakthrough innovation. However, local media reports suggest that DeepSeek recently decided to open up its cap table to outside investors in order to secure a well-defined market value, a step that is essential to attracting and retaining top talent in the fiercely competitive AI job market.
The valuation surge reflects not only DeepSeek's technical achievements but also the broader global excitement around generative AI. DeepSeek disrupted the AI industry last year with the release of highly powerful and cost-efficient models that challenged the assumption that cutting-edge AI required enormous computational budgets. Its models are seen as a proof point that Chinese firms can compete with American giants such as OpenAI, Google and Meta, even with certain restrictions on advanced chip access.
Founder and High-Flyer Quant connection
Liang Wenfeng is not a typical AI startup founder. He previously built High-Flyer Quant into one of China's most successful quantitative hedge funds, leveraging machine learning and big data to generate outsized returns in the financial markets. That background gave him an unusual combination of technical expertise and financial firepower. High-Flyer Quant reportedly provided the initial capital and computing resources that allowed DeepSeek to train large-scale models without relying on venture capital.
The connection to High-Flyer Quant also helped DeepSeek secure a huge cluster of GPUs before US export controls tightened. Reports have suggested that High-Flyer had accumulated thousands of Nvidia chips, giving DeepSeek a significant early advantage in model training. This infrastructure enabled the startup to move quickly from research to deployment, something that many well-funded rivals struggled to achieve.
Breakthrough models and price efficiency
DeepSeek's reputation was built on a series of model releases that combined strong performance with remarkable cost efficiency. The startup's V4 model, released in April, was particularly notable for its ability to compete with leading frontier models while using far fewer computing resources. At the time of the V4 release, DeepSeek said that prices would drop significantly in the second half of this year, when Huawei's Ascend 950PR supernodes are expected to ship at scale. That statement was widely interpreted as a signal that DeepSeek is actively optimising its models to run on domestic Chinese chips, reducing its dependence on Nvidia's high-end processors.
DeepSeek's approach to model development is markedly different from Western AI labs. Instead of chasing ever-larger parameter counts, the company has focused on architectural innovations, data efficiency and inference-time optimisation. This strategy has allowed it to offer API access at prices far below competitors, prompting a wave of adoption among Chinese developers and enterprises. The models have also been used to power applications in translation, code generation, customer service and financial analysis.
China's local AI supply chain
The reported funding round comes at a time when Chinese chipmakers are racing to close the gap with Nvidia. Huawei, Cambricon and MetaX have all announced that they are optimising their AI processing chips for DeepSeek's V4 model. This collaboration is part of a broader national drive to build a locally developed AI supply chain, reducing China's reliance on imported semiconductors amid US export controls.
Huawei's Ascend 950PR supernodes are particularly important for DeepSeek's rollout plans. These supernodes are designed to deliver high-throughput performance for large-scale AI training and inference, and they are expected to become widely available in the latter half of the year. If DeepSeek can seamlessly deploy its V4 model on Ascend hardware, it would be a major milestone for China's self-sufficiency in AI infrastructure.
Other Chinese chipmakers are also deepening their ties with DeepSeek. Cambricon, which specialises in AI processors, has been developing software stacks that are compatible with popular AI frameworks used by DeepSeek. MetaX, a newer entrant in the AI chip market, has reportedly been working closely with DeepSeek to optimise memory bandwidth and kernel performance for the V4 architecture. These efforts are not purely technical; they carry strategic significance for China's goal of creating a competing AI ecosystem to the NVIDIA CUDA platform that dominates the global market.
Talent retention and market value
One of the key reasons DeepSeek has decided to accept external investors is to address the intense competition for AI talent. The market for AI researchers and engineers in China is extraordinarily competitive, with both domestic tech giants and well-funded startups offering lucrative compensation packages and equity incentives. A clearly defined valuation allows DeepSeek to create a meaningful stock option pool for its employees, which is essential for keeping top talent from being poached by larger rivals such as Alibaba, Baidu or ByteDance.
Liang Wenfeng has spoken in the past about the importance of retaining researchers who are willing to tackle ambitious, long-term projects. The funding round will also allow the company to expand its research team, scale its computing infrastructure and invest in new areas such as multimodal AI, robotics and autonomous agents. With over 50 billion yuan in new capital, DeepSeek will be in a strong position to accelerate its roadmap and extend its lead in cost-efficient AI development.
The participation of state-backed funds adds another layer of significance. While final terms have not been settled, the expected involvement of government-affiliated investment vehicles signals that Beijing sees DeepSeek as a strategic asset for the nation's AI ambitions. This aligns with broader policy efforts to support indigenous innovation in critical technologies, from semiconductors to large language models.
Investor landscape
Tencent's expected 10bn yuan investment makes the internet giant one of the largest external shareholders in DeepSeek. Tencent has been aggressively building its AI capabilities, including the development of its own Hunyuan large language model and deep integration of AI into WeChat, gaming and cloud services. Holding a substantial stake in DeepSeek would give Tencent access to one of the most efficient model architectures in China, complementing its internal research efforts.
NetEase and JD.com, each contributing about 3bn yuan, are also strategic investors. NetEase has been exploring AI-driven content generation and interactive media, while JD.com relies heavily on AI for logistics, supply chain management and customer service. CATL, the EV battery giant, may seem like an unusual investor, but the company has been investing heavily in AI for battery research, manufacturing automation and energy management. These diverse investors reflect the broad applicability of DeepSeek's technology across multiple industries.
Existing venture capital firms in the round, such as IDG Capital and Loyal Valley Capital, have long track records in backing Chinese technology companies. Their participation lends credibility and financial governance to DeepSeek as it transitions from a research lab backed by a quant fund to a commercially oriented AI company.
Competitive dynamics and global impact
DeepSeek's rise has reshaped the competitive landscape for AI both in China and globally. Its models have demonstrated that high performance can be achieved at a fraction of the cost of US rivals, prompting companies like OpenAI and Google to re-evaluate their pricing and efficiency strategies. In China, DeepSeek is often compared with Alibaba's Qwen, Baidu's Ernie, and ByteDance's Doubao, but it stands out because of its strong focus on open-weight releases and developer-friendly APIs.
The startup has also become a symbol of China's resilience in the face of semiconductor sanctions. By optimising its technology to run on domestic hardware, DeepSeek has shown that Chinese companies do not necessarily need the most advanced US chips to produce world-class AI. This is a message that resonates not only with Chinese investors but also with governments and enterprises in other countries that are wary of relying on US-dominated AI infrastructure.
At the same time, DeepSeek faces challenges. International expansion could be complicated by data governance regulations and by security scrutiny from Western governments concerned about the transfer of sensitive technology. The company must also navigate an increasingly complex regulatory environment in Beijing, where AI models are subject to safety assessments and content controls. Maintaining its independence while bringing on powerful corporate and state-backed investors will require careful balancing.
The reported funding round is still subject to final agreement, and the exact vehicles through which state-backed funds will provide capital have not been confirmed. But the direction is clear: DeepSeek is preparing for its next stage of growth, with a valuation that places it among the most valuable AI companies in the world. The coming months will show how the new capital helps the startup scale its operations, retain its talent and continue to push the boundaries of cost-efficient AI.
Source:Silicon UK News