DeepSeek, the Chinese artificial intelligence startup that shook the global technology sector with its open-weight models, is finalising its first external funding round at a valuation of nearly $60 billion. The company is raising more than 50 billion yuan, or approximately $7.4 billion, according to people familiar with the matter. That marks a more than six-fold increase from the $10 billion valuation the company commanded just two months ago, reflecting the astonishing pace of adoption and strategic importance of its technology.
A landmark funding round with heavyweight backers
Investment in the round is being led by some of the most prominent names in Chinese technology and manufacturing. Tencent Holdings, the internet and gaming conglomerate, is expected to invest 10 billion yuan. The report also names NetEase and JD.com, each likely to contribute around 3 billion yuan. These stakes underline the growing significance of artificial intelligence infrastructure to virtually every sector of the digital economy.
CATL, the world’s largest electric vehicle battery manufacturer, is also expected to inject approximately 5 billion yuan. The participation of CATL is particularly notable because it signals crossover interest from the energy and automotive sector. AI data centres, which require substantial amounts of electricity and advanced cooling systems, are becoming a emerging area of interest for energy giants. By investing in DeepSeek, CATL is likely seeking a foothold in the compute-intensive ecosystem that will underpin the next generation of smart manufacturing and autonomous vehicles.
Venture capital firms including IDG Capital, Monolith, Loyal Valley Capital and Shixiang Tech are also reported to join the round. Their presence is a reminder that DeepSeek is also attracting conventional technology investors who have long backed high-risk, high-reward startups. The involvement of these funds suggests that the financial community sees DeepSeek not merely as a research lab but as a scalable commercial enterprise.
DeepSeek’s founder and chief executive Liang Wenfeng is reportedly contributing around 20 billion yuan of his own capital. Liang, who previously built the highly profitable quantitative hedge fund High-Flyer Quant, is one of China’s most private and understated tech founders. His willingness to put his own money at risk reinforces his conviction in the company’s long-term mission. It also sends a powerful signal to outside investors who might worry about founder commitment.
State-backed funds are also expected to take part, although final terms, including which specific vehicles will provide the capital, have not yet been settled. The participation of state-backed institutions is a common feature of large Chinese technology financing rounds, especially for companies deemed strategically important to national competitiveness. DeepSeek has already been celebrated by Chinese officials as an example of the country’s ability to innovate in advanced artificial intelligence despite restrictions on cutting-edge chip imports.
Why DeepSeek changed course on external capital
For years, DeepSeek avoided outside investment altogether. The startup was comfortably financed by Liang’s High-Flyer Quant, which generated enormous revenues from algorithmic trading. This financial independence allowed DeepSeek to pursue long-term research without the pressure of quarterly earnings guidance or investor demands for fast commercialisation.
However, local media reports indicate that DeepSeek recently decided to secure external funding in order to establish a well-defined market valuation. This is a crucial step for talent retention. Equity allocation becomes a powerful tool for attracting and keeping top engineers, researchers, and executives, particularly in China’s fiercely competitive AI sector. A clearly valued company can offer employees a credible path to significant financial upside, something that standalone startups often lack.
The company’s change of heart also reflects the shifting realities of the global AI race. DeepSeek’s research breakthroughs require substantial computing power, which in turn requires capital to purchase or rent advanced silicon. The new round will give DeepSeek the war chest needed to expand its infrastructure, fund research and development, and potentially commercialise its models in new ways.
Liang’s dual role as both founder of DeepSeek and operator of High-Flyer Quant has always raised intriguing questions. The hedge fund acted as an early source of funding and compute resources, allowing DeepSeek to function as an independent research outfit. Yet for many years, this arrangement was unusual and at times opaque. External investors will now bring greater scrutiny and governance expectations, potentially professionalising DeepSeek’s operations while preserving its unconventional culture.
Disruptive technology that changed the AI landscape
DeepSeek first entered the global spotlight in late 2024 with the release of models that demonstrated surprising performance at a fraction of the cost of American rivals. Its innovative use of mixture-of-experts architectures, advanced distillation techniques, and optimised training methods challenged the conventional wisdom that world-class AI research requires billions of dollars in training runs. Within months, deepseek became shorthand for resource-efficient model development.
The company has continued to release models at a rapid cadence. Its V4 model, introduced in April, further cemented its reputation for cost efficiency. DeepSeek said at the time of the V4 release that prices would “drop significantly” in the second half of this year when Huawei’s Ascend 950PR supernodes are expected to ship at scale. Those supernodes are a key component of China’s effort to build a self-reliant AI supply chain.
The price decline in inference and, in some cases, training costs is expected to ripple across the AI ecosystem. Lower prices make DeepSeek’s models accessible to smaller developers, academic researchers, and enterprises that previously regarded advanced AI as too expensive. This could accelerate the adoption of AI across China’s enormous industrial base, from automated customer service to precision agriculture and scientific discovery.
DeepSeek's technology also has geopolitical significance. Western governments have raised concerns about open-source or open-weight models that can be used by adversaries to generate disinformation or develop cyber tools. Yet the response from Chinese industry has been overwhelmingly enthusiastic. For Beijing, DeepSeek represents a proof-point that Chinese companies are not simply following in the footsteps of OpenAI or Anthropic but are actually leading in areas such as training efficiency, sample efficiency, and algorithmic innovation.
China’s chip industry rallies around DeepSeek
The new funding round should also accelerate DeepSeek’s collaboration with Chinese chipmakers. Reports indicate that Huawei, Cambricon, and MetaX are all optimising their AI processors for DeepSeek’s V4 model. This is a notable reversal of the usual dynamic, where model developers tailor their software to existing chips. Here, chipmakers are actively adapting their hardware to maximise performance for a specific model architecture.
The reason is straightforward: DeepSeek has become a key piece of China’s national AI infrastructure. By optimising for V4, domestic chipmakers can ensure that their products run the most popular and efficient models on the market. This creates a virtuous flywheel. DeepSeek benefits from better performance and lower cost per token on Chinese hardware. Chipmakers benefit from having a flagship workload that showcases their capabilities to the rest of the industry.
Huawei’s Ascend series is particularly crucial. The Ascend 950PR supernode, expected to ship at scale in late 2025, could provide a serious alternative to Nvidia’s high-end offerings that remain largely unavailable to Chinese buyers due to export controls. If DeepSeek can demonstrate that its models run efficiently on Huawei’s hardware, that would be a massive validation of China’s ability to innovate in artificial intelligence even without cutting-edge Western fabrication.
But there are significant hurdles. Software toolchains for domestic or Chinese chips are still less mature than Nvidia’s CUDA ecosystem. Developers face a steep learning curve when migrating to Huawei or Cambricon processors. DeepSeek’s willingness to optimise for those chips could lower the barrier for others, but the process is not trivial. Moreover, domestic chipmakers continue to grapple with lower yields and higher costs compared to their US counterparts, at least for the most advanced nodes.
Still, the combined effort of DeepSeek and Chinese chipmakers may be accelerating China’s path toward semiconductor self-sufficiency in the AI domain. The launch of V4 at a time of tightened US export controls was itself a bold statement. DeepSeek achieved its performance using an ingenious mix of hardware resources, some of which were likely older-generation Nvidia chips that were stockpiled before the ban. But the next wave of innovation must be built on Chinese-made silicon if the country’s AI industry is to scale without relying on smuggled or illegally acquired processors.
The impending decline in token prices, paired with the rise of domestic hardware optimised specifically for DeepSeek, could have far-reaching implications for the global AI market. Chinese developers are already building new products on DeepSeek’s open-weight models, such as virtual assistants, coding tools, and scientific software. As those products mature and expand, the demand for DeepSeek’s API and its open-source releases will only grow. The new funding round gives the company the resources to nurture this ecosystem.
For US-based AI giants, the rise of DeepSeek is a call to action. Many have responded by accelerating their own efficiency improvements, yet they still face the challenge of far higher cost structure. The possibility that a Chinese startup could continue to match or beat Western models at a fraction of the cost intensifies the pressure on American laboratories to justify their enormous valuations and capital expenditures. The next few years will reveal whether DeepSeek’s model of frugal innovation can scale beyond the research phase into a sustainable, broadly adopted platform.
In the meantime, DeepSeek’s first external funding round represents a rare moment of convergence: a world-leading research startup, a deep-pocketed founder, an ambitious array of corporate and state investors, and a geopolitical race that has placed artificial intelligence at the center of national industrial policy. The final terms are still being settled, but one thing is certain: DeepSeek is no longer just a fascinating experiment. It has become a commercially vital, strategically essential enterprise with a global reach and a $60 billion price tag.
Source: Silicon UK News