DeepSeek, the Chinese artificial intelligence startup that has rapidly become a key player in the global AI race, is poised to achieve a valuation of $45 billion (£33 billion) in its first-ever round of external funding. According to a report by the Financial Times citing unnamed sources familiar with the matter, the China Integrated Circuit Industry Investment Fund, commonly known as the “Big Fund,” is in advanced talks to lead the round. This significant state-backed investment vehicle, which has previously supported major semiconductor companies like SMIC and YMTC, is now turning its attention to the software side of AI infrastructure.

The valuation growth has been nothing short of meteoric. When discussions first began a few weeks ago, DeepSeek initially sought a valuation of at least $10 billion. By late April, that figure had reportedly climbed to $20 billion, and now it has more than doubled to $45 billion. This dramatic increase reflects both the intense investor demand for frontier AI models and the strategic importance of DeepSeek’s work in the context of China’s broader technological ambitions. Other Chinese tech giants, including Tencent Holdings and Alibaba Group, are also reportedly in talks to participate in the round, further underscoring the significance of this funding event.

Unlike many of its Chinese competitors, such as Moonshot and Hong Kong-listed Zhipu AI, which are actively pursuing broad commercialization and product integration, DeepSeek has maintained a sharp focus on developing advanced frontier models. Zhipu AI, for instance, is currently valued at around $52 billion and has a wide array of product offerings. DeepSeek’s strategy appears to prioritize pushing the boundaries of model capability over immediate market penetration. This approach has not deterred investors, who are betting on the long-term potential of DeepSeek’s innovations to shape the future of AI, both in China and globally.

The involvement of the Big Fund is particularly noteworthy. The fund, formally established in 2014 as part of China’s ambitious semiconductor self-sufficiency plan, has typically invested in hardware giants like contract chipmaker SMIC and memory manufacturer YMTC. Its move to invest in an AI software company signals a strategic shift: the Chinese government is now urging domestic tech players to collaborate and create an integrated ecosystem around AI that can compete with US offerings. DeepSeek, with its emphasis on model efficiency and open-source contributions, fits neatly into this vision.

DeepSeek’s Technological Breakthroughs

DeepSeek first gained international attention in late 2023 with the release of its DeepSeek-V2 model, which demonstrated competitive performance against leading open-source models from Meta and Mistral. The startup, founded in 2023 by Liang Wenfeng, a former Chinese hedge fund executive, quickly established a reputation for innovation in model architecture. Its Mixture-of-Experts (MoE) design allowed it to achieve high performance while using significantly fewer computational resources than traditional dense models. This efficiency became a key selling point, especially as global supply chains faced constraints on advanced GPUs due to US export controls.

In January 2025, DeepSeek made headlines with the release of DeepSeek-R1, a reasoning model that rivaled OpenAI’s o1 in certain benchmark tasks. The model’s ability to perform chain-of-thought reasoning and its transparent inference process garnered praise from the AI research community. Suddenly, DeepSeek was no longer just a Chinese curiosity but a serious contender in the global frontier model race.

More recently, the launch of DeepSeek-V4 has further solidified the company’s position. The company stated that V4 was optimized to run inference on Huawei’s Ascend 950PR chips. This is a significant development for China’s tech independence efforts. Huawei has overtaken Nvidia in market share within China, as Nvidia’s most advanced chips, such as the H100 and B100, remain largely unavailable in the country due to US export restrictions. The optimization for Huawei’s Ascend platform means that DeepSeek can offer high-performance AI inference without relying on Nvidia hardware, a crucial step for Chinese organizations that seek to reduce their dependence on foreign technology.

The Role of Government and Big Tech

The Chinese government has been actively fostering a domestic AI ecosystem that can stand toe-to-toe with the US. This includes not only hardware independence but also software collaboration. The Big Fund’s potential investment in DeepSeek is part of a larger pattern of state-directed capital flowing into strategic sectors. In addition to the Big Fund, China has established multiple national AI platforms and has provided billions of dollars in subsidies and grants to AI startups.

Tencent and Alibaba’s interest in DeepSeek is also strategic. Both companies have their own large language models—Tencent’s Hunyuan and Alibaba’s Qwen series—but they recognize that DeepSeek’s frontier models could complement their offerings or serve as the foundation for applications across their ecosystems. Alibaba, in particular, has been investing heavily in AI infrastructure and cloud services, and integrating DeepSeek’s models could enhance its competitive position against Baidu’s ERNIE and ByteDance’s models.

The influx of capital from state and corporate sources will likely accelerate DeepSeek’s research and development. The company has maintained a relatively lean operation, with a team of around 200 researchers and engineers. With a $45 billion valuation, DeepSeek will have the resources to scale up its workforce, acquire more computing power, and expand its model training efforts. However, the increased attention also brings scrutiny, particularly regarding the potential for the technology to be used in ways that align with state objectives, such as surveillance or censorship.

Market Context and Competition

DeepSeek’s valuation surge comes at a time when the global AI market is experiencing a correction after the initial euphoria of 2023. Many startups have struggled to justify high valuations without clear revenue models. DeepSeek, by contrast, has not only gained significant traction but has also demonstrated that frontier model development can attract substantial investment even in a more cautious funding environment.

Competitors in China include Zhipu AI, which has a valuation of around $52 billion and a broader product portfolio that includes enterprise-grade solutions. Baidu, with its ERNIE 4.0 model, commands a huge user base but has faced criticism for its conservative approach. Moonshot AI, another high-potential startup, has focused on long-context models and consumer-facing applications. DeepSeek’s niche remains cutting-edge research and open-source contributions, which have earned it a loyal following among developers and researchers.

Internationally, DeepSeek competes with the likes of Anthropic, Cohere, and Mistral AI, as well as the major tech players like Google, Meta, and Microsoft-backed OpenAI. The landscape is fiercely competitive, but DeepSeek’s efficient model architecture and low-cost inference give it an edge, particularly for customers who are price-sensitive or concerned about vendor lock-in.

Implications for China’s Tech Ecosystem

The Big Fund’s involvement in DeepSeek signals a broader policy shift: China is no longer content to be a follower in AI innovation. The government is actively working to create a vertically integrated ecosystem that spans from chip design (with companies like Huawei and SMIC) to model development (DeepSeek, Zhipu) to application layers (Tencent, Alibaba). This cohesive strategy aims to reduce vulnerabilities caused by US export controls and to position China as a leader in the next generation of AI technology.

DeepSeek’s optimization for Huawei’s Ascend chips is a key milestone in this effort. While Nvidia still dominates the global AI chip market despite the US bans, Huawei has been rapidly improving its Ascend series. The 950PR chip, specifically designed for inference tasks, offers competitive performance. By ensuring that its models run efficiently on Chinese hardware, DeepSeek helps create a self-reliant supply chain that can operate independently of US sanctions.

Furthermore, DeepSeek’s open-source approach has incentivized a broader community of developers in China to build applications on its models. This accelerates adoption and creates a virtuous cycle where feedback from real-world use improves the model’s capabilities. The Chinese government has also encouraged open-source AI development as a way to quickly disseminate technology across industries, from manufacturing to healthcare to finance.

Yet, there are challenges ahead. The US government has already indicated it will tighten export controls on AI software and model weights, making it harder for Chinese companies to access the most advanced technologies. DeepSeek’s reliance on Huawei chips may also face limitations if Huawei itself encounters production bottlenecks or further US sanctions. However, the investment from the Big Fund provides a financial buffer that can help DeepSeek navigate these turbulent waters.

In summary, DeepSeek’s first funding round is not just a financial milestone but a strategic pivot for China’s AI ambitions. The startup’s ability to attract top-tier investors, develop world-class models, and align with national goals has set it on a trajectory that could reshape the global AI landscape. As the funding round closes, all eyes will be on DeepSeek’s next moves—whether it will continue to focus on research, expand into commercial products, or become a key supplier for the Chinese government’s AI initiatives. The race is far from over, but DeepSeek has firmly established itself as a contender to watch.

Source: Silicon UK News