DeepSeek, the Chinese artificial intelligence startup known for its frontier research and open-weight models, is reportedly heading toward its first external funding round at a valuation of $45 billion (about £33 billion). According to unnamed sources close to the matter, the China Integrated Circuit Industry Investment Fund, widely referred to as the "Big Fund," is in talks to lead the investment round. Earlier reports also indicated that internet giants Tencent Holdings and Alibaba Group are in discussions to participate, marking a major step for a company that has largely operated without outside capital since its founding.
The reported valuation has grown dramatically in a short period. When DeepSeek first broached the idea of raising external funds a few weeks ago, it was seeking a minimum valuation of $10 billion. By late April, that figure had reportedly risen to $20 billion. Now, with state-backed investors and major private technology companies lining up, the valuation appears to have more than doubled again to $45 billion. The rapid escalation underscores the intense investor appetite for high-caliber AI research labs in China, particularly those that have demonstrated technical leadership comparable to leading U.S. models.
State funding and strategic backing
The Big Fund, formally known as the National Integrated Circuit Industry Investment Fund, is China's largest state-backed semiconductor investment vehicle. Established to bolster the country's semiconductor supply chain, the fund has historically backed major chip manufacturers such as Semiconductor Manufacturing International Corporation (SMIC) and memory chip producer Yangtze Memory Technologies Co. (YMTC). Its involvement in an AI software startup marks a notable expansion of its focus, reflecting the Chinese government's view that AI and semiconductors are deeply intertwined.
Beijing has been increasingly vocal about the need for domestic technology companies to collaborate and build an integrated AI ecosystem that can compete with products from the United States. The government has also encouraged state funds to invest in strategic sectors, particularly those where China has historically lagged, such as advanced chips and foundation models. DeepSeek, with its highly efficient AI models and close ties to High-Flyer, a quantitative hedge fund that seeded the project, is viewed as a crown jewel in China's AI research community.
Tencent and Alibaba's potential participation adds a strategic dimension beyond state backing. Both companies have heavily invested in cloud computing and enterprise AI services. For Tencent, DeepSeek offers an opportunity to integrate advanced models across its social media, gaming, and enterprise software ecosystem. Alibaba, which has developed its own Qwen family of models, may be looking to deepen its portfolio of AI offerings and maintain its position as a leading cloud provider in China. The participation of these commercial giants would give DeepSeek access to vast distribution channels and real-world deployment scenarios that pure research labs often lack.
Competitive landscape in Chinese AI
Investors are reportedly betting on DeepSeek's potential despite its primary focus on developing frontier models rather than broad commercialization. In this respect, DeepSeek differs from some of its Chinese rivals, such as Moonshot AI and the Hong Kong-listed Zhipu AI. Zhipu is currently valued at roughly $52 billion, making it one of the most valuable AI startups in the country. Moonshot, known for its Kimi assistant, has also raised substantial capital from leading Chinese investors.
The Chinese AI sector has become crowded in recent years, with companies vying to produce models that can match or surpass American counterparts like OpenAI's GPT series, Anthropic's Claude, and Google's Gemini. DeepSeek has carved out a niche by emphasizing efficiency and open-source releases. Its V3 model, released in late 2024, demonstrated that high-performance results could be achieved with lower training costs than many competitors. The subsequent R1 reasoning model drew worldwide attention for its ability to perform complex logical and mathematical tasks, challenging assumptions about the necessity of enormous computational budgets.
More recently, DeepSeek launched its V4 model, which reportedly includes optimizations for running inference on Huawei's Ascend 950PR chips. This is seen as a significant boost for China's technological self-reliance efforts, as it signals DeepSeek's willingness to support domestic hardware platforms despite the longstanding dominance of Nvidia GPUs in AI training and inference. The decision to optimize for Huawei's chips is both a technical and political statement, aligning with Beijing's push for a less restricted supply chain in the face of U.S. export controls.
Huawei optimization and China's tech independence
Huawei has overtaken Nvidia in market share in China, a dramatic shift driven by U.S. restrictions that prevent Nvidia from selling its most advanced chips to Chinese customers. Nvidia's top-tier products, such as the H100, H200, and Blackwell architecture chips, are largely unavailable in China without special licenses, which the U.S. government has been reluctant to grant. While Nvidia has developed lower-performance variants specifically for the Chinese market, such as the H800 and the more recent H20, customers have nonetheless shown increasing interest in domestic alternatives.
Huawei's Ascend lineup, developed by its HiSilicon subsidiary, has emerged as the leading domestic option. The Ascend 950PR is designed to handle large-scale AI workloads, and DeepSeek's optimization of V4 to run inference on this chip is a strong indicator of the ecosystem's maturation. Inference, the process of running a trained model on new data, is particularly important for real-world applications like chatbots, recommendation systems, and autonomous agents. By optimizing for Ascend, DeepSeek is effectively validating Huawei's hardware as a viable competitor to Nvidia for at least certain types of workloads.
However, industry analysts note that training still remains a significant challenge on domestic hardware. While inference requirements are relatively more flexible, training frontier models requires massive clusters with high-speed interconnects and deeply optimized software stacks. Huawei has been working on its MindSpore framework and other tools to close the gap, but the company's chip supply remains constrained by its own production capacities, as advanced lithography machines are restricted under U.S. export rules. Nonetheless, the DeepSeek-Huawei optimization is a symbolic win for China's tech independence narrative.
Broader implications for the global AI race
The news of DeepSeek's colossal valuation comes at a time when the global AI industry is undergoing a paradigm shift. After a period of relentless hardware investment, investors are increasingly questioning the return on the massive capital expenditures required to train frontier models. DeepSeek's reported ability to achieve strong performance with relatively limited training budgets has already pressured some U.S. AI companies and led to debates about whether the high-cost, resource-intensive model of frontier AI research is sustainable.
China's approach to AI development has historically relied on a combination of state support, catch-up research, and a vast industrial base. The deepening collaboration between state funds, technology corporations, and AI research labs is likely to accelerate in the coming months. DeepSeek's funding round is not just about money; it is a strategic alignment aimed at creating a self-sufficient AI ecosystem that can survive and thrive in a fragmented global market.
From a technical perspective, DeepSeek has already made waves with its R1 reasoning model, which many observers saw as a significant advance in neuro-symbolic reasoning and chain-of-thought capabilities. The V4 model's release further demonstrated the company's commitment to pushing the boundaries of what is possible with open-source AI. By making models available to the broader research community, DeepSeek has also contributed to the global democratization of AI, even as Western governments debate the risks of open-weight models.
The potential $45 billion valuation places DeepSeek in the upper echelons of global AI startups. For comparison, Anthropic, one of OpenAI's main competitors, was valued at around $60 billion in early 2025, while OpenAI itself has seen valuations as high as $300 billion in private transactions. DeepSeek's valuation is particularly remarkable given that the company has not yet generated significant revenue and has not yet filed for an initial public offering. The high valuation suggests that investors are willing to look beyond immediate monetization and instead bet on the long-term strategic value of proprietary AI capability.
The involvement of the Big Fund also signals a departure from the previous pattern where state-backed funds focused exclusively on hardware. As AI models become more central to national competitiveness, software is increasingly seen as a critical piece of the semiconductor ecosystem. The Big Fund's investment could pave the way for other state-backed entities to support AI model providers, potentially leading to a more coordinated national strategy in artificial intelligence.
Nevertheless, the funding round has yet to be finalized. Negotiations could still fall through due to valuation disagreements, regulatory scrutiny, or other obstacles. In previous instances, Chinese startups have seen funding rounds collapse when government agencies stepped in to review strategic investments. Given DeepSeek's prominence and the potential involvement of foreign-linked entities like Tencent and Alibaba (both listed in Hong Kong and subject to international regulations), the deal may attract close antitrust and national security reviews, both in China and abroad.
For now, the international AI community remains closely watchful. DeepSeek's success would not only reshape China's internal AI landscape but also influence the global balance of power in artificial intelligence. With computing resources becoming increasingly contested and export controls continuing to tighten, the convergence of state capital and cutting-edge research in China could redefine how AI innovation is funded and deployed.
As part of the broader push, the Chinese government has been urging domestic tech players to avoid wasteful competition and instead focus on building integrated platforms that can rival U.S. offerings. This collaborative spirit is evident in the reported participation of both Tencent and Alibaba in the same round, which was rare in the past. Previous funding rounds for Chinese AI startups typically featured one primary strategic investor, but the involvement of multiple giants suggests a higher level of coordination, potentially orchestrated by government policy.
The talk of a $45 billion valuation also raises questions about sustainability and market rationality. Some analysts worry that Chinese AI valuations have entered bubble territory, especially when only a few companies have demonstrated meaningful revenue from AI products. However, proponents argue that the strategic value of owning foundational AI technology is worth the premium, particularly in a geopolitical climate where access to leading models cannot be taken for granted. The Chinese domestic market, with a population of over 1.4 billion and strong demand for digital services, offers a sufficient testing ground for new AI applications.
DeepSeek's future will likely be shaped by how it balances its research ideals with the practical demands of commercial investors. The company initially grew out of a quant hedge fund and was not intended to become a conventional business. The decision to raise outside capital marks a turning point, one that may force DeepSeek to embrace product development and customer delivery at a pace it has previously avoided. At the same time, the resources provided by a $45 billion valuation could enable the company to attract world-class talent and build the compute infrastructure necessary for next-generation models.
The next steps in the funding process are expected to be announced in the coming weeks. If the round concludes as reported, it would be one of the largest AI investments ever made in China and a clear signal to the global market that China intends to remain a frontline competitor in artificial intelligence.
Source: Silicon UK News