Chinese AI Labs Pursue Custom Chips to Lower Costs

London, United Kingdom Jul 12, 2026 ALN: Chinese AI labs are increasingly pursuing proprietary chips, mirroring a global trend towards software-hardware integration, but industry insiders and analysts warn that the strategy carries risk due to the massive upfront investments required. This shift towards custom chip development is not only a response to the growing demands of artificial intelligence (AI) applications but also reflects a broader ambition within China to gain technological independence and leadership in the AI sector.

The core motivation for choosing in-house chips lies in pursuing greater hardware-software synergy and lowering long-term operating costs. As AI applications become more complex and resource-intensive, the need for specialized hardware that can efficiently handle these demands has become paramount. Proprietary chips can be optimized for specific tasks, potentially leading to faster processing times and reduced energy consumption. Arisa Liu, chief director and research fellow at Taiwan Industry Economics Services, emphasizes that this strategic move aims to enhance operational efficiency and reduce reliance on foreign chip manufacturers.

In recent years, the global semiconductor industry has faced significant disruptions, primarily due to geopolitical tensions and supply chain vulnerabilities exposed by the COVID-19 pandemic. These challenges have prompted many countries, including China, to invest heavily in domestic chip production capabilities. China's government has recognized the importance of semiconductor technology as a critical component of national security and economic competitiveness. As a result, there has been an influx of funding and resources directed towards the development of homegrown chip technologies.

Paul Triolo, a partner and technology policy lead at DGA-Albright Stonebridge Group, noted in his personal newsletter AIStackDecrypted that the proprietary efforts underscore how China’s leading model developers increasingly view silicon as a strategic extension of the model stack rather than simply another infrastructure input. This perspective indicates a significant shift in how AI companies conceptualize their technology stacks, integrating hardware development into their core strategy rather than treating it as a peripheral concern.

DeepSeek, a Hangzhou-based start-up, exemplifies this trend as it has been quietly hiring chip-design talent without posting public job openings, according to two individuals familiar with the situation who declined to be named due to the private nature of the matter. The company’s plans for a customized AI inference chip began roughly a year ago and remain at an early stage. This indicates a growing recognition among AI firms that control over hardware can provide a competitive edge in a rapidly evolving market.

Meanwhile, Zhipu AI, the Beijing-based developer of the high-performance GLM-5.2 model, is reportedly in early talks with domestic chip-design companies about tailored AI processors. The discussions come amid a sharp increase in its daily token usage, highlighting the urgent need for enhanced processing capabilities to support its growing user base and application demands. The move towards custom chips could allow Zhipu AI to better manage its computational resources, optimize performance, and reduce costs associated with third-party hardware solutions.

Despite the potential benefits of developing proprietary chips, industry analysts caution that the strategy carries significant risks. The upfront investments required for research and development, manufacturing, and talent acquisition can be substantial. Additionally, the semiconductor industry is characterized by rapid technological advancements and fierce competition, making it challenging for new entrants to keep pace. Companies must not only invest in the design and production of chips but also in the ongoing research necessary to innovate and improve their offerings continuously.

Moreover, the complexity of chip design and manufacturing means that companies must navigate a steep learning curve. Developing a competitive chip requires expertise in various fields, including materials science, electrical engineering, and computer science. This complexity can lead to delays and increased costs, particularly for companies that are new to the semiconductor space.

Furthermore, the geopolitical landscape adds another layer of uncertainty. As tensions between China and other countries, particularly the United States, continue to rise, there may be implications for the availability of essential technologies and materials. Export restrictions and trade policies can significantly impact the ability of Chinese companies to access the necessary components and technologies to develop their chips.

The push for proprietary chips also aligns with China's broader strategy to enhance its technological capabilities in various sectors, including telecommunications, computing, and AI. The Chinese government has been vocal about its desire to achieve self-sufficiency in technology, reducing reliance on foreign suppliers, particularly in critical areas like semiconductors. This strategic vision has been articulated in national policies such as the "Made in China 2025" initiative, which aims to elevate the country’s manufacturing capabilities and foster innovation across high-tech industries.

As the global race for AI supremacy intensifies, the development of custom chips is seen as a crucial component for maintaining competitive advantage. AI applications, such as machine learning, natural language processing, and computer vision, require immense computational power, which can be more efficiently delivered through specialized hardware. Hence, companies that can effectively integrate chip design with AI algorithms may find themselves at the forefront of technological advancements.

However, the road to developing proprietary chips is fraught with challenges. The semiconductor industry is notoriously capital-intensive, with significant investments required not just for the design phase but also for fabrication, testing, and marketing. For instance, the cost of setting up a semiconductor fabrication plant (fab) can run into billions of dollars. This financial burden can be particularly daunting for start-ups or smaller firms that may not have the same financial backing as established players in the industry.

In addition to the financial implications, there is also the issue of talent acquisition. The semiconductor sector requires a highly skilled workforce, including engineers and researchers with expertise in chip design, fabrication processes, and software integration. The demand for such talent is high globally, leading to fierce competition among companies to attract and retain skilled professionals. This talent shortage can hinder the efforts of companies trying to establish their own chip manufacturing capabilities.

Moreover, as AI applications continue to evolve, the requirements for chips will also change, necessitating ongoing research and development efforts. Companies must remain agile and responsive to these changes, which can be a significant challenge in an industry characterized by rapid technological progress. The ability to pivot and adapt to new advancements in AI and chip technology will be critical for the success of proprietary chip initiatives.

In conclusion, the pursuit of proprietary chips by Chinese AI labs reflects a broader trend of integration between hardware and software in the tech industry. While the potential benefits of custom chip development are significant, including improved efficiency and reduced long-term costs, the associated risks cannot be overlooked. As these companies navigate the complexities of chip design and manufacturing, they will need to balance their ambitions with the realities of the semiconductor market and the geopolitical landscape. The outcome of these efforts could have far-reaching implications for the future of AI development in China and the global tech industry as a whole. The strategic decisions made by these AI labs will not only shape their own futures but could also influence the trajectory of technological innovation and competition on a global scale.