This pair of entrepreneurs from Hubei, born in the 1960s, have quietly become essential partners for major AI players. When discussing artificial intelligence, most people think of advanced components like NVIDIA's graphics cards, rarely noticing the fundamental printed circuit boards that enable these systems to function. For electronic devices, PCBs serve as both the skeleton and the highway network.
A Chinese company specializing in building these skeletons for AI servers has gained significant attention. The manufacturer is progressing toward a Hong Kong listing following a remarkable turnaround. Guanghe Technology's story represents a classic reversal of fortune rather than a typical tech startup narrative. A factory that was nearly shuttered in late 2012, on the brink of collapse, was acquired by a visionary couple. Over a decade, they expertly capitalized on the AI computing boom, transforming it into a leading global supplier of PCBs for AI servers.
The turnaround began in 2013 when Xiao Hongxing and his wife, Liu Jinchan, took a substantial risk. They invested over $25 million of their own capital to acquire a majority stake in the struggling facility, which had Taiwanese corporate origins. Both graduates of South China University of Technology with backgrounds in chemistry and chemical engineering, the couple had previously established a successful chemical materials business in Dongguan during the 1990s, accumulating both capital and industry expertise.
Their decision to acquire the failing PCB factory was met with skepticism. However, Xiao Hongxing recognized the intrinsic value: existing production qualifications, basic equipment, and, crucially, the growing importance of circuit boards in an increasingly data-driven world. He believed the server market held substantial potential.
After the acquisition, the couple prioritized two critical steps. First, they assembled a core technical team, recruiting experienced professionals and alumni from their university to instill a strong technological foundation. The goal was clear: shift from low-end contract manufacturing to focus on high-end, technologically complex products.
Second, they made a strategic pivot. While the consumer electronics market for PCBs was crowded and offered thin margins, Xiao Hongxing insisted on concentrating all resources exclusively on server-grade PCBs. This decision required significant courage, as server PCB technology is challenging, client requirements are stringent, and certification cycles are long, often taking one to two years to generate revenue. However, these high barriers to entry also meant less competition and stronger client loyalty with better profitability once established.
This strategic focus proved to be a turning point. The company invested heavily in updating equipment, mastering complex processes, and rebuilding client trust. By 2015, Guanghe Technology had achieved profitability. As cloud computing and big data gained momentum around 2020, the company's early specialization made it a highly sought-after supplier. When ChatGPT ignited the global AI frenzy in 2022, Guanghe was already an established leader in its niche.
The financial results underscore the success of this strategy. Annual revenue grew from 1.3 billion yuan in 2019 to 3.73 billion yuan in 2024. More impressively, net profit surged from 280 million yuan in 2022 to 676 million yuan in 2024, demonstrating significantly enhanced profitability.
The company's success is underpinned by the evolving demands of AI technology. Traditional server PCBs, with 8-16 layers, were sufficient for standard data processing. AI servers, however, require PCBs that function like sophisticated, multi-layered transportation hubs—often exceeding 20 layers—to handle immense data volumes at high speeds while managing significant heat generation. This necessitates advanced materials with minimal signal loss and extremely precise manufacturing tolerances.
Consequently, the value and margins for these high-end PCBs have skyrocketed. While a standard server's PCB might cost around 3,400 yuan, a PCB for an AI training server can be two to three times more expensive, with some top-tier AI server PCBs valued in the tens of thousands of US dollars.
The market opportunity is substantial. The global AI server PCB market was valued at approximately $3.2 billion in 2024 and is projected to reach $7 billion by 2029. Chinese manufacturers dominate this high-end segment, with Guanghe Technology ranking among the top three globally. It supplies eight of the world's top ten server brands, meaning its foundational technology supports a vast majority of cloud services.
Guanghe's competitive edge is built on several factors. The stringent, lengthy certification process for server components creates high entry barriers but also fosters long-term, stable client relationships once approved. The company's gross margin exceeded 33% in 2024, with margins for core AI server PCBs above 36%. Its net profit margin surpassed 18% in the first three quarters of 2025, exceptional for a manufacturing business. This profitability fuels further investment in research and development and advanced equipment.
Strategically, the company is expanding capacity intelligently, using funds from its listing to establish a new factory in Thailand and upgrade its facility in Guangzhou, focusing on high-layer-count, complex, and high-margin products.
The story of Guanghe Technology and the couple behind it epitomizes the journey of many dedicated Chinese manufacturers. It is not about disruptive business model innovation but about decades of deep expertise, focusing on perfecting a core product until it aligns perfectly with a major technological wave. While attention often focuses on flashy AI applications, the critical role of foundational infrastructure providers like Guanghe—the companies that supply the essential tools and pathways—is increasingly recognized as indispensable. From Hubei, Xiao Hongxing and Liu Jinchan seized the opportunity presented by the AI era, demonstrating that every groundbreaking achievement in technology relies on the most fundamental and reliable components created by such dedicated builders.
Comments