Woofun AI reports that a semiconductor company founded in Xianyang, Shaanxi, has evolved from a cash-strapped startup with fewer than 20 employees into a critical supplier for NVIDIA's most expensive AI servers, achieving a gross profit margin of 77.81% in the first quarter of 2026. This entity, which once operated out of a facility so modest it utilized an old Otto to transport international experts, now commands a market capitalization of 220 billion yuan, representing a thirtyfold increase from its initial public offering valuation of 7 billion yuan five years prior. The trajectory of this firm, driven by Mr. Zhang, illustrates a stark contrast between the early skepticism of global partners and the current dominance in high-speed optical chip manufacturing.

The narrative begins in 2010 when Mr. Zhang, a 40-year-old director of research and development with a doctorate from the University of Southern California and a bachelor's degree from Tsinghua University, attempted to launch a chip company in Beijing. Trusting a familiar friend as general manager while he remained in the United States, Mr. Zhang suffered a catastrophic betrayal when the manager secretly transferred more than 7 million yuan of company funds to settle personal debts. By the time the embezzlement was discovered, the capital was entirely spent, resulting in the bankruptcy of the venture and the loss of two years of development. This event established a foundational principle for Mr. Zhang's subsequent entrepreneurial strategy: absolute control over financial and operational lifelines is non-negotiable.

Undeterred by this failure, Mr. Zhang returned to his hometown of Xianyang at the end of 2012 with two partners to capitalize on local policy support. In 2013, the new company was officially registered in the Xixian New Area, focusing on DFB laser chip technology, a niche field virtually non-existent in China at the time. These chips function as critical components in optical fiber networks, converting electrical signals into light signals capable of switching billions of times per second. At that juncture, 100% of such high-end chips were imported from the United States and Japan, creating a complete dependency on foreign supply chains. Mr. Zhang rejected low-cost contract manufacturing in favor of a resource-intensive vertical integration model, covering chip design, wafer fabrication, manufacturing, and packaging.

Financial constraints forced the team to rely on second-hand equipment from closed-down factories overseas, often requiring Mr. Zhang to lead the team in modifying machines whose parameters did not match their specific needs. The old Otto used to pick up a Japanese semiconductor materials supplier engineer in the summer of 2013 served as a tangible symbol of these difficult beginnings, prompting the visitor to hesitate before entering the vehicle due to suspicions of deception. Survival during this equipment adjustment phase was only secured when a Zhongguancun fund in Beijing invested capital based on Mr. Zhang's technical background. In early 2014, Mr. Zhang resigned from his senior executive role at a foreign company to devote himself full-time to operations in Xianyang, marking the transition from a part-time endeavor to a dedicated industrial pursuit.

Prior to 2018, the domestic market for mid- and low-speed optical chips, specifically 2.5G and 10G variants, was dominated by Japanese giants such as Sumitomo and Mitsubishi, alongside the American firm Lumentum. These overseas monopolies reaped substantial profits until Mr. Zhang's team achieved technological parity and began mass production. In response, the established giants launched an aggressive price war, utilizing their financial advantages to cut prices of similar products by half, sometimes approaching cost levels, in an attempt to crush the fledgling Chinese competitor. Despite immense financial pressure on the Xianyang factory, Mr. Zhang's vertical integration strategy proved decisive; by controlling the entire production chain, the team optimized wafer structures and significantly improved yield rates, achieving a lower cost per chip than their overseas rivals.

By 2021, the dynamics of the price war shifted decisively in favor of the Xianyang company. The American and Japanese giants realized that their high labor and operating costs had rendered their mid- and low-speed optical chip businesses unprofitable, forcing them to reduce production capacity and withdraw from the Chinese market. This retreat handed over more than half of the market share to the Chinese firm, which successfully went public in December 2022.

However, the period between 2023 and 2024 saw a downturn in the communications industry, with major customers cutting spending. In 2024, despite increased revenue, the company reported a net profit loss of 6.13 million yuan, leading market observers to question whether the firm was merely a pseudo-high-tech entity built on a catchy concept.

The turning point arrived with the explosion of large-scale artificial intelligence models in 2025, which created an exponential demand for fast data transmission speeds. Mr. Zhang's engineers possessed two powerful assets that became critical during this shift: high-power CW laser chips with output levels of 70 mW and 100 mW. These specifications were the key to entering NVIDIA's supply chain, distinguishing the firm from competitors.

Woofun AI data shows that in NVIDIA's latest GB200 supercomputing rack, the high-power CW laser chips from Xianyang provide the necessary power for data transmission, working in tandem with AWG chips from a rival firm in Hebi, Henan, which manage vast data traffic. These two companies, though located in different third- and fourth-tier cities, are not competitors but indispensable components within the same server architecture.

In the first quarter of 2026, the company's financial results reflected the massive impact of this strategic pivot. Quarterly revenue reached 355 million yuan, representing a threefold increase year-on-year, while quarterly net profit surged to 179 million yuan, an almost twelvefold increase compared to the previous year. The gross profit margin climbed to a staggering 77.81%, a figure that underscores the scarcity of global manufacturers capable of producing such chips and grants Mr. Zhang significant bargaining power against American and Japanese giants. This performance contrasts sharply with the 6.13 million yuan loss recorded in 2024, validating the long-term investment in high-power chip technology.

Capitalizing on this momentum, Mr. Zhang announced a major investment plan in March 2026 to expand a new production base in the Xixian New Area. The plan allocates 1.251 billion yuan to develop the next generation of 1.6T superfast optical chips, an amount equivalent to twice the company's annual revenue in 2025. Almost simultaneously, the leading optical chip company in Hebi, Henan, announced a parallel expansion plan of 1.265 billion yuan. Both firms, originating from traditional industrial or resource-based cities that have experienced population loss, are investing over 1.2 billion yuan each to close the gap in production capacity for global AI hardware competition.

The economic environment of these locations plays a crucial role in sustaining such hard-tech endeavors. Data from May 2026 indicates that the average price of second-hand houses in Qindu District, Xianyang, had dropped to 5,847 yuan per square meter, a nearly 40% decrease from previous highs. This decline in housing prices, which fell from 9,000 yuan to 5,000 yuan in the broader context of the region, allows young engineers and technicians who cannot afford homes in major cities to settle in Xianyang and Hebi. The extremely low cost of living provides a stable foundation for companies that require years of dedication to succeed, contrasting with the high overheads of metropolitan hubs.

Thirteen years after the Japanese engineer's skeptical arrival in an old Otto, the chips produced in the clean rooms of Xianyang are now installed in NVIDIA's servers, representing the pinnacle of human computing power and being shipped globally. The rise of China's hard-tech industry is not defined by fancy office buildings but by hardworking engineers in anti-static clothing overcoming challenges in affordable cities. This trajectory marks a definitive shift in the global semiconductor landscape, where third-tier cities are becoming the engines of AI infrastructure.