China’s Biren Technology, a domestic GPU startup founded by Wall Street veteran Zhang Wen, is making waves in the semiconductor world by assembling a “dream team” of international chip experts and developing high-performance GPUs that rival global leaders.
Its recent Hong Kong IPO, which valued the six-year-old company at HKD 90 billion, highlights the country’s push for technological self-reliance amid rising geopolitical and supply chain pressures.
Founded just over six years ago, Biren Technology has become a symbol of China’s AI ambitions, leveraging a mix of international talent and local capital to challenge global players such as Nvidia and AMD. The company’s IPO propelled its market capitalization to HKD 90 billion in a single leap, reflecting the fervor surrounding domestic AI chipmakers.
Although Biren promotes the dream of a “Chinese chip,” the company bears an unmistakably global signature in its management style and executive backgrounds.
Biren Technology was founded by Zhang Wen, who holds an MBA from Columbia University and a JD from Harvard Law School. After starting his career on Wall Street, Zhang returned to China and transitioned into the semiconductor and AI industries. Over the years, he has built an influential network of investors and talent, attracting experts from Nvidia, AMD, Oracle, Samsung, IBM, and international investment banks.
Zhang’s trajectory is notable for its breadth rather than technical expertise. Unlike other GPU founders with deep engineering backgrounds, Zhang is primarily a cross-sector integrator, connecting capital, technology, and talent. Sources indicate that he earned an undergraduate degree in Electrical Engineering from Hefei University of Technology and later pursued a PhD in Electronic Science and Technology at Shanghai Jiao Tong University’s School of Integrated Circuits.
After returning to China in 2011, Zhang co-founded InnoLight Photonics with SMIC founder Zhang Rujing, raising yields for domestically produced LED chips from 70% to over 95% and narrowing the gap with global technology leaders. He also served as chairman of private equity firm Dingyu Hengrui and as president of AI startup SenseTime, leading major city-level collaborations in Shanghai and Chengdu.
Zhang’s experiences convinced him that China was entering a historic wave of AI development. In September 2019, he registered Biren Technology, aiming to produce world-class GPUs while leveraging his international experience and networks.
Biren’s early focus was on talent recruitment. Zhang reached out to alumni from Harvard and Columbia to identify top GPU experts, assembling what industry insiders call a “Dream Team.” Among the early recruits was Hong Zhou, a former Nvidia engineer who contributed to Tesla architecture and CUDA development.
Other high-profile hires included Jiao Guofang, former Qualcomm Adreno GPU lead, and Xu Lingjie, a UC Berkeley graduate with experience at Nvidia, AMD, Samsung, and Alibaba Cloud. Despite resignations in recent years, Zhang continued to attract talent, adding former executives from AMD, Samsung, IBM, Oracle, Synopsys, and Nvidia.
“From architecture design to hardware, SoC, engineering, software ecosystems, financing, and team building, the team Zhang Wen has assembled at Biren Technology is the most comprehensive,” said Zhou Zhifeng, a managing partner at Qiming Venture Partners, Biren’s early investor.
The company’s ability to attract top talent has been instrumental in convincing private equity and venture capital firms to invest heavily, cementing Biren’s role in China’s AI chip history.
Biren Technology’s ambitions surpass those of other domestic GPU makers. While companies like Moore Threads aim to become the “Nvidia of China,” Biren has sought to develop the “most powerful domestic GPU,” with full-stack in-house R&D spanning architecture, hardware, software, and ecosystems.
In 2022, Biren unveiled its BR100 series, including BR104 and BR100 products, China’s first GPUs to adopt chiplet technology and TSMC’s 7nm process. The BR100 claimed a peak performance of up to 10 petaflops, putting it on par with Nvidia’s H100.
However, U.S. technology restrictions in 2023 cut off access to TSMC’s advanced manufacturing, forcing Biren to pivot toward self-reliance. Its BR106 chip achieved mass production in early 2023, and subsequent models, including the BR166, combined chiplet technology and high-bandwidth memory to double performance. The next-generation BR20X is slated for 2026, targeting large-model training comparable to the H100.
Biren has also developed a self-contained ecosystem. Its proprietary GPGPU architecture, BiliRen, is paired with the BIRENSUPA software platform, supporting domestic AI models such as Tongyi Qianwen and DeepSeek. At the cluster level, the BIRENCUBE system enables ultra-large-scale training with thousands of GPUs.
Full autonomy has made Biren attractive to state-owned enterprises. In 2024, the company deployed a 1,024-GPU cluster for China Telecom, operating for over 30 days without failure. Financial reports show revenue from intelligent computing solutions rising to 337 million yuan in 2024, with first-half 2025 revenue at 58.9 million yuan. Biren currently holds 24 active orders worth 822 million yuan, along with 5 framework agreements and 24 contracts totaling 1.241 billion yuan.
Biren’s IPO was measured compared to the more frenzied listings of Moore Threads or Muxi, but its solid order backlog sets it apart. Analysts say the company represents a more sustainable, execution-focused player among China’s four major GPU firms.
“Biren is building a foundation that balances technological ambition with market credibility,” said an industry analyst familiar with the company.
The company’s name, Biren—meaning “thousands of feet of cliff”—was inspired by Wuyi Mountain in Fujian province, a UNESCO World Heritage site. Zhang reportedly meditated beneath a cliff inscribed with the phrase “Steep cliffs rising thousands of feet,” a metaphor for the lofty ambitions of his company and China’s domestic GPU industry.
“The ambition behind the name speaks to Zhang Wen’s vision,” said the analyst. “For China to compete globally in AI and semiconductors, this kind of grand ambition is essential.”
However, analysts caution that such achievements remain contingent on attracting international-level talent. At present, Biren’s integration of top technical experts from around the world has made it a rare example of a domestic GPU company capable of competing on a global scale.
China’s drive to build domestic GPUs is part of a broader AI and semiconductor push, motivated by U.S. technology restrictions and the desire for self-reliance. The market’s response to Biren’s IPO underscores investor confidence in well-capitalized, talent-rich startups, even amid geopolitical tensions and supply chain uncertainties.
As the domestic GPU sector heats up, Biren Technology illustrates the blend of strategic vision, international networks, and technological ambition required to position China at the forefront of AI chip development.
Explore more exclusive insights at nextfin.ai.
