There is a long-held belief in the semiconductor industry that each generation of materials, devices, and applications corresponds to a specific stage of development. However, fourth-generation semiconductors remain an area where most market participants are hesitant to invest because advanced results are currently confined exclusively to university laboratories. Shanghai has accelerated this process in the Lingang area by planning the first systematic scheme for March 2024, the first cluster of the future industry by May 2025, and the plan for building the first cluster in July 2026.
Fourth-generation semiconductors include types with ultra-wide and ultra-narrow band gaps: the former includes gallium oxide, diamond, and aluminum nitride, while the latter includes indium arsenide and gallium arsenide. Yu Jinxiu, Project Manager for the Fourth Generation at the Shanghai Science and Technology Commission, presented a timeline: third-generation research began in laboratories back in the 1980s, but industrial implementation only scaled up in 2018 when Tesla used silicon carbide in the Model 3, a transformation that took over thirty years. Regarding gallium oxide, the first global device result was recorded in 2012; following this trajectory, application may not appear until after 2040. Therefore, it was decided to start work in 2024, focusing on scenarios that seem more than a decade away; this stage is in the period of investment observation, and the leading scheme is capable of attracting attention.
End consumers do not focus on material generations; their main need is to achieve required characteristics while significantly reducing costs. Cost is a key factor for the fourth generation: gallium oxide allows for liquid-phase processing, unlike the strictly gas-phase method for silicon carbide, which provides an order-of-magnitude cost advantage after mass production begins. In high-frequency communication filters, existing materials are approaching physical limits, forcing system developers to seek new, advanced materials. The Lingang cluster plan for 2026–2028 aims to attract no fewer than 50 companies and achieve an industrial volume exceeding 5 billion yuan by 2028.
Scientific achievements from universities and institutes do not guarantee industrial competitiveness on their own. Li Sixhua, CEO of the Sinan semiconductor incubator, notes that researchers who create companies often lack experience based on market demand. In 2025, Sinan launched a fourth-generation conceptual verification center in the city, creating a closed loop from selection to pilot production and transforming results around pathways related to gallium oxide, diamond, aluminum nitride, and arsenides. Xinjing Energy, founded by Professor Jiaotong Hu Zhijiu of Shanghai University, is a prime example: just days after winning the Create-in-Shanghai competition last year, it settled in Lingang. Hu's team developed a thermoelectric chip integrating 60,000 units; Sinan linked this to the need to utilize waste heat from Baosteel, promoting its utilization for the steel, cement, and thermoelectric industries, where China's annual losses exceed losses equivalent to more than 100 Three Gorges dams. Regarding policy, Lingang offers grant-to-investment conversion: the government supports transformations through R&D grants, and then the Qihang angel fund, involving external investors, makes the initial investment, with the principal of the grant being returned. Sinan also anticipates a 500 million yuan angel fund from Lingang Group, into which five projects in the incubator have been invested; 1200 square meters of conceptual verification volume were created in a year, and one project was transformed in the first half of the year.