New Photonics Laboratory in Shanghai China Accelerates the Chip Race with New Advancements in Photonics

Related Vendor

FAULHABER Drive Systems

With photonic chips, China aims to close a technological gap that traditional silicon chips alone can no longer fill. The increasing computational demand of AI data centers, limited scaling, and US sanctions are adding pressure. In Shanghai, a new laboratory is set to bring optical computing closer to industrial applications.

China is increasingly focusing on the research and development of photonic chips to meet the growing computational demand of its AI data centers. Following several recent technological advancements, a new laboratory for applied research into computing with light particles has now been established in Shanghai.

The new "Shanghai Key Laboratory of Integrated Photonic Computing Chips and Systems" began operations in early June 2026, according to Shangguan Xinwen, an online portal of the government-affiliated Jiefang Ribao. It is described in the report as the country's first platform where universities and industry "jointly work on optical computing."

The laboratory is supported by Shanghai Jiao Tong University, one of the leading institutes for academic photonic research in the People's Republic, and the company Lightelligence, a Chinese start-up for photonic computing that went public in Hong Kong in April 2026.

"Optical computing is an important approach to achieving breakthroughs in computing performance, with advantages in bandwidth, latency, and energy efficiency," the South China Morning Post quoted Zou Weiwen, the director of the new laboratory. He is a professor of photonics at Shanghai Jiao Tong University, where the laboratory is also located.

More Computing Power Despite Physical Limitations

The AI boom is also driving rapidly increasing demand for more computing power in China, while at the same time Moore's Law, the continued performance improvement of silicon chips through miniaturization, is reaching its limits. Photonic chips are considered one of several alternative technologies that could solve this problem.

While conventional chips send electrons through silicon circuits, these new chips instead use light particles, or photons, to transmit and process data. This offers several advantages. Light is fast and generates less heat. Optical chips operate with very low latency and high energy efficiency. Most importantly, calculations can be executed simultaneously with light, as signals on different wavelengths do not interfere with each other.

This massive parallelism is the advantage that has particularly intrigued Chinese researchers for several years, as it aligns perfectly with the requirements of artificial intelligence. Large AI models consist primarily of huge, parallelizable matrix computations. For this, computing with light is ideally suited.

Urgent Need

In China, technology companies are striving, just like their U.S. counterparts, to rapidly expand their computing capacities in order to train and operate ever-larger AI models. However, the high-performance GPUs from Nvidia, which are urgently needed for this, have been placed on Washington's boycott lists for China since 2022 in an effort to hinder its progress in the AI sector as much as possible.

Since then, the Chinese central government in Beijing has further accelerated its photonics efforts, which began a decade ago. The new laboratory in Shanghai can also be seen as one of many concrete responses to the American containment strategy. Optical chips are far less dependent on the most advanced lithography machines from ASML, whose delivery to China the U.S. government has also been preventing for years through diplomatic pressure on the Netherlands.

The performance of photonic chips depends more on how cleverly the light paths are designed, not on how many transistors are packed into a small space. If computing with light can be made practical for widespread computational applications, the impact of American technology boycotts against China could be significantly reduced.

Mandated Drive for innovation

Beijing is pursuing a strategy in several areas of not competing head-to-head with U.S. or European companies like Nvidia or ASML but instead replacing their technologies with alternatives. The logic behind this was explained by Huawei founder Ren Zhengfei several years ago in one of his rare interviews. His country was still "a generation behind the United States" in conventional chips, Ren said to the People's Daily.

This lag is to be made up by "replacing physics with mathematics, Moore's law with non-Moore approaches, and the limits of individual chips with computing in networks." Photonics is one of these bets against Moore. Since 2015, the People's Republic has been promoting photonics as a national priority. The 14th Five-Year Plan, which has just been replaced by the 15th, already declared it a strategic technology, and since then, funds from the programs of the Ministry of Science and the National Research Foundation have been directed into its own laboratories.

Since the U.S. tightened its sanctions against modern chip manufacturing in China, the pace has accelerated. In 2024, the country's first pilot factory for photonic chips was established in Wuxi. The facility, operated by Jiao Tong University, announced its first batch of domestically produced six-inch wafers and a series production of high-speed chips in June 2025, which state media celebrated as a leap from "technology latecomer" to "industry leader" in high-quality optical components.

A team at Tsinghua University in Beijing developed two variants of an optical chip named Taichi, the second of which is said to be capable of training AI models solely with light. According to Chinese media reports, Taichi II operates about a thousand times more energy-efficiently than Nvidia's high-end H100 chip, a claim that is difficult to verify. Additionally, researchers from Fudan University presented a chip designed to accelerate data within an AI cluster.

Parallelism in focus

How much the aforementioned parallelism is at the center of China's interest in photonics is demonstrated by the domestically developed chip "Meteor-1." According to its own statements, a team led by Xie Peng at the Shanghai Institute of Optics and Fine Mechanics of the Academy of Sciences built the "first highly parallel integrated optical computing chip," which performs more than a hundred computing processes simultaneously and, in purely computational terms, achieves a peak performance of 2,560 TOPS, comparable to Nvidia's most advanced graphics processors.

Here, too, researchers deliberately focused on parallelism because there is little room for further progress in chip size and optical clock frequency. The "chip war" between the West and China, often highlighted by Western media, is more complex than the debates about nanometer generations or EUV machines might suggest.

American companies like Lumentum, Coherent, Intel, and Lightmatter cover the entire value chain of photonic computing. European providers are very strong in optical components and materials. However, China is rapidly catching up in all these areas and has simultaneously developed its own strength in the supply chain. In thin-film lithium niobate, a key material in photonics, Chinese manufacturers already account for around 42 percent of global capacity.

Only at the high end of the scale is there still a gap from a Chinese perspective. In the market for particularly fast laser chips, the domestic Chinese share is only about four percent. It will still be a long and arduous journey before China's R&D efforts in photonics reach the everyday operations of AI data centers. There is a lack of a mature ecosystem of software and algorithms that can fully utilize the optical hardware, admitted the newly appointed laboratory head Zou in Shanghai in an interview. This is precisely the gap the newly opened photonics laboratory in Shanghai aims to help close. The next round in the technology race may be decided less in nanometers than in wavelengths of light. (sb)