Hard Krypton Exclusive | Secured 100-million-yuan Seed Round Financing Within 4 Months of Establishment, Shanghai Jiao Tong University-affiliated Enterprise Proposes a New All-Optical Interconnection Architecture
Yingke learned that Xinjie Guanghe, an optical interconnection enterprise, recently completed several rounds of seed - round financing, with a total financing amount reaching hundreds of millions of yuan. The investors include Qigao Capital, a fund affiliated with Shanghai Jiao Tong University, Hanyuan Assets, Xiaomiao Fund, Tongding Group, an investor in the optical communication industry, and the market - oriented institution Junding Fund. This round of financing will be mainly used for the expansion of the core team, chip tape - out, engineering verification, and product R & D.
Xinjie Guanghe was founded in January 2026. The core team is composed of the scientific research team from the National Key Laboratory of Optical Communication at Shanghai Jiao Tong University and industry professionals with experience in hard - tech product R & D, enterprise operation, and industrialization. The company focuses on building the Photonic Nexus all - optical interconnection platform, providing all - optical interconnection solutions from silicon - photonics chips, optical engines to optical switching systems for AI computing power clusters, high - performance computing, and next - generation intelligent computing centers.
The company's founder, Ying Ying, and co - founder, Shi Bo, were classmates at Peking University. After graduation, they successively worked at international technology companies such as IBM and Xilinx and held core management positions in hard - tech startups. Each of them has over 20 years of experience in hard - tech product R & D, commercialization, and enterprise operation. Professors Lu Liangjun and Li Yu, the co - founders, are both from the National Key Laboratory of Optical Communication at Shanghai Jiao Tong University. This team is one of the earliest in China to conduct research on silicon - photonics chips. The two professors have nearly 20 years of experience in the fields of optical transceiver chips, optoelectronic co - packaging, and OCS chips, and their technical level ranks among the international advanced levels. Before joining Shanghai Jiao Tong University, Professor Li Yu worked at AMF and Marvell for 6 years, leading the R & D of silicon - photonics chips.
Driven by the continuous growth of AI computing power, the trend of "optical in, copper out" is extending from the cabinet - level and board - to - board level to the inside of chip packages. On June 10, 2026, the Ministry of Industry and Information Technology issued the "Implementation Opinions on the Innovative Development of 'Artificial Intelligence + Information and Communication' (2026 - 2028)", regarding optoelectronic chips, OCS devices, CPO devices, etc. as the foundation for the development of artificial intelligence, and proposing to strengthen the R & D of high - end optoelectronic chips and devices, as well as the R & D and verification of technologies and products such as all - optical switching devices.
From optical modules, NPO, CPO to OIO, optical interconnection technologies are getting closer to computing power chips, aiming to achieve higher bandwidth, lower latency, and lower power consumption over shorter distances.
The core idea of the Photonic Nexus all - optical interconnection platform proposed by Xinjie Guanghe is to further introduce OCS (optical switching) in addition to OIO, breaking through the limitations of traditional point - to - point optical connections. By using optical switches to achieve direct optical routing, the "optical - electrical - optical" conversion in the traditional electrical switching process can be reduced, thereby reducing link latency and system power consumption and better adapting to the AI cluster Scale Up scenario.
In terms of the OCS route selection, Xinjie Guanghe adopts the silicon - photonics solution. Compared with the common optical switching solution based on MEMS (micro - electro - mechanical systems), silicon - based OCS has more advantages in terms of size, cost, reliability, and large - scale integration potential. Currently, the company has completed the R & D of a 32×32 silicon - based OCS chip, with key indicators such as insertion loss leading the industry. It is promoting the transformation of the chip into a commercial system and is also planning products with 64 ports and higher port scales.
As the products enter the engineering and customer verification stages from the technical verification stage, Xinjie Guanghe will continue to expand its teams in silicon - photonics chip design, optoelectronic co - packaging, OCS systems, test verification, supply chain, and market applications. The company hopes to attract more talents with backgrounds in chips, optical communication, AI infrastructure, and system engineering to jointly participate in the industrialization of the next - generation all - optical interconnection platform.
The following is an excerpt from the interview between Yingke and founder Ying Ying:
Yingke: The all - optical interconnection Photonic Nexus platform of Xinjie Guanghe can be understood as "OIO + OCS". What are its differentiated features and advantages compared with pure OIO?
Ying Ying: Our technology is targeted at the high - bandwidth, low - latency, and low - power - consumption interconnection scenarios in the next - generation AI computing power clusters. OIO (Optical I/O) mainly solves the problem of converting high - speed electrical signals around GPUs or other computing power chips into optical signals. However, if data exchange still relies on the traditional electrical switching layer, there will still be power consumption and latency caused by optoelectronic conversion in the link.
The all - optical interconnection architecture of Xinjie Guanghe adds OCS (optical switching) in addition to OIO, completing data scheduling through optical paths and replacing some traditional electrical switching functions. This means that GPUs, computing power chips, memory, or other key units can achieve interconnection with lower power consumption under higher bandwidth density. This type of architecture is naturally suitable for the Scale Up scenario and is also the direction actively explored by global leading AI infrastructure manufacturers.
Yingke: The company has laid out multiple optical interconnection products. What is the logic behind the product roadmap? What is the current progress?
Ying Ying: Our product layout revolves around two main lines: OIO and silicon - based OCS. The technical foundation is reused, both based on MZI (Mach - Zehnder modulator) and MRR (micro - ring modulator). On the OIO side, we first promote the "narrow and fast" high - speed silicon - photonics PIC, with the single - wavelength rate evolving from 100G, 200G to 400G. These products are highly compatible with existing applications such as optical modules and NPO, can enter industrial verification quickly, and are also important core components of next - generation CPO/OIO products. This part of the products will quickly enter mass production.
Secondly, the company will rely on self - developed high - speed silicon - photonics transceiver chips, combined with electrical chips and advanced 3D optoelectronic co - packaging technology, to launch high - bandwidth, high - density optical engine products for CPO, used in the low - power - consumption interconnection scenarios of intelligent computing centers. We are conducting joint development with customers and plan to produce in small batches next year.
Thirdly, we will further promote the high - density optical transceiver engine for the GPU side, that is, OIO, targeting the GPU cluster Scale Up interconnection, memory pooling, and CXL expansion. Through the micro - ring modulation scheme and optoelectronic co - design, we will achieve a greater improvement in power consumption and bandwidth density, and it is expected to be produced in small batches the year after next.
On the OCS side, the company has completed the R & D of a silicon - based 32×32 strictly non - blocking OCS chip, with key technical indicators such as insertion loss leading the global industry. It plans to launch a 32×32 OCS system early next year. The port scale of the subsequent silicon - based OCS chips will continue to develop towards 64 ports, 128 ports, or even higher port numbers. This series of products will depend on the market maturity, and it is expected to be sampled next year and enter small - batch mass production the year after next.
Yingke: When technology moves from the laboratory to large - scale commercialization, it often gets stuck on yield, cost, and manufacturability. How will Xinjie Guanghe handle these issues?
Ying Ying: The scientific research team from Shanghai Jiao Tong University has laid a solid foundation at the laboratory stage. They not only conduct scientific research but also have rich experience in the engineering field. The team has been deeply involved in the optical interconnection field for 20 years, continuously improving chip performance and yield through multiple tape - outs, design iterations, and process optimization. More importantly, we are not only focusing on single - point devices but also have long - term experience in chips, packaging and testing, and system applications.
The team began to consciously accumulate engineering capabilities several years ago and has formed relatively complete experience in chip packaging and testing, system verification, and application scenario adaptation. On the foundry side, the company cooperates with leading domestic and foreign foundries with rich experience and has also recruited an experienced engineering team. We have shifted from pursuing "the highest indicators" to focusing on the balance of specification satisfaction rate, yield, cost, reliability, and delivery cycle, establishing the core logic oriented towards mass production.
Yingke: Compared with other optical interconnection enterprises, what do you think is the core advantage of Xinjie Guanghe?
Ying Ying: Our core advantage lies in the combination of technical accumulation and industrialization capabilities. Although the company has been established for a short time, it is backed by 20 years of silicon - photonics and optical interconnection technology accumulation. The core R & D team has long been deeply involved in silicon - photonics OIO, OCS, and optical interconnection systems, with the ability to conduct chip design, process collaboration, and system - level verification. The top - notch scientist team has continuous innovation capabilities and has long maintained a leading position in cutting - edge technology exploration. The company has extremely rich technical reserves, with a wide technical moat both in the short - term and long - term.
At the same time, the company's operation team consists of senior industry professionals with many years of experience in hard - tech entrepreneurship and industrialization. We believe that the combination of the scientists' underlying technical capabilities and the industrial team's productization, customer - orientation, and organizational management capabilities is the key for hard - tech companies to move from the laboratory to the market. The company's leading optoelectronic co - packaging experience and technologies and products such as the 32×32 silicon - based OCS also lay the foundation for us to conduct joint verification with leading customers and ecological partners.
Yingke: The computing power scenario is regarded as an important market for OIO. When do you think OIO will truly explode on a large scale?
Ying Ying: The industry generally believes that OIO will be accelerated in overseas leading AI computing power clusters in the next one to two years. Affected by the supply chain, production capacity, system architecture, and customer verification rhythm, there is usually a certain time lag in large - scale commercialization in China. Therefore, starting to layout and conduct R & D at this time is a critical window period to enter the market.
We believe that as the scale of AI models expands and the computing power clusters continue to expand, the pressure on traditional electrical interconnection in terms of power consumption, bandwidth density, and latency will become more prominent. OIO and OCS are not just single - device upgrades but part of the upgrade of the next - generation computing power infrastructure architecture.
Yingke: What are the company's plans for products and the market next?
Ying Ying: In terms of products, we hope to become a leading domestic provider of optical interconnection technology solutions, providing customers with high - performance, low - power - consumption, and scalable products. The company's optical interconnection product line is being advanced in sequence: the single - wavelength 200G/400G silicon - photonics chip products are being iterated and are planned to enter mass production. The CPO optical engine products are expected to be sampled by the end of the year, and the OCS system is expected to enter the sampling and customer verification stage early next year.
In terms of the market, we focus on serving leading cloud providers, AI computing power cluster operators, server and GPU platform manufacturers, and optical communication equipment ecological partners. The company will promote joint verification around scenarios such as Scale Up interconnection, low - power - consumption interconnection in intelligent computing centers, high - density optical engines, and all - optical switching, participate in defining the next - generation optical interconnection standards, and accelerate large - scale implementation through collaboration with industrial partners.
Investors' views:
Qigao Capital: Xinjie Guanghe is a highly scarce target in the silicon - photonics track that we have seen. The company's core technology team comes from the long - term accumulation in the silicon - photonics field at Shanghai Jiao Tong University, and the operation team also has experience in hard - tech industrialization. We are optimistic about the company's development of product lines such as optical engines, optical interconnections, and optical switching based on the silicon - photonics platform, and believe it will become an important infrastructure provider for the next - generation AI computing power clusters.
Hanyuan Assets: The Shanghai Jiao Tong University fund has long been concerned about the transformation of hard - tech achievements. Relying on more than 20 years of technological accumulation at Shanghai Jiao Tong University, Xinjie Guanghe promotes the industrialization of scientific research achievements in the fields of silicon - photonics and optical interconnection. The full - stack layout of OIO and OCS directly addresses the bottlenecks of AI computing power consumption and interconnection. We are optimistic about the team combination of "scientists + industry professionals" and look forward to the company promoting the large - scale application of domestic silicon - photonics technology.
Xiaomiao Fund: In hard - tech entrepreneurship, the team is the first thing to consider. Xinjie Guanghe is jointly promoted by academic experts with long - term scientific research accumulation and an entrepreneurial team with industrialization experience, combining underlying technical capabilities and commercial execution. We believe that this combination has the opportunity to form a sustainable competitive advantage in the optical interconnection track.