Company Profile
Jiuwei Guangzi Technology (Kunshan) Co., Ltd. (referred to as "Jiuwei Guangzi ") was founded in 2024, with its core technology route focusing on self-developed phase-change photonic materials. The company's business is divided into two major sectors: next-generation photonic switching chips for AI data centers, and maskless nano-lithography equipment, with R&D, trial production, and market implementation proceeding simultaneously. The technology implementation goals include two aspects: on the one hand, optimizing the optical interconnection links of computing centers to improve data transmission efficiency and energy consumption performance; on the other hand, promoting the localized industrialization of high-end nano-lithography equipment to fill the domestic supply gap for related equipment.
Editor's Note
AI computing clusters are advancing from the scale of tens of thousands of cards to millions. Traditional electrical switching has approached its physical limits in terms of port density and energy efficiency metrics, making all-optical switching shift from an "alternative" to a "must-have". Among the current mainstream OCS technology routes, the MEMS micromirror solution led by Google has achieved large-scale deployment, but controversies over its long-term reliability and system collaborative scheduling capabilities persist. In this issue of "Chip Unveiled · Experts Talk Chips", we interview He Qiang, Associate Professor at Huazhong University of Science and Technology and CTO of Jiuwei Guangzi, to deeply explore whether phase-change silicon photonics technology, with its differentiating highlights of low cost, high integration, and fast switching, can break through the limitations of existing routes and become a core variable in the upgrade of the next-generation global computing network.
*The following is a transcript of the podcast, confirmed and authorized by the guest.
Phase-Change OCS: "Optical Switches" at the Atomic Scale
Huan Shi: Professor He, I understand that you are engaged in research related to phase-change materials. In daily life, water turning into ice is a phase change. Is there any essential difference between the phase-change photonics you study and the phase change of water?
He Qiang: The principle is somewhat similar, both involving the state transition of matter from disorder to order. However, we focus on a new type of nanomaterial, whose scale is only a few hundredths of a human hair, enabling structured state switching at the atomic level. This change in microscopic state can directly affect optical properties—producing a significant regulatory effect on the propagation of light. Therefore, we have applied this phase-change material to optical path switching in AI data centers, super-resolution laser direct-write lithography, and other areas.
Huan Shi: So "phase-change photonics" is the core label of your technology. Before hearing you introduce this term, I didn't know what it meant. Could you help us break down the application directions you just mentioned in more detail?
He Qiang: Sure. Phase-change photonics is currently a frontier research hotspot in the photonics field both domestically and internationally. As a foundational material technology, it possesses strong technical extensibility and can be widely applied in multiple core tracks such as optical computing, optical path switching in AI data centers, super-resolution laser direct-write lithography, and new displays.
However, the precise control of phase-change materials is a core difficulty in the industry. How to achieve stable switching between dual-state and multi-state of the material through optical and electrical methods is the core technical challenge.
Huan Shi: Everyone says Moore's Law has reached its limit and we can only rely on light, but light is not so "obedient".
He Qiang: Yes, it requires joint efforts from both the industry and academia to tackle and solve.
Huan Shi: Has the phase-change photonic product you developed moved from the laboratory to industrialization?
He Qiang: Not yet. However, relying on the long-term technical accumulation and industry-university-research collaboration advantages of Huazhong University of Science and Technology, we are continuously promoting the industrialization of phase-change photonic technology. Currently, we have achieved phased progress in two core scenarios with high maturity and high demand.
First, in the AI computing scenario, we have made targeted optimizations to the OCS all-optical switch. Compared with the current mainstream MEMS technology, the phase-change photonic solution performs better in switching latency, power consumption, and integration. The team has currently completed the technology iteration and will launch a new generation of scaled port optical switching chip, officially entering the stage of commercial implementation.
Second, in the nano-lithography scenario, our 100nm maskless direct-write lithography equipment has already secured overseas orders. In the future, we hope to consolidate the overseas market while continuously expanding domestic customers to promote the scaled implementation of the technology.
Huan Shi: As a company that has only been established for two years, how did you win your first batch of customers?
He Qiang: Mainly because we accurately hit the market demand. Currently, in the field of maskless lithography, traditional optical solutions struggle to achieve high-precision lithography of 100nm and below, unless electron beam lithography equipment is used. But electron beam lithography has two core problems: first, the equipment price is extremely high, making it unaffordable for small and medium-sized R&D institutions and enterprises; second, electron beam lithography is a "choke point" technology under overseas embargoes, severely restricting domestic industrial development.
Jiuwei Guangzi , leveraging the nonlinear absorption effect of phase-change materials, has achieved spot resolution beyond the diffraction limit. This not only avoids the technical restrictions of electron beam lithography but also keeps costs very low. This is also an important reason why we can achieve rapid commercial implementation.
Huan Shi: Once this technology of yours matures, will the cost be even lower? After all, there is no expensive expenditure for lithography masks.
He Qiang: Exactly. There is still significant room for optimization in R&D costs and equipment production costs. The optical path structure of the entire equipment is relatively simple, with low technical complexity. It can achieve 100nm high-precision lithography at a low cost, possessing high market competitiveness.
Is the Cost of Phase-Change OCS Much Lower Than That of MEMS?
Huan Shi: This OCS technology route is being hotly discussed in the industry, and Google is also leading one of the directions. What do you think of the development prospects of this route? Do you think OCS will become the mainstream all-optical switching solution in the future?
He Qiang: I believe it will.
First, all-optical switching has larger bandwidth. With the rapid popularization of AI large models and supercomputing power, we are already accustomed to using AI tools like DeepSeek and Doubao to solve problems in daily life. In fact, every task you submit to AI will be uploaded to the data center for calculation and processing, and what supports the operation of the data center is tens of thousands of GPU cards. Frequent data interaction is required between cards. As the interaction frequency and data traffic surge, traditional electrical switching solutions are struggling to expand bandwidth.
Second, all-optical switching has lower latency. Currently, electrical switching is the main approach, where signals need to be repeatedly converted between optical and electrical forms. Latency, power consumption, and bandwidth constitute three bottlenecks that are difficult to break through. Optical signals are inherently the core carriers of fiber optic communication, with naturally abundant bandwidth and transmission speeds at the speed of light. Using light directly for switching eliminates the optical-electrical conversion process, significantly reducing latency. Reflected at the data center level, this means reduced energy consumption and optimized costs. Falling on the ordinary user side, it means lower usage costs and faster responses for AI applications.
From these two perspectives, all-optical switching is an inevitable trend for the future.
Huan Shi: How is the maturity of this route now?
He Qiang: As an emerging technology direction, OCS is currently not mature overall, but the demand is very strong. For example, the MEMS solution being used by Google, which is mechanical micromirrors, deflects light beams through mirrors. Although it already has a mature supply chain, it can only solve basic optical path switching and has limitations in system collaboration and intelligent scheduling.
The phase-change photonic solution is different. Its chips will be made more compact, with lower power consumption and faster switching speeds. The cost is even a fraction or even tens of times lower than that of traditional MEMS solutions. Because of its high integration, a single wafer can integrate hundreds to thousands of chips, greatly reducing equipment volume and enabling the deployment of more GPU computing power in limited rack space. Therefore, it is actually a technology direction with lower costs.
Huan Shi: If we make chips with this special process, are there tape-out conditions domestically?
He Qiang: Currently, domestic tape-out conditions are quite good. Compared to electrical chips, the gap between domestic and foreign in the optical chip field is much smaller, basically slightly behind, or even running in parallel. We have platforms like Jiufengshan Laboratory domestically for foundry work.
Huan Shi: That's great. We might achieve technological catch-up in the optical chip era.
He Qiang: Yes, this is a structural opportunity in the optical chip track—less reliance on advanced lithography, making it easier for the domestic supply chain to follow up.
Huan Shi: We don't need EUV and won't be "choked" by others, right? Then let's accelerate embracing "light" and achieve technological transcendence earlier.
He Qiang: Yes, "light" is definitely the future. The slogan of Jiuwei Guangzi is also that we want to "build the optical foundation of the intelligent world".
Huan Shi: Although this path sounds even more fantastic, I believe you have already laid a solid theoretical foundation.
From Academia to Mass Production: Overcoming the Short Board of the Wooden Barrel
Huan Shi: Professor He, from academia to industrialization, what is the biggest challenge you are currently facing?
He Qiang: This involves the essential difference between academia and industrialization. Academic research focuses more on principle verification and technical exploration, while industrialization is completely different—the core criterion is customer satisfaction, requiring products to be stable, reliable, without short boards, and capable of scaled mass production. The essence of this transition is the leap from "0" to "1" and then to "100". During this period, a large number of engineering problems need to be solved, which requires continuous resource investment and sufficient time for iterative trial and error. This is the core challenge of industry-university-research transformation. Our products must not have any short boards like the wooden barrel effect and must be customer-demand oriented.
Huan Shi: In the OCS track you are in, what is the R&D progress of domestic and foreign peers? Is the difference between them very large?
He Qiang: We think the difference is not very large. The concept of OCS was indeed first proposed internationally, but we have a mature supply chain system, multi-level policy support, and a complete mechanism for industry-university-research collaboration domestically. Our promotion speed and development acceleration are very fast. I believe we can achieve technological catch-up in the short term.
Huan Shi: It's good to have confidence. So what do you think is the most core competitive barrier of Jiuwei Guangzi ?
He Qiang: Relying on the technical accumulation of Huazhong University of Science and Technology and the industrial cluster advantages of Wuhan Optics Valley, we have built three core competitive barriers:
First, the full-chain technology barrier. Our team has mastered the technology closed loop from phase-change materials, process preparation, chip devices to equipment terminals, as well as the end-to-end systematic capability of scenario applications.
Second, the composite talent barrier. The core team of our company comes from Huazhong University of Science and Technology and leading domestic communication and semiconductor companies. They have both academic and industrial backgrounds, balancing technological innovation and engineering mass production capabilities, which is extremely rare among domestic start-ups.
Third, the industrial ecology barrier. We are rooted in the complete optoelectronic industry chain of Wuhan Optics Valley, relying on local foundry platforms such as Jiufengshan Laboratory to significantly shorten the chip iteration cycle and quickly respond to market demands.
Huan Shi: Okay. If we look forward to five years from now, what kind of company do you hope Jiuwei Guangzi will grow into? What is your ultimate imagination for this phase-change OCS technology route? How do you think it can ultimately change the world in the future?
He Qiang: Then let me boldly imagine it. Looking into the future, phase-change photonic materials are defined as the next-generation core optical field regulation materials. Compared with technical solutions such as liquid crystals, traditional silicon photonics, and MEMS micromirrors, it has comprehensive advantages in three core dimensions: switching speed, energy consumption control, and integration. The application prospects cover multiple frontier fields such as terahertz regulation, visible light regulation, all-optical computing, and super-resolution lithography. The potential is infinite, and it is worth the whole industry working together to develop it.
For enterprise development, Jiuwei Guangzi will seize the era opportunity of the AI computing power explosion, continue to deeply cultivate the two core tracks of OCS all-optical switching and phase-change lithography, promote technology from the laboratory to the production line, and form differentiated competitiveness as soon as possible. At the same time, it will accelerate commercial implementation, provide domestic wafer fabs with autonomous and controllable new lithography solutions, and assist in the domestic substitution of the optoelectronic industry.
Huan Shi: Your imagination is wonderful, with both long-term and short-term visions. I am quite happy today to hear Professor He choose a very important technology route in this era and persist in it.
He Qiang: This is a great era for starting businesses and making achievements.
Huan Shi: Okay, thank you very much, Professor He, for sharing so much with us. We also look forward to you coming to our "Chip Unveiled" to share again when your products have further results. Finally, I wish you more and more customers.
He Qiang: Thank you.
Interview | Huan Shi Editor | Xiao Qie Review | Huan Shi