Lightstandard is betting that the next bottleneck in artificial intelligence will not be only the number of available accelerators, but the speed and energy cost of moving data between them. The Chinese optical-computing company said it had completed multiple financing rounds during the past year and was approaching a post-money valuation of RMB20 billion. TMTPost reported that the company will use the funds to scale delivery of its optical-computing systems and advance research on glass-substrate optical-computing chips.
The announcement comes as AI workloads place increasing pressure on memory, networking, and interconnects. Training and serving large models requires chips to exchange enormous volumes of information. When that data movement becomes too slow or too power hungry, faster processors alone do not solve the system-level problem. Optical technologies aim to use light rather than electrical signals for parts of that connection layer, potentially raising bandwidth and reducing some energy costs.
Lightstandard’s financing is a company disclosure reported by TMTPost, and the valuation should be understood as a private-market estimate, not a public-market price. The company says it has delivered more than 10 optical-computing systems, has orders or intended orders worth hundreds of millions of yuan, and serves large technology firms, vertical AI companies, and intelligent-computing centers. Those figures suggest commercial traction, but they do not independently establish the technical performance or market leadership of the company’s products.
Funding Backs Optical Systems and Glass-Substrate Chip Development
TMTPost reported that Lightstandard has raised funding through several rounds over the past year. The company’s stated priorities are expanding large-scale system delivery and continuing research and development on optical-computing chips that use glass substrates. Founder and chairman Xiong Yinjian is named in the report.
Optical computing can describe several related approaches. Some companies build photonic processors that perform mathematical operations with light. Others focus on optical interconnects that move information between processors, memory, and network equipment. Lightstandard’s announcement links system delivery with chip development, indicating that it is pursuing both a product layer and a longer-term component roadmap.
Glass substrates are relevant because advanced chips need a stable base for dense interconnects and thermal management. The material is often discussed as a possible alternative or complement to traditional organic packaging substrates for high-performance computing. A glass-substrate optical chip remains an R&D objective in Lightstandard’s announcement, not a mass-market product with independently verified specifications. That distinction matters in a sector where laboratory demonstrations can take years to become reliable commercial hardware.
The company’s near-RMB20-billion valuation nevertheless shows how investors are treating optical technologies as part of AI infrastructure rather than a niche hardware category. Chinese AI investment has expanded beyond model developers to include memory, networking, server integration, power, cooling, and interconnects. EastFrontier’s coverage of Lightelligence’s optical-link strategy for AI data centers described the same pressure point: AI clusters need to exchange information quickly enough that communication does not overwhelm compute gains.
AI Clusters Need Faster Data Movement, Not Only Faster Chips
The case for optical systems begins with the economics of AI clusters. A large model may be spread across many accelerators, and each accelerator must repeatedly exchange activations, weights, gradients, or inference data with other parts of the system. As the number of chips rises, that traffic can become a constraint on latency, throughput, and energy consumption.
Conventional electrical connections remain essential throughout data centers. Optical technologies do not remove that infrastructure overnight. But they can help at the points where moving data over copper traces or cables becomes difficult at high bandwidth. The business opportunity is therefore tied to a broader AI buildout: more complex models and larger clusters create demand for better ways to link computing resources.
Lightstandard says it has delivered more than 10 optical-computing systems. The report also says it has won or agreed intended orders totaling hundreds of millions of yuan. Those statements provide a starting point for assessing commercial progress. They do not reveal customer names, shipment volumes, system utilization, or recurring revenue, so readers should not infer a fully mature market from the announcement alone.
Still, systems matter because they move the technology beyond a component roadmap. A customer buying an optical-computing system needs integration with servers, networks, software, cooling, and operating procedures. The company’s attempt to combine system delivery with chip R&D may therefore give it an opportunity to learn from deployments while it develops more specialized hardware.
China’s AI infrastructure race has made these integration questions more important. EastFrontier previously examined Kiwimoore’s planned Hong Kong listing for AI network infrastructure, where the commercial case also depended on supporting the connections around AI compute. Lightstandard’s focus on optical systems is a different technical route, but it reflects the same market view: the value of an AI cluster depends on the network and interconnect layer as well as the chips themselves.
A High Valuation Raises the Need for Operational Evidence
A valuation nearing RMB20 billion puts pressure on Lightstandard to show more than a promising technical thesis. Investors will want evidence that the company can deliver systems at scale, secure durable customers, manage supply chains, and make its technology operate reliably in production environments. In hardware, that path can be difficult because prototypes, pilot shipments, and mass deployments have very different cost structures.
TMTPost describes Lightstandard as supplying customers among large technology companies, vertical AI firms, and intelligent-computing centers. If those relationships develop into repeat business, they could give the company an important foothold in China’s AI infrastructure market. If intended orders do not translate into revenue, the distance between a private valuation and an operating business will become more apparent.
The glass-substrate chip work is a longer-horizon bet. It could help Lightstandard differentiate itself if it produces components that improve bandwidth, integration density, or energy efficiency. But the company has not provided a public timeline for commercial release or independently comparable performance data. Its claims about technical advantages should therefore remain company claims until customers, third-party testing, or product filings provide more detail.
The verified facts are already meaningful. Lightstandard has disclosed multiple new funding rounds, a post-money valuation close to RMB20 billion, a plan to scale optical-computing-system delivery, and a program to develop glass-substrate optical chips. It says it has delivered more than 10 systems and secured orders or intended orders worth hundreds of millions of yuan.
Those facts place the company at an increasingly important layer of China’s AI supply chain. As models grow and clusters expand, computation will be constrained by the movement of data as well as its processing. Lightstandard’s investors are betting that optical systems can solve enough of that problem to become a core piece of AI infrastructure. The next proof will come from disclosed deployments, product reliability, and whether the company can turn an optical-computing narrative into sustained commercial delivery.