Hangzhou Xinliu Weilan Intelligent Technology has completed its first funding round worth tens of millions of yuan, with Qigao Capital as the sole investor. The company positions itself as an AI-for-IC design startup.
Founded in 2026 by Sun Qi
According to Touzijie AI, growing chip design complexity has made traditional manpower-heavy approaches harder to sustain. Against that backdrop, Zhejiang University researcher Sun Qi founded Xinliu Weilan in 2026 to combine AI with experience in electronic design automation and build a new generation of intelligent chip design platforms.
The company describes its long-term vision as “using AI to design AI.” It says the team is now among the relatively small number in the industry that have run an industrial-grade EDA toolchain end to end and deployed agents in advanced-process chip projects.
Founding and technical team
Founder Sun Qi is a researcher and doctoral supervisor at Zhejiang University and a former postdoctoral researcher at Cornell University. He has worked in the AI-plus-IC field for years and has published nearly 100 papers in top international conferences and journals. His previous work has received the First Prize of Zhejiang Science and Technology Progress, the Huawei Spark Award, the EDA Young Scientist Award, and eight best paper awards or nominations at international academic conferences, according to the article.
Technical lead Chen Zhengrui, PhD, proposed what the article describes as Agentic EDA infrastructure built for real industrial toolchains. He has delivered production-grade standard cell libraries and back-end optimization solutions for major IDM companies and has experience in tape-out and industrial deployment.
The project has also won recognition from Wu Hanming, an academician and director of Zhejiang University’s Faculty of Information Science, and Zhuo Cheng, vice dean of integrated circuits at the university. As strategic advisers and senior industry figures, the two will support the company in technology roadmap planning, industrial resource connections and enterprise deployment.
Focus on AI-driven EDA tools
EDA, short for electronic design automation, sits at the upstream end of the chip industry chain. It uses computer software for integrated circuit design, simulation and verification, and it affects chip performance, yield and development cycles.
Xinliu Weilan frames its broader ambition as “Generalized AI for Design.” The company says it wants to build AI-driven EDA tools that lower the barrier to chip design, shorten development cycles, move more innovative chip ideas from concept to mass production, and support China’s push for self-controlled semiconductor development.
Its IC design platform is built to understand design targets, plan workflows, orchestrate tools, analyze results and keep optimizing through repeated iterations. Under the company’s description, engineers provide goals and constraints, and the agent can break down tasks, call industrial tools to execute the design, interpret report data and decide on the next optimization step.
Product stack: Wavelet and Orbit
The company says its product is built around two parts: a proprietary vertical model and an agent engineering framework, designed to answer two questions — how to interpret what is happening in a design flow, and how to keep a long and complex process moving forward.
On the model side, the team has developed an IC design large model called Wavelet. It is responsible for professional understanding and reasoning in chip engineering. According to the company, the model can process context involving netlists, place-and-route, timing and power, and it has data handling capabilities for advanced VLSI process nodes.
On the workflow side, Xinliu Weilan has built an engineering-grade long-horizon process control system called Orbit. The system supports stable, long-duration operation across tools and across stages, along with fault diagnosis and strategy adjustment. It is also designed to keep learning from chip design projects and turn senior engineers’ tacit knowledge into reusable capabilities.
Claimed deployment in sub-10nm projects
The company says its approach differs from projects that remain at the level of tool invocation or proof of concept. Instead, it uses real industrial toolchains and combines models, data, toolchains, process control and chip engineering experience through work in actual industrial environments.
According to figures cited in the article, the system has already completed the full physical design flow in real projects below 10nm and at a scale of nearly 10 million gates. It has continued to optimize PPA — performance, power and area — while maintaining stable and high-quality long-duration operation. The system also supports optimization for advanced devices such as GAA and optimization for ultra-high-dimensional industrial processor microarchitectures.
How the team sees agents in chip design
The article argues that while EDA software has grown stronger over the past decades, chip scale has kept increasing and design flows have become more automated, a critical phase still depends on large engineering teams to fix a wide range of violations before a project can converge on schedule.
The reason, as described in the piece, is that even with abundant server resources and tool licenses that allow many convergence strategies to be tested in parallel, teams still need to interpret reports, locate root causes, modify designs and verify the results. Those tasks continue to depend heavily on the judgment of experienced engineers.
In that context, the team sees AI agents as a natural fit for chip design workflows. The process involves many repetitive iterations, the optimization path varies from project to project, and the result of each step can be checked against explicit engineering metrics. The workflow described in the article starts with understanding the current design state and analyzing the source of a problem, then choosing the next strategy, calling real EDA tools to execute it, and using the new results to judge whether the move worked before entering another round.
From that perspective, the article says companies first need a system capable of carrying complex engineering processes in a highly constrained chip-design setting, linking design goals, domain knowledge, professional tools, process status and verification feedback into a traceable engineering loop.
What the investor and founder said
Dr. Zhang Yong, founder and managing partner of Qigao Capital, said: “Chip design is moving from tool automation toward process intelligence centered on large models and agents. The barrier to entry is high because it requires capabilities across multiple domains. The Xinliu Weilan team combines long-term academic accumulation, industrial delivery experience and real-world validation. We are optimistic about Agentic EDA’s long-term reshaping of the chip R&D paradigm, and we also expect Xinliu Weilan to grow into an important infrastructure company in the global IC agent field.”
The article extends that engineering architecture beyond semiconductors, arguing that chip design is only one branch of complex engineering. In other complex systems, engineers also face changing states, conflicting constraints, fragmented professional tools and verification processes that cannot be replaced by a single generated answer.
That is why Xinliu Weilan treats “generalized intelligent design” as its long-term anchor. Starting from chip design, it aims to build what it calls a generalized intelligent design operating system for complex system design. In the article’s framing, “generalized” means organizing design goals, professional knowledge, engineering tools, operating states and verification mechanisms from different fields into a reusable agent engineering architecture.
Sun Qi said: “We hope to build an IC agent system that is truly usable and trustworthy for design engineers. It should be able to amplify engineers’ efficiency by dozens of times while also turning years of accumulated enterprise experience into system capabilities, so that scarce senior engineering experience no longer becomes the bottleneck in complex projects and human value can move from implementation to innovation.”
The original article was published by the WeChat account Touzijie AI and written by Yu Mengying.