高瓴出手 AI芯片“卖铲人”汪大祥拿到钱了

Summary 1: Langshi Technology has completed a nearly RMB 200 million Pre-A+ financing round, with participants including Hillhouse Ventures and SenseTime Guoxiang Capital, bringing total funding to nearly RMB 300 million. The company focuses on silicon capacitors for AI chip power delivery, and its products have achieved large-scale deployment in CPU and GPU applications.

1. Product value: Silicon capacitors act like a reservoir close to the water user. By mounting on the surface, attaching to the back, or embedding in the substrate, they sit close to the main chip, shortening the power delivery path and mitigating current fluctuations. Compared with traditional ceramic capacitors, they are smaller, thinner, and easier to integrate.

2. Key data: The mass-produced silicon capacitor achieves a capacitance density of 2μF/mm², with the new generation reaching 3μF/mm². The company has over 20 customers, more than 200 mass-produced models, monthly capacity exceeding 500 million units, and has secured orders worth over RMB 100 million.

Summary 2: For general readers, the most useful insight is how this company turned technology into a business.

1. Founder Wang Daxiang has a background in HiSilicon large-chip design, GlobalFoundries, and NAURA. He started by building a process platform, then developed standard products, winning customers with measured data.

2. The cycle from testing to entering a customer's supply system takes about one to one and a half years. In 2025, the company mainly relies on validation orders; revenue is expected to reach tens of millions of RMB in 2026, showing that hard-tech startups face long validation cycles with volume growth coming later.

Summary 1: Brand owners can learn from Langshi Technology that in the B2B hard-tech market, brand trust comes from verifiable technical metrics and customer success stories.

1. Product positioning: The company targets the power delivery gap in large AI chips, using silicon capacitors placed close to the chip with high-frequency decoupling to solve the path loss issues of traditional ceramic capacitors. By reaching a capacitance density of 2μF/mm² and developing a 3μF/mm² product, it creates a differentiated selling point.

2. Customer validation: Leading customers in CPU and GPU fields have recognized the technology. Multi-terminal substrate-embedded capacitance products have achieved volume delivery and secured orders exceeding RMB 100 million, proving that "entering a major customer's supply chain" is the strongest brand endorsement.

Summary 2: Brand owners should also watch user demand and trend shifts.

1. Demand trends: Rising AI compute power makes current transient fluctuations faster. Customers want capacitors with higher density and smaller footprint, while still designing the power network with ceramic capacitors. Price fluctuations in ceramic capacitors have also amplified the commercial value of silicon capacitors.

2. R&D direction: The company is expanding from silicon capacitors to silicon resistors, silicon inductors, and other silicon-based passive components, while entering optical communications and power applications. The successful adaptation of silicon capacitors for 1.6T optical modules, with high frequency and low insertion loss, represents new opportunity in emerging scenarios.

Summary 1: For sellers, Langshi Technology reveals a high-growth supply chain opportunity: silicon-based passive components for AI chip power delivery are ramping up, with a domestic substitution window.

1. Growth market: AI compute and advanced packaging raise the requirements for power delivery networks. The company first achieved volume delivery in CPUs and GPUs, then expanded into optical communications and power. It has successfully adapted to 1.6T optical modules and is shipping in volume to leading overseas optical module customers.

2. Opportunity signal: Its multi-terminal substrate-embedded capacitance products have secured orders over RMB 100 million, with more than 20 mass-production or validation customers and monthly capacity exceeding 500 million units, showing that once validation with leading customers is passed, order volume can grow rapidly.

Summary 2: The other side is the long cycle and strong synergy risks that sellers need to see.

1. Risk warning: The product validation cycle is about one to one and a half years. In 2025, revenue still comes mainly from million-level validation orders, and only in 2026 is it expected to reach tens of millions. Sellers entering a similar supply chain need to reserve cash flow and patience.

2. Learnable point: Langshi follows a model of "developing a process platform first, then standard products, and jointly validating with wafer fabs, substrate makers, and packaging houses." This collaborative approach can reduce the friction of entering a customer's system.

Summary 1: For factories, Langshi Technology demonstrates the specific production and design requirements of silicon-based passive components from R&D to mass production.

1. Product requirements: The company uses semiconductor processes to build capacitors on silicon substrates, using three-dimensional structures to increase capacitance per unit area. Mass-produced capacitance density reaches 2μF/mm² and the new product reaches 3μF/mm², while making the device smaller and thinner and controlling parasitic resistance and parasitic inductance.

2. Manufacturing barriers: Products must go through manufacturing, packaging adaptation, and reliability validation. Once embedded in a substrate, capacitors are hard to rework, and one failure can affect the entire substrate. Factories need to collaborate with wafer fabs, substrate makers, and packaging houses to validate the complete packaged state.

Summary 2: Factories can also find commercial opportunities here.

1. Volume opportunity: With over 200 mass-produced models of silicon capacitors and silicon resistors, monthly capacity exceeding 500 million units, and orders above RMB 100 million, advanced packaging power devices are proven to be a scalable incremental category.

2. Platform insight: The company drives productization via a "process platform + standard products + measured verification" approach, then extends silicon capacitors to silicon resistors and silicon inductors, forming a silicon-based passive component platform. This is a useful model for factory product line planning.

Summary 1: Service providers can see from Langshi Technology the technical service demand created by AI computing: high-performance chip power delivery networks are shifting from traditional ceramic capacitors to silicon-based passive components.

1. Industry trend: Changes in chip workload cause current demand to fluctuate within very short time spans. Power delivery networks need proximity to the chip and high-frequency decoupling. Advanced packaging raises power delivery requirements, creating an opportunity for domestic substitution of silicon-based passive components.

2. Customer pain points: Customers need high capacitance density to save space, low parasitic resistance and inductance, and reliability after embedding into the substrate. Validation scope has extended from individual devices to the complete packaged state consisting of substrate and main chip.

Summary 2: Solutions and progress that service providers can focus on.

1. Technical solution: 3D silicon capacitors use deep trench processes to increase capacitance density and are placed close to the main chip via surface mounting, backside mounting, or substrate embedding. They can also be combined with ceramic capacitors to design the power delivery network.

2. Progress: The product validation cycle is about one to one and a half years, and both product paths have passed customer validation. The company has over 20 mass-production or validation customers, successful adaptation to 1.6T optical modules, and is building a silicon resistor and silicon inductor platform.

Summary 1: Platform operators can learn from Langshi Technology's collaborative supply chain approach and consider how to build a validation and service ecosystem around advanced packaging components.

1. Demand insight: What customers ultimately purchase must meet electrical performance, withstand subsequent processing, and maintain consistency. Validation extends to the complete packaged state of capacitors, substrate, and main chip, requiring collaboration among wafer fabs, substrate makers, and packaging houses.

2. Platform approach: Langshi chose to develop a process platform first, then standard products, using measured data to verify capacitance density and other performance. Platform operators can prioritize validation data, reliability, and delivery capability as key service offerings.

Summary 2: Platform operators should also watch risk and market direction.

1. Risk: The cycle from testing to entering a customer's supply system is about one to one and a half years. Langshi's 2025 revenue is still mainly from validation orders at the million level. Platforms entering this field must manage the long timeframe.

2. Opportunity: AI computing and advanced packaging raise power delivery requirements across CPU/GPU, optical communications, and power markets. Capital players like Hillhouse and SenseTime Guoxiang are entering, indicating the sector is heating up.

Summary 1: The Langshi Technology case shows how AI computing drives upstream passive component industry upgrades and forms a hard-tech innovation model.

1. New industry dynamic: Silicon capacitors replace ceramic capacitors using semiconductor processes, achieving miniaturization and high capacitance density to meet CPU/GPU power delivery needs. Applications have expanded from consumer electronics to AI large chips and further into optical communications and power.

2. Business model: Starting with silicon capacitors, the company extends toward a silicon-based passive component platform covering silicon resistors and silicon inductors. It develops a process platform first, then standard products, and goes from validation orders to volume delivery through joint validation with wafer fabs, substrate makers, and packaging houses.

Summary 2: From a research perspective, this case also reveals industry issues that deserve attention.

1. Long validation cycle: The product takes about one to one and a half years from testing to entering a customer's supply system. Once embedded in a substrate, rework is difficult, so reliability validation extends to the complete packaged state.

2. Implications: Industrial policy and supporting funds should focus on cross-stage collaborative validation capabilities, key equipment procurement, and long-term R&D investment rather than only short-term shipment volume. Domestic substitution in silicon-based passive components is accelerating, requiring more upstream-downstream joint validation earlier in the cycle.