1. Google Chain Fever: From Technological Breakthroughs to Industry Resonance
Recently, the Google Cloud ecosystem has been continuously generating excitement, with significant technological breakthroughs translating into substantial impacts on the capital markets. The stock price of semiconductor giant Broadcom surged by 12% in a single day, nearing its historical high market capitalization. This market reaction essentially validates the commercial value of Google’s self-developed chips and AI computing power systems.
From a technical core perspective, Google’s latest release, the Gemini 3 model, has exceeded performance expectations, showcasing exceptional computational efficiency during large model training with its self-developed TPU (Tensor Processing Unit) chips. Additionally, Google’s introduction of OCS (Optical Circuit Switching) technology allows for dynamic customization of network topology, significantly reducing operational costs and energy consumption for ultra-large-scale computing clusters. These two technological breakthroughs not only solidify Google’s competitive advantage in the AI computing power domain but also send a clear signal to the industry — self-developed ASICs (Application-Specific Integrated Circuits) have become a key pathway for tech giants to overcome computing power bottlenecks and build differentiated competitive barriers. This is expected to accelerate the pace of self-developed chip layouts for major internet companies like Meta, Amazon, and Microsoft, propelling the global ASIC industry chain into a period of accelerated growth.
2. Domestic Substitution: From Breakthroughs of “0 to 1” to Volume Expansion of “1 to 10”
Against the backdrop of a clear global AI industry trend, AI computing power has become one of the core areas of technological competition between China and the United States. In the domestic market landscape, NVIDIA previously held over 90% of the AI hardware computing power market share, posing significant supply chain security risks. Currently, driven by top-down policy guidance and industrial demand, the process of domestic AI computing power substitution is accelerating significantly, with the potential to unleash enormous market elasticity.
After a year of supply chain adjustments and technological breakthroughs in 2025, the domestic ASIC industry chain has achieved a critical breakthrough from “0 to 1”: on one hand, the chip design phase has realized autonomous control of core IP, with significantly improved tape-out success rates; on the other hand, manufacturing capacity is gradually being released, with advanced process yields at foundries like SMIC and Hua Hong Semiconductor continuously optimizing, laying the foundation for the large-scale production of domestic ASIC chips. Based on the assessment of industry cycles and technological maturity, the domestic ASIC industry chain is now equipped to transition from “technology validation” to “large-scale commercial use,” and is expected to enter an accelerated volume expansion phase of “1 to 10” in the next 1-2 years, becoming a core growth driver for semiconductor domestic substitution.
3. Overview: Key Links in the Industry Chain
Companies in the domestic ASIC industry chain with technological barriers and mass production capabilities are detailed as follows:
1. Chip Design Phase: Customization Capabilities Drive Order Explosion
Canxin TechnologyCore Business : A leading ASIC design service provider in China, focusing on high-performance chip customization, covering core scenarios such as AI training/inference chips and data center acceleration chips, and has established stable partnerships with several top AI companies.Competitive Barriers : Possesses full-process service capabilities from front-end design to back-end tape-out, with a wealth of successful cases accumulated in FinFET advanced process nodes, enabling rapid response to customer demands for high-performance, low-latency customization.Latest Developments : Achieved total operating revenue of 468 million yuan in Q3 2025, with continuous revenue growth in the first three quarters, total assets reaching 1.677 billion yuan, and a solid financial structure supporting large ASIC customization projects.Chipone TechnologyCore Business : Focused on one-stop chip customization services, specializing in AI cloud-side/edge ASIC chip design, with clients including global tech giants like Samsung, Google, Baidu, and Tencent. In the first half of 2025, AI computing power-related revenue accounted for 52% of total revenue.Competitive Barriers : Owns six types of controllable processor IP and over 1,600 mixed-signal IPs, ranking first in China’s semiconductor IP licensing market share, achieving successful tape-out of 5nm FinFET chips, with cumulative shipments of nearly 200 million NPU IPs.Latest Developments : Orders have shown explosive growth, with new signed orders for ASIC design exceeding 700 million yuan in Q2 2025, a 700% increase quarter-on-quarter; as of September 11, new signed orders in Q3 reached 1.205 billion yuan, with AI computing power-related orders accounting for 64%, and 81% of orders expected to convert to revenue within a year.Aojie TechnologyCore Business : Focused on cellular baseband technology, specializing in RF and baseband ASIC chip development, with main products covering 4G Cat.1/Cat.4 IoT chips, achieving large-scale commercial use in automotive networking and industrial interconnect fields.Competitive Barriers : Possesses integrated design capabilities for RF and baseband, with products known for high integration and low power consumption, having passed rigorous validation in automotive scenarios, entering the supply chains of mainstream automakers such as Chery, Chang’an, and Dongfeng Nissan.Latest Developments : Revenue in the first half of 2025 increased by 14.67% year-on-year, with shipments of 4G Cat.1 chips surging over 50% year-on-year, and Q2 shipments increasing by 40% quarter-on-quarter, with overseas market demand becoming a significant growth engine.
2. Wafer Manufacturing Phase: Capacity and Process Drive Substitution
SMICCore Business : The largest and most advanced wafer manufacturing enterprise in China, providing foundry services from mature to advanced processes for domestic ASIC chips, serving as the core hub in the AI chip manufacturing phase.Competitive Barriers : Has achieved mass production of 14nm FinFET processes, continuously breaking through in mid-to-high-end process fields such as 28nm and 22nm, meeting the advanced process requirements for high-performance ASIC chips, with capacity scale and yield levels ranking first in China.Industry Value : As the “capacity cornerstone” of the domestic ASIC industry chain, its process iteration rhythm directly determines the mass production efficiency of domestic chip design companies, and it has established long-term foundry cooperation with several AI chip companies.Hua Hong SemiconductorCore Business : A leading wafer manufacturing company known for its specialty processes, focusing on nodes from 0.35μm to 55nm, with global competitiveness in embedded non-volatile memory and power devices, providing stable capacity for mid-to-low-end ASIC chips.Competitive Barriers : Self-developed super junction MOSFET technology breaks foreign monopolies, with mature process yields stable at industry-leading levels, meeting the mass production needs of ASIC chips for IoT and automotive electronics scenarios.Latest Developments : In response to market demand, it is advancing towards advanced nodes, having initiated R&D for 28nm and 22nm processes, and will further cover mid-to-high-end ASIC chip manufacturing needs, forming a complementary process pattern with SMIC.
4. Risk Warning
Industry demand may fall short of expectations : If global AI industry growth slows or domestic companies’ computing power procurement demand decreases, it may lead to order conversion for design companies like Chipone Technology and Canxin Technology falling short of expectations, affecting the growth rhythm of the industry chain;Intensifying industry competition : With NVIDIA launching the Rubin architecture inference chip to counterattack the ASIC market, coupled with an increase in new domestic entrants, it may trigger a price war, squeezing profit margins for companies;Technological iteration risks : TSMC and Intel have initiated plans for 2nm and below processes; if domestic companies lag in advanced process R&D, it may widen the technological gap with international giants and delay the domestic substitution process;Supply chain dependency risks : Companies like Chipone Technology still rely on external cooperation in areas such as high-speed SerDes IP; if supply chain fluctuations occur, it may impact the design progress of ASIC chips.Disclaimer: The content of this media is for recording and sharing purposes only and should not be used for commercial purposes. All content is sourced from the internet. If any text, images, or materials infringe on rights, please contact for modification or deletion. The content of the article does not constitute any investment advice.