Stock code: 688521 Stock abbreviation: VeriSilicon Announcement No.: 2025-012

VeriSilicon Microelectronics (Shanghai) Co., Ltd

Announcement on the Voluntary Information Disclosure of Recent Business Results

The board of directors and all directors of the company guarantee that there is no false record, misleading statement or material omission in the content of this announcement, and assume legal responsibility for the authenticity, accuracy and validity of its content in accordance with the law.

VeriSilicon Microelectronics (Shanghai) Co., Ltd. (hereinafter referred to as the "Company" or "VeriSilicon" or "VeriSilicon") is an enterprise that provides customers with platform-based, all-round, one-stop chip customization services and semiconductor IP licensing services based on its own semiconductor IP. The company has its own controllable graphics processing unit IP (GPUIP),

Neural Network Processor IP (NPU IP), Video Processor IP (VPU IP), Digital Signal Processor IP

(DSP IP), Image Signal Processor IP (ISP IP), and Display Processor IP (DisplayProcessorIP)

These six types of processor IP, as well as more than 1,600 digital-to-analog hybrid IP and RF IP. With a long-term retention of 30%

VeriSilicon has always been at the forefront of technological development and has been leading the AI industry in terms of high R&D investment

domain has a wealth of achievements. For example, VeriSilicon's world-leading NPUIP has 142 AI in 82 customers

More than 100 million AI chips integrated with VeriSilicon's NPU IP have been shipped, and DeepSeek is now running.

VeriSilicon's GPUs and GPGPU-AI IP for high-performance AI computing have also been re-architected multiple times around the world

Licensed, and used in a wide range of high-performance computing products. VeriSilicon's GPU IP is also outstanding, being widely adopted in data centers, automotive electronics, wearables, PCs, and more, with more than 2 billion customer chips with built-in VeriSilicon GPUs shipped worldwide.

Based on its own IP, the company has a wealth of software and hardware chip customization platform solutions for artificial intelligence (AI) applications, covering real-time online (always-on) lightweight spatial computing devices such as smart watches and AR/VR glasses, high-efficiency end-side computing devices such as AIPC, AI mobile phones, smart cars, and robots, as well as high-performance cloud-side computing devices such as data centers/servers. In addition to serving IC design customers, about 40% of VeriSilicon's revenue in recent years comes from non-IC devices such as system vendors, Internet companies, cloud service providers, and car companies

Approximately 80% of the company's chip customization services business revenue is based on advanced processes of 14nm or below. Based on the company's unique chip design platform as a service (SiPaaS) business model, the company's main business applications include consumer electronics, automotive electronics, computers and peripherals, industry, data processing, Internet of Things, etc., and the main customers include chip design companies, IDMs, system manufacturers, large Internet companies, cloud service providers, etc.

In recent years, the automotive industry is undergoing a transformation of "electrification, intelligence, unmanned, and networking", and the era of intelligent travel has arrived. Focusing on the fast-growing field of autonomous driving, the company's automotive-grade high-performance autonomous driving system-on-chip (SoC) design platform has been verified and successfully implemented in customer projects. Based on VeriSilicon's Silicon Design Platform-as-a-Service (SiPaaS) business model, the platform can provide powerful technical support for scenarios requiring high-performance computing such as autonomous driving and intelligent driver assistance systems (ADAS).

VeriSilicon's chip design process has been certified by the ISO26262 Automotive Functional Safety Management System, providing one-stop customization services for automotive chips that meet functional safety requirements for global customers in terms of chip and IP design and implementation, software development, etc. Combined with the company's rich automotive-grade IP portfolio and a complete autonomous driving software platform framework, VeriSilicon can provide customers with full-process support from chip design, verification to automotive certification, including safety requirements analysis, architecture design and certification support.

VeriSilicon's automotive-grade, high-performance autonomous driving SoC design platform adopts a flexible and configurable architecture, supporting multiple coprocessors such as high-performance multi-core central processing units (CPUs), image signal processors, video codecs, and neural network processors to work together efficiently, with optional built-in ASILD-grade functional safety islands, and support for high-speed and high-bandwidth storage subsystems, with excellent data throughput and real-time processing capabilities. The platform is also optimized for advanced automotive processes, including 5nm and 7nm, with superior power, performance, and area (PPA) characteristics.

At present, global car companies are actively deploying ADAS and autonomous driving technologies to seize the next development highland of the automotive industry, and with the help of related technology accumulation, they are further moving towards the application of robots. ADAS and autonomous driving chips face challenges such as long design cycles, low single-chip (large chip) yields, limited production, and difficulty in scaling computing power, and the adoption of chiplet technology can effectively solve these problems, and has been used in the automotive industry

Consensus: In October 2023, the automotive ecosystem will hold its second automotive chiplet conference at IMEC.

The consensus is that "Chiplets will be part of the car of the future, no matter what".

In November 2023, Renesas announced its fifth-generation automotive chip strategy, making it clear that it will adopt chiplets

architecture, and provide SoC, MCU and other series of products based on Chiplet architecture in 2027; October 2024

This month, Bosch and Tenstorrent announced a collaboration to develop a standardized platform for chiplets for automotive chips. Chiplet has become the consensus of the world's high-end intelligent driving, and it is also one of VeriSilicon's strategic development directions.

The company began to develop chiplet technology five years ago. VeriSilicon is currently working on "IP as a chip."

(IPasaChiplet)", "ChipletasaPlatform" and "Platformas an Ecosystem" as the guiding principles for action, starting from interface IP, chiplet chip architecture, advanced packaging technology, AIGC and smart mobility solutions, etc., continue to promote the development and industrialization of the company's chiplet technology and projects. Expand a broader market space through competitiveness upgrading. At present, VeriSilicon has achieved leadership in the two major tracks of chiplet-based generative AI big data processing and high-end intelligent driving, and is advancing the R&D projects of chiplet architecture, chip platform for intelligent driving system and AIGC high-performance computing, which have achieved excellent phased results.

Based on the company's automotive-grade certified chip design process, automotive-grade IP, and complete software services, VeriSilicon has successfully provided customers with customized autonomous driving chips based on advanced automotive-grade processes. At present, the R&D and industrialization of the chiplet platform for high-end intelligent driving are being promoted. In the future, VeriSilicon will continue to deepen technological innovation in the automotive field and help the smart automotive industry achieve a higher level of safety and intelligence.

The announcement is hereby made.

Board of Directors of VeriSilicon Microelectronics (Shanghai) Co., Ltd

April 8, 2025