In its debut at Auto Shanghai, Intel today unveiled its second-generation AI-enhanced SDV (software-defined vehicle) SoC (system-on-chip) that it says uses the automotive industry’s first multi-process-node chiplet architecture. The new SoC is engineered to meet the growing demands of intelligent, connected vehicles by providing its auto customers with scalable performance, advanced AI capabilities, and optimized cost efficiency.

“Intel is redefining automotive compute with our second-generation SDV SoC, combining the flexibility of chiplet technology with our proven whole-vehicle approach,” said Jack Weast, Intel Fellow, Vice President and General Manager of Intel Automotive.

The new chiplet architecture helps enable automakers to tailor compute, graphics, and AI capabilities to their needs while reducing development costs and accelerating time-to-market. By combining best-in-class silicon for each functional block, the architecture offers up to ten times the AI performance for generative and multimodal AI (vs. an MBL i7-13800HAQ CPU+GPU with Turbo off), up to three times the graphics performance for richer HMI (human-machine interface) experiences (vs. MBL i7-13800HAQ with Turbo off), and 12 camera lanes for increased camera input and image processing capabilities.

Intel says its growing list of ecosystem partners reflects the company’s expanding influence in the automotive space. At Auto Shanghai, Intel’s customer announcements included new strategic collaborations with leading automotive innovators ModelBest and Black Sesame Technologies to further expand its automotive ecosystem and accelerate innovation in AI-powered cockpits, integrated ADAS (advanced driver-assistance system) solutions, and energy-efficient vehicle compute platforms.

“Together with our partners, we’re solving real industry challenges—from energy efficiency to AI-driven experiences—to make the software-defined vehicle revolution a reality for all,” added Weast.

ModelBest’s GUI Intelligent Agent, powered by Intel’s SDV SoC and Arc graphics for automotive, brings an on-device LLM (large language model) to enable offline, AI-enhanced voice control and personalized interaction, even without network connectivity. The agent enhances voice interaction by accurately understanding natural language in complex situations, ensuring an intuitive cockpit experience. This collaboration leverages ModelBest’s success with Intel’s AI PC acceleration program, optimizing AI for an optimal out-of-the-box experience on Intel’s automotive platform.

Black Sesame is developing a central compute platform that fuses ADAS and immersive cockpit experiences into one seamless, energy-efficient system with stable, high-speed, low-latency connections by combining its autonomous driving technology with Intel’s SDV SoC and Intel Arc graphics for automotive. The new collaboration builds on the company’s innovations that started with Huashan Series high-computing power SoCs for autonomous driving and later Wudang Series cross-domain SoCs to address more diverse and sophisticated demands for advanced functionalities of intelligent vehicles.

Intel says that these announcements mark a major milestone in the company’s automotive journey, underscoring its commitment to delivering the compute power and flexibility required for future mobility and helping drive the transformation toward software-defined vehicles. They build on the expanded product portfolio and new partnerships the company unveiled at CES 2025. There, Intel said it now offers a whole-vehicle platform, including high-performance compute, discrete graphics, AI, power management, and zonal controller solutions alongside the Intel Automotive Virtual Development Environment co-developed with Amazon Web Services.

“Our whole-vehicle approach, combined with cloud integration, delivers a complete solution that drives down total cost of development and deployment while empowering automakers to build the future of mobility faster, more efficiently and more profitably,” said Weast, at CES.

Intel says its whole-vehicle platform reduces the inefficiencies of traditional fragmented approaches to vehicle architecture. Supporting the platform, the company introduced the availability of an ACU (adaptive control unit) designed for electric vehicle powertrains and zonal controller applications. It calls the ACU U310 a new kind of processing unit that supports the consolidation of multiple real-time, safety-critical, and cybersecure functions, applications, and domains into a single chip.

When used in an electric vehicle powertrain, the ACU U310 supports advanced algorithmic solutions that reduce vehicle energy demand from the battery, automatically adapting high voltage and control frequencies to individual driver styles and road conditions. The ACU reduces cost per kilowatt and enhances energy efficiency, allowing the vehicle to reclaim up to 40% of the powertrain system energy losses, delivering a 3-5% efficiency boost during the Worldwide Harmonized Light Vehicles Test Procedure. This is said to translate to increased range, faster charging, and a more responsive driving experience while significantly reducing per-vehicle bill of materials (BOM), electric motor size and battery costs compared to traditional approaches.

Karma Automotive announced at CES support for Intel’s ACU, showcasing an Intel co-branded inverter featuring optimal pulse pattern control algorithms to improve efficiency and enable four unique driving profiles and includes innovative features like torque ripple reduction and range boost. Last month, the OEM revealed that its Kaveya super coupe, which will arrive in 2027, and the Ivara GT-UV, coming in 2028, will both incorporate an SDV architecture developed in collaboration with Intel Automotive.