On January 9, 2024 local time, Intel announced at the CES 2024 exhibition that it plans to promote the company's "AI Everywhere" strategy to the automotive market, including the acquisition of Silicon Mobility, a fabless chip design and software company specializing in smart electric vehicle (EV) energy management SoC. At the same time, Intel also announced a new family of AI-enhanced software-defined automotive system-on-chips (SDV SoCs), and Zeekr will be the first automaker to adopt a new SoC to deliver generative AI-driven in-cabin experiences.
Acquisition of Silicon Mobility
Founded in 2015 and headquartered in Sophia Antipolis, France, and with offices in Silicon Valley, Germany, Japan and China, Silicon Mobility is an automotive chip design and software company that designs chips that are key components in controlling the powertrain of electric vehicles.
Silicon Mobility claims to be the inventor of the Field Programmable Control Unit (FPCU), a unique semiconductor architecture that removes the biggest hurdles encountered in the process of mass vehicle electrification. The OLEA FPCUs it offers are designed to be flexible and customizable, designed to process critical information faster with 100% availability and accuracy.
In Silicon Mobility's view, traditional automotive microcontrollers, which were the backbone of all automotive control systems, proved insufficient and unable to adapt to changing environments, including optimization for hot engine control and its many sensors, actuators, complex timing, and low response times and real-time processing needs. Recognizing the urgent need for a revolution in automotive semiconductors, Silicon Mobility reimagined the foundations of electric, autonomous, and connected vehicle control systems, building a team of experts specializing in system-on-chip design, embedded software, and power electronics systems, and introducing FPCUs. According to Silicon Mobility, the platform, unified in a design-for-design and development environment under one model, is a game-changer.
According to reports, FPCU is a breakthrough achievement in the field of automotive control technology, which has become the first control chip to provide a hybrid architecture heterogeneous processing platform by combining the ** processing unit (CPU) with the flexible logic unit (FLU), digital signal processing (DSP) accelerator and mathematical coprocessor. The chip can efficiently distribute tasks across its different processing resources. CPUs, with their versatility, can handle general tasks, and mathematical coprocessors are responsible for fast calculations, from trigonometric functions to matrix multiplication. Programmable logic (FLU) and DSP units intervene in real-time, compute-intensive tasks. The result is that the inverter and motor can be controlled with an amazing 250kHz magnetic field-oriented control with zero CPU intervention.
In 2016, Silicon Mobility launched its first FPCU product, the OLEA T222, followed by the second-generation FPCU product, the Olea U FPCU family, in 2023, which is designed to meet the current and future needs of the automotive industry for real-time, secure, and reliable control applications and functional grouping, and also introduces advanced edge AI-based control algorithms to protect against post-quantum cybersecurity threats.
Intel is taking a 'full truckload' approach to solving the industry's biggest challenges. "AI solutions that drive innovation across the entire vehicle platform will help the industry transition to electric vehicles," said Jack West, vice president and general manager of Intel Automotive. The acquisition of Silicon Mobility aligns with our sustainability goals while addressing the industry's critical energy management needs. ”
Intel also said in a press release that Silicon Mobility's SoCs are equipped with industry-leading accelerators designed for energy delivery and co-designed with highly advanced software algorithms to significantly improve vehicle energy efficiency. Silicon Mobility's technology portfolio will extend Intel's automotive presence from high-performance computing to intelligent and programmable power devices.
It is unclear what the deal was for Intel's acquisition of Silicon Mobility. However, according to the official website, Silicon Mobility announced in 2018 that it had completed a $10 million Series B funding round led by Capital-E and Cipio Partners. Obviously, Intel's acquisition** could be in the hundreds of millions of dollars.
It is important to note that the acquisition has not yet been completed, as the necessary approvals are still to be obtained.
Support for software-defined vehicles
In Intel's view, the future of the car is defined by software, and in the future, every function is an application running on a high-performance computing platform.
Historically, automakers have relied on microcontroller-based architectures that require more than 100 microcontrollers and more than a mile of cable to deliver modern experiences and features. With the development of electric vehicles, more and more components and wiring harnesses are required, and the answer to this increasingly complex problem is software-defined vehicles (SDV).
The SDV architecture operates automotive features and functions as software workloads that can be consolidated on high-performance computing platforms. This makes vehicle functions and systems easier to develop, scale, and upgrade, and simplifies cloud integration.
Intel blazed a new trail in SDKV development, pioneering software-defined architecture in the data center, and now this expertise provides a flexible path for SDV transformation. Intel Automotive enables a single Intel system-on-chip (SoC) platform to be partitioned and stitched together to support multiple concurrent automotive workloads and support a range of SDV use cases. And it does so without impacting critical workloads or security measures.
Intel's SDV demo, which runs up to 12 advanced workloads on a single system, enables automakers to consolidate their ECU architectures and achieve greater efficiency, manageability, and scalability. Various automotive applications, such as Rightware's Kanzi One Instrument Cluster, Thundersoft's Surround View and Driver Monitoring apps, Android Automotive apps, and the Over the Air update feature, are all available in the QNX RTOS with Ubuntu and Android Virtual Machines hypervisor platform.
The Intel Automotive SDK demo also runs multiple operating systems simultaneously with different workloads and hybrid criticality, all on the same piece of silicon. Linux virtual machines (VMs) run clusters of digital instruments. The automotive RTOS runs additional security software to verify the integrity of the dashboard display. The Android virtual machine runs navigation, infotainment, and passenger seat gaming solutions. The Linux container in the Android VM runs real-time AI inference to analyze the data, while the second Linux VM runs AAA games and conversational AI.
Intel's Open Platform Brings AI PC Experiences to the Car
Intel's January 9, 2024, CES 2024 presentation also introduced a family of AI-enhanced software-defined vehicle system-on-chips (SDV SoCs) with AI acceleration capabilities from Intel's AI PC roadmap to enable the most desirable in-vehicle AI use cases, such as driver and passenger monitoring.
Intel also showcased 12 advanced workloads based on the SDV SoC system, including generative AI, e-mirrors, HD** conference calls, and PC gaming, which run simultaneously on multiple operating systems, including a mix of key use cases. The demo also showcases how automakers can integrate traditional electronic control unit (ECU) architectures to improve efficiency, manageability, and scalability, while integrating their own custom solutions and AI applications.
We're bringing AI PCs into cars," and "However, we can't simply put a PC in a car. As I mentioned, we know that automakers need to be able to redesign their models to make them software-defined. "Intel's AI-enhanced SDV SoC combines the best of AI PCs and Intel data center technologies needed to support true software-defined vehicle architectures," said Jack West. ”
"Today, Intel SoCs are used in more than 50 million vehicles, powering infotainment, displays, digital instrument clusters, and more." Tomorrow, Intel's expanded AI-enhanced 'full vehicle' roadmap will propel the industry toward a more scalable, software-defined, and sustainable future." ”
Zeekr is the first to bring Intel-powered GenAI experiences to the next generation of electric vehicles
Geely's new energy vehicle brand Zeekr will be the first OEM to use Intel's new SDV SoC series.
Andy An, President of Geely Holding Group and CEO of Zeekr***, explained how forward compatibility of Intel systems combined with Intel AI Acceleration will allow Zeekr to continuously expand and upgrade its services to enable the next-generation experiences customers demand, such as generative AI-based voice assistants.
Open standards are the key to the success of the industry
To drive a faster and smoother transition to electric vehicles and sustainable SDVs, Intel and SAE International announced the formation of a committee to provide automotive standards for vehicle platform power management (J3311). Intel will chair the committee.
Inspired by the proven power management technologies in the PC industry's ACPI standard, the new SAE standard will accelerate progress by adopting and enhancing the PC industry's advanced power management concepts to help all electric vehicles become more energy-efficient and sustainable.
The Standards Committee currently includes industry representatives from Stellantis, HERE, and Monolithic Power Systems (MPS). The committee is open to more industry involvement, with the goal of delivering the first draft standard within 12 to 18 months.
Intel pushes for an open automotive chiplet platform
Intel also announced a commitment to deliver the industry's first UCIe-based open chiplet platform for SDV. Intel will work with IMEC to ensure that Intel's advanced chiplet packaging technology meets the stringent quality and reliability requirements required for automotive use cases.
The move underscores Intel's commitment to becoming the first automotive vendor to support the integration of third-party chiplets into its automotive products. This gives OEMs the freedom to integrate custom chiplets into Intel's roadmap products at a fraction of the cost of a fully custom SoC. The ability to mix and match chiplets further eliminates the risk of vendor lock-in and promotes a more scalable software-defined architecture.
Intel IFS will manufacture automotive chips for Valens
It is worth mentioning that at the same time, Intel also announced that its Foundry Services (IFS) division will use its advanced process technology to produce a new generation of MIPI A-PHY chips for VALENS foundry to meet market demand. This collaboration will further deepen the partnership between VALENS and IFS.
According to reports, A-PHY, led by the MIPI Alliance, is the first automotive industry standard developed for in-vehicle high-speed sensor connectivity. Since its publication in 2020, it has attracted an ecosystem of companies that design products based on this technology. The MIPI A-PHY standard is rapidly evolving into one of the most powerful and reliable long-range connectivity standards for ADAS, autonomous driving, in-car entertainment, and other automotive applications. This standard dramatically simplifies the installation and deployment of lidar, mmWave radar, and cameras in vehicles, facilitates autonomous driving, and improves the link to high-resolution and infotainment displays. Valens, a key contributor to the standard, is the first company on the market to offer A-PHY compliant chips with the VA7000 series.
As part of the strategic partnership, Valens will use Intel's leading process technology to produce its second-generation A-PHY standard chips, the report said. Intel's ecosystem and advanced process technologies will enable significant reductions in cost and power for this chipset family while achieving the highest performance and perfect electromagnetic compatibility (EMC). With a significant increase in bandwidth and other architectural requirements, this unique set of capabilities ensures that link performance and passenger safety are maintained at all times.
"One of the goals of Intel Foundry Services is to identify transformative technologies and leverage Intel's breakthrough process technologies and global ecosystem to accelerate innovation and unlock new opportunities for our customers," said Stuart Pann, senior vice president and general manager, Intel Foundry Services. The MIPI Alliance's A-PHY standard chip is a clear fit, as it is a next-generation technology with the potential for broad market adoption to help improve automotive performance and safety, and further advance the vision of autonomy. ”
The market expects that this collaboration will play an important role in promoting the development of chiplet architecture in the automotive industry. With its reliance on high-end packaging integration and the use of standardized interfaces such as UCIe, the chiplet architecture will be key to standardizing the MIPI Alliance A-PHY standard and the best way to meet the automotive market's requirements for high reliability, high performance, and low latency.
Editor: Xinzhixun-Rogue Sword.