As we celebrate Micron’s Micron’s 30-year anniversary in the automotive industry, we recognize and appreciate that our journey would not have been possible without the close partnerships, collaboration and relationships that have been foundational to our collective success. As part of our celebration, we would like to extend a heartfelt “thank you” from all of us to all the participants in this journey and give a warm welcome to those who are recent acquaintances.
Memory and storage are fueling automotive industry transformation
Who could have known 30 years ago that qualifying a 16-Kbit EEPROM for a powertrain would mark the beginning of the biggest transformation of the automotive industry since the first Model T rolled off the assembly line? More than 100 years later, intelligence has become the new horsepower for the auto industry. The amount of data generated and software complexity in cars has grown exponentially; today’s high-end luxury vehicle contains approximately 100 million lines of code. And this is expected to grow to more than 300 million lines as the industry pushes toward fully autonomous vehicles.
Connected cars generate up to 25GB of data an hour.1 Depending on the architecture of the vehicle, advanced driver-assistance system (ADAS) and autonomous vehicle (AV) sensors can generate between 4TB and 20TB each day.2 This data must be processed in real time so that compute-centric, artificial intelligence (AI)-powered systems can react instantaneously, making modern vehicles the ultimate intelligent edge devices. The amount of memory and storage required to fuel these data-intensive systems has led to a $2.5 billion automotive memory industry,3 making it one of the fastest-growing segments of the semiconductor market.
Today’s average vehicle contains 4GB of DRAM and 256GB of storage. Looking forward, fully autonomous (Level 5) vehicles are projected to support an aggregate memory bandwidth that will exceed 1 terabit per second (Tbps). The total memory, storage and system bandwidth are astounding — and a stark contrast to a time when a 16-Kbit EEPROM was considered a high-density device.
The earliest dreams of self-driving cars
Almost as soon as the first Model T rolled off the assembly line, the world began to dream of a vehicle that would be self-driving. While personally owned vehicles proved to be transformative, over the first 100 or so years after introduction of the mass-produced vehicle, nominal progress was actually made in achieving the vision of a self-driving car. One of the earliest documented cases of an attempt to design a self-driving car was in the 1920s.4 However, it has only been in the past 10 years, with the confluence of multiple technological advancements and efficacy in sensors, compute processing, AI algorithms, memory bandwidth and wireless communication, that the early vision of a self-driving car has come closer to reality.
Today, the mass-produced self-driving car is literally “just around the corner.” Realizing this vision, however, requires processing massive amounts of data from the wide range of sensors that are used to detect, recognize, track and predict the actions of objects on the road. A single high-resolution camera sensor can generate 1.4 GBps of data. A fully autonomous vehicle may have more than 40 sensors — which can generate in excess of 50 gigabytes of data each second. Fully autonomous driving requires extremely high real-time AI compute performance on the order of 500 tera (trillions) operations per second (ToPS).
A quieter revolution in in-vehicle experiences
And while autonomous driving (AD) rightfully captures all the fanfare, it’s important to recognize that ADAS and the enriched cabin are driving the majority of growth in semiconductor content today — well ahead of fully autonomous vehicles. The vehicle interior/cockpit and “in-vehicle experience” (IVE) have become major factors in the consumer’s purchasing decision and an OEM’s brand identity.
Consumers expect increased connectivity, access to information on their fingertips and a seamless smartphone-like user experience, with native in-vehicle support for all their favorite smartphone applications. These apps should be supported on an increasingly larger number and size of high-resolution displays that provide an immersive experience. The innovation in IVE grows unabated as OEMs recognize that the future of self-driving cars will increasingly value a cabin rich with information and passenger experiences as the focus shifts from the actual driving experience to an experience based upon individual taste, entertainment and productivity.
Software code, user personas and 5G
As the industry gains a foothold in delivering mobility as a service (MaaS), it is expected that the consumer’s choice of a transportation service provider will be influenced as much by vehicle aesthetics as continuity in the in-vehicle experience. This again reflects a significant shift for an industry that has historically been based on a “bending metal” mindset to one where “dicing silicon” now establishes brand and identity.
As the software footprint of the automobile expands to over 300 million lines of code, to remain current with the latest updates and enhancements, it will be essential to support over-the-air (OTA) updates. It will also be key to support capabilities such as software as a service (SaaS) where autonomous driving and other features can be enabled via monthly subscription fees and software downloads. For MaaS, user personas will be stored in the cloud and retrieved whenever drivers enter their own vehicles or those of the same brand that they may be renting. To realize these capabilities, high-performance 5G connectivity and system-level security will be paramount — even nonnegotiable. Micron’s unique security solution, Authenta, which provides root-of-trust-based security, was developed specifically to address the needs of this complex edge compute environment.
Securing a safer future in automotive
Ensuring that the hardware and software are both secure and reliable has prompted a keen focus on both security and functional safety. This complex, high-performance software compute platform on wheels requires the utmost in reliability and is going through a complete overhaul of the underlying vehicle architecture. We’re currently watching the evolution from distributed to centralized/zonal-based architectures that will combine infotainment, ADAS and other systems into a single architecture.
Achieving the requisite functional safety (FuSa) levels demands a greater focus on the impact and role of memory in these increasingly complex embedded software platforms. Adoption of ISO 26262-evaluated solutions for both traditional systems on chips (SoCs) and memories will prove to be essential in meeting these FuSa levels. This, of course, is all built on a mindset that aspires to achieve and deliver a zero-defect level of quality.
Planning for many more years of leadership
For 30 years, Micron has been honored to have a front seat in the transformation of the automotive industry. Becoming the top memory supplier to the automotive industry with over 40% global market share5 happened through consistent focus, investment and collaboration. Through this close collaboration, we have established a view of how semiconductors will fuel the data-driven car today and for the next 30 years. An efficient design cycle helps Micron execute new technology with high quality, low cost and a reduced time to scale.
With many trillions of miles traveled on Micron memories, we are at the forefront of the auto industry that is poised to be measured not by the horsepower or acceleration but by the compute performance and in-vehicle experience.
Whether the focus is on safety, security, performance, quality, longevity, innovation or customer support, we are committed to continuing to be the automotive memory leaders for the next 30 years, delivering the industry’s leading portfolio of innovative, auto-qualified DRAM-, NAND- and NOR-based solutions addressing today’s and tomorrow’s requirements.