The future of the automotive industry is software-led

The past 20 years have generally been good for carmakers, but the next 20 years will likely be very different as software begins to control and shape a vehicle’s experience and development more visibly. A key element of software-defined vehicles (SDVs) is the ability for users to fine-tune various levels of personalization, such as comfort, convenience, safety, and performance. Consequently, our future cars will be able to adapt and improve post-sale, retaining or even increasing in value over time, as opposed to remaining a traditionally depreciating hardware asset.

Every automotive original equipment manufacturer (OEM) is somewhere on its SDV journey with interesting offerings found in digital cockpits, driver assist, and electric vehicle (EV) features. Each is running a major cultural and organizational transformation to become a fast-paced technology company that can challenge the conventions of a century-plus-old industry and expand to unfamiliar business models.

In particular, the automotive product-development ecosystem (research and development [R&D], advanced technologies, IT, and engineering) must pivot in unison to a software-led mindset. It must excel at building software-driven features that redefine the brand and not allow hardware to limit the experience. Ultimately, the goal is to create a vehicle with the capacity for continuous upgradability.

On-the-ground incumbent carmakers are “climbing a mountain” as they invest in a long-haul software-driven transformation.¹ Most notably, revenue is hardware-based; so product definition, R&D, manufacturing, and supply chain still live in a world of parts and components. Incumbents are also saddled with legacy organizational structures that make scaling decisions across multiple brands costly and complex. Finally, for safety reasons, the automotive sector has an entrenched mindset that shipping software post-production is in direct opposition to a stable product.

Nevertheless, car brands have put into motion ambitions and strategies that emphasize a need to dominate and gain control over their SDV destiny. For example, some automakers have elected to create separate software R&D entities to accelerate the transition. Others are in the midst of digital transformations that span manufacturing to engineering and direct-to-consumer sales models. Each journey will differ depending on collaboration with existing suppliers and an open marketplace ecosystem, encompassing software, silicon, and a diverse range of technology partnerships.

While all OEMs have outlined their desired positions in the automotive software landscape by 2030, current economic conditions have affected most of the action plans. There is also genuine uncertainty concerning the willingness of different customer segments to pay for one-time or recurring software features. That being said, regardless of the scope or pace of the transformation, becoming a prominent player in the SDV arena is recognized as a prerequisite for achieving competitive success. One should also not discount the fact that vehicle self-awareness is about not only revenue generation but also self-healing, as over-the-air (OTA) updates also target preventive fixes significantly reducing cost of quality.

We have seen four priorities emerge from our conversations with OEMs, tier 1 suppliers, and technology companies as they work to not only get ready for a significant SDV push but also compete with a new class of product—one that must conform to unique schedule, scope, and resource constraints.

Software-led product development mindset

Simultaneously achieving the adaptability of a software or technology startup, combined with the opportunities and influence of a global OEM, and delivering robust software capabilities, represent the complexities faced by incumbents in the automotive industry. While there won’t be a single path to achieving these objectives, we have identified a common set of success factors for OEMs, tier 1 suppliers, and technology companies:

• Implement a business model where software becomes the primary source of cost and value; transform the traditional hardware-based approach; and redefine sales strategies, dealer interactions, and direct-to-consumer pricing.
• Initiate product design with a focus on features as the benchmark, managing software and hardware cohesively to enhance the overall car experience rather than financing the investment by vehicle model preference.
• Adopt life cycle costing for software to shift system complexity from monolithic applications to micro services, enriching the automotive experience and including more hardware capacity upfront than the minimum to run the software at vehicle launch.
• Transition from a software R&D startup to production-ready, adhering to the rigorous “V-model,” automotive regulations and safety requirements while gradually integrating cloud-based V-model components.
• Establish and manage robust alpha and beta test capabilities and communities in compliance with regulatory and safety standards, enabling customers to participate in field beta testing to minimize surprises in the initial version of feature releases.
• Center development and release processes on features closely linked with reliable OTA update campaigns, minimizing the need for in-person servicing and ensuring clear product-led ownership and governance of feature definition; features are not only driving experiences but also the “health” of the vehicle, regardless of life cycle states (design, manufacturing, in-field or in-service).
• Develop software that is decoupled from dependencies as much as possible from the exact versions of other hardware, configurations, calibrations, and software in the system.
• Develop the ability to deploy software efficiently, whether at the end of the production line, at charging stations or service departments, or primarily through OTA updates, ensuring scalability across different brands and product variants.

Quality needs to guide the ongoing transformation

In the context of an automotive OEM or tier 1 supplier, customers have high expectations regarding the availability, predictability, and reliability of functions and services. Quality management and software bug resolution become even more crucial due to the safety-critical mandate that must be upheld. Utilizing virtual environments for verification and validation, along with phase-specific software key performance indicators, can aid in preventing recalls, enhancing code coverage, and minimizing the blast radius. Other quality-related factors contributing to success include the following:

• Design and deploy advanced hardware and software systems emphasizing fault tolerance and resilience.
• Utilize high-quality components, incorporate redundancy when needed, and apply rigorous testing and validation processes to ensure consistent performance under varying conditions.
• Implement quality management to accommodate new software platforms and legacy hardware.
• Cover the entire product life cycle in quality activities, from concept and development to product enhancement, customer satisfaction monitoring, external standards and regulations shaping, supplier quality evaluations, embedded quality, and launch stabilization.
• Utilize new low-latency input channels as a proxy for the voice of the customer, incorporating complaints, vehicle data, and user data monitoring.
• Establish efficient systems engineering management to address hardware and software integration end to end (from requirements to release and from the car to the cloud), supported by comprehensive project-management tools (digital thread, process/data models, etc.).
• Employ predictive analytics, machine learning, and real-time data to proactively identify potential issues before they arise, recognizing patterns and trends that could lead to system failures.
• Leverage compute power to provide ability for vehicle to self-test, self-predict, and self-fix issues along the life cycle. Zero-forward quality becomes the standard versus just an aspiration.
• Provide dedicated and responsive customer support, such as in-vehicle assistance, remote diagnostics, and a network of skilled service technicians.
• Continuously monitor and analyze customer feedback and usage data to pinpoint areas for improvement. Ensure ongoing customer satisfaction by regularly updating software and refining existing features and services.

Platform simplification and evolution to software

The major difference between hardware and software is that while hardware generally requires minimal maintenance, software often requires ongoing updates and fixes. So, an upgraded vehicle’s electrical and electronic architecture will be the foundation of an SDV that allows for feature upgradability and reduced hardware component count (e.g., from 60–150 ECUs² in a lower-to-luxury-tier car to a few high-performance domain controllers in an SDV). Some of the necessary prerequisites for a scalable car software platform and operating system include the following:

• Establish one uniform software architecture for the entire vehicle range, making the management and maintenance of vehicle software more scalable (achievable in modules, updates/maintainability, and decoupling hardware).
• Promote a multi-company system-on-a-chip (SoC) strategy not tied to a value chain, with the capability for the OEM to design, code, and use the compute infrastructure (e.g., major updates beyond product launch).
• Implement a progressive and continuous software development workflow, incorporating insights and signing up beta customers post-SOP (e.g., feature flags, OTA), and going from “code to road” in seconds, a single day, or four to five times per day.
• Evolve vehicle electrics and electronics (E/E) architectures to allow software teams to better utilize high-performance hardware with appropriate interfaces and compatibility across multiple hardware generations.
• Rightsize ADAS/AD development activities with prioritized and differentiated focus areas and use cases, partnering with silicon companies, evaluating the latest SoCs, and teaming with the ecosystem to share development costs and accelerate progress.
• Establish a stable software architecture, including core software, runtime environment, drivers, etc., ensuring downward compatibility across hardware generations while maintaining performance.
• Transition from basic telemetry and data collection to advanced big data loop architectures, capable of producing training data sets and enhancing embedded software test suites.
• Incorporate real-time operational design domain (ODD) considerations (weather, carbon footprint, terrain, load/duty) into vehicle design, manufacturing, and in-field performance with ability to control vehicle behavior/features based on real-life scenarios. For example, vehicle prevents features based on ODD input—why should one be able to open a sunroof when it’s raining?

Controlled pivot to cloud-based environments—vehicle onboard and offboard

One of the key advantages of SDVs is the accelerated transformation of the automotive product development workflow to incorporate the efficiency benefits of a cloud native from vehicle onboard system to offboard and cloud. This involves establishing a consistent platform for in-vehicle operating system (OS), containers, DevOps, and micro services to promote a “build once and deploy anywhere” approach. However, this can be easier said than done, due to the necessity for engineering platforms (e.g., APICE, requirements and systems engineering, product life cycle management, safety code, testing) that meet automotive-grade standards. Some critical elements of a cloud-centric approach for both vehicle onboard and offboard include the following:

• Establish accurate R&D planning and development schedules (across production, supply chain management, and R&D) for fully manufactured vehicles that are predominantly made up of software capabilities.
• Enhance developer productivity by using cloud platforms to speed up build systems and shorten development cycles, including collaboration with suppliers in different ways.
• Utilize hardware-free prototyping through emulation and software-in-the-loop testing and simulation to achieve higher code quality before physical tests.
• Utilize generic hardware in the cloud as a proxy for first module samples, as the ability of cloud hardware to more closely mimic vehicle compute rapidly improves.
• Move to a completely virtual development environment in the cloud where developers can deploy anywhere on multiple operating systems (QNX, Android, Wind River, Automotive Grade Linux, RedHat, etc.) with automated build toolchains.
• Provide on-demand visibility of vehicle configuration and change history by tracking genealogy data during the product development and manufacturing life cycle. Make the vehicle master for what is installed inside it in any mismatch between car and cloud versioning.
• Eliminate downtime during the build or post-build phase due to incompatible software versions, minimizing the effort required to diagnose and resolve software/hardware issues due to frequent software version changes.
• Integrate cybersecurity and safety criticality into a streamlined homologation workflow and implement automotive-grade development pipelines that balance risk, speed, and adherence to regulations (e.g., SOTIF, ISO 26262).
• Increase the frequency and expand the coverage in larger OTA upgrade campaigns, improving proficiency over time and emphasizing stability.

Becoming a league player in SDV

Over the next 20 years, the factors that distinguish winners and leaders from laggards in the automotive industry will largely depend on the SDV capabilities of their product portfolio and their ability to transition product development to a customer-relevant and software-driven mindset. The cost and number of cars produced used to be key factors of commercial success, but now it will be the ability to create a scalable organization that follows a similar growth pattern to a technology company.

As software becomes the primary factor in a vehicle’s value proposition and drives new business models, there is a need for streamlined and agile end-to-end automotive product development workflow. OEMs should design their organizations with the understanding that the ecosystem is ever-changing and there will be a continuous evolution of high-performance computing and more open software platforms.

Endnotes

¹ Nathan Eddy “Continental sees software as Everest of challenges,” Automotive News, March 11, 2023.
² Dr. Harald Proff and Philipp Wolf, “Software is transforming the automotive world,” Deloitte Insights, June 18, 2020; Dr. Harald Proff, Thomas Pottebaum, and Philipp Wolf, Autonomous driving: Moonshot project with quantum leap from hardware to software & AI focus, Deloitte, 2019; Christoph Hammerschmidt, “Number of automotive ECUs continues to rise,” eeNews Europe, May 15, 2019.

Authors