Car Companies Can’t Take on Software Alone: Lessons from Tesla, Toyota, and Volkswagen

At the beginning of April 2024, Tata Technologies and BMW Group announced a 100-employee strong joint venture that is aiming to target Software-Defined Vehicle (SDV) solutions, with rapid growth in headcount in the coming years reaching the thousands. The venture is BMW’s step toward technology stack ownership for its premium vehicle range and business Information Technology (IT) operations, following an industry trend of automotive Original Equipment Manufacturers (OEMs) taking control of more components in their supply chain. Other software ventures by OEMs include Volkswagen’s (VW) CARIAD, and Woven by Toyota, which were created to build a full technology stack for their respective OEMs. The appeal of ownership is strong as the industry transitions to SDVs, where creating your own Operating System (OS) and cloud architecture, as well as the underlying computing hardware, will more easily enable the continuous development and upgrading of features in the car, across several functions from automated driving to in-vehicle voice assistants. Efficiency aside, technology stack ownership also endows OEMs with a stronger sense of supply chain security—something they experienced a lack of during operations in the COVID-19 era. However, this approach has not been as simple as OEMs thought, with many facing delays, bugs, and misalignment of priorities as a result of in-housing software development.

The outcomes of VW’s CARIAD and Woven by Toyota, previously known as Woven Planet, initially left much to be desired. Both began as software subsidiaries with optimistic goals of unified, full-stack ownership that can proliferate throughout their vehicle lines, but were marred by software glitches and bugs that delayed launch of the ID.3, and misalignment of expectations pushed deadlines of Toyota’s Arene software platform from 2025 to 2027. While they were launched as independent agents that could be free from the size and complexity of their OEMs, corporate/organizational restructuring and realignment of strategic goals have been the common themes of both subsidiaries, making them more closely resemble a traditional automotive subsidiary than a tech startup. It became clear that closer integration and smaller, incremental goals had to be implemented. CARIAD chief, Dirk Hilgenberg, called its new approach a “triple jump,” with the next step initially intending to be the launch of the Porsche Macan Sport Utility Vehicle (SUV) this year with VW’s premium software architecture. However, unsatisfied with delays of more than 2 years that are estimated to have cost VW €3 billion, Porsche opted to use Google’s Android Automotive OS (AAOS) as the basis for the Macan’s software platform instead. Similarly for autonomous driving, the premium brands of the Volkswagen Group adopted Mobileye’s SuperVision and Chauffeur technologies after internal efforts to develop and launch its own autonomous driving systems were underwhelming, as detailed in ABI Insight “Mobileye’s Market-Ready Solutions Surge as Automotive OEMs Struggle to Digest “Open” Platforms.”

Even the best of the industry’s software efforts, Tesla, is struggling to convert its performance into long-lasting advantages compared to the rest of the automotive market. Tesla’s leadership in the “computer on wheels” narrative was often validated by consumers with impressive performance in global markets for new vehicle sales, but China’s BYD recently overtook Tesla for global Battery Electric Vehicle (BEV) sales. While still leading over any other car company in their market value, the gap is closing, and it is becoming more common for stakeholders to evaluate Tesla as a car company, rather than a software company, despite Elon Musk’s recent declaration that Tesla is an “AI/robotics and sustainable energy company.” Waning investor confidence is not a rejection of Tesla’s software expertise coexisting with automotive expertise, but it seems that Tesla is betting on autonomous vehicles to deliver future growth, as other OEMs continue to close the gap in the EV race. While Tesla did blaze the trail for Over-the-Air (OTA) updates and software experiences in vehicles, OEMs should now look to other software-defined devices and industries for their long-term inspiration as the automotive industry matures.

The desire for technology stack ownership will not go away for OEMs, who assumed that the problem with software engineering in the automotive industry came from their slow suppliers. After trying it for themselves, they are now realizing that it is the industry’s engineering practices that have impeded progress and delayed their launches. The answer to this can be inspired from software-defined devices in other industries, with stakeholders already taking steps to adapt their software practices to the automotive industry. For example, several open-source initiatives have been created across different software domains from hypervisors to application layers for infotainment. More approaches are described in ABI Insights “The Changing Approach to Software Development in the Automotive Industry” and “Addressing the Software Skills Shortage in the Automotive Industry.” Moreover, Elektrobit’s recent announcement of EB corbos Linux for Safety Applications shows that open-source development is a promising avenue to tackle even the more safety-critical domains in the automotive software stack.

Another answer to rectifying the industry’s challenges with software is the more modular, “small steps to success” approach that CARIAD is attempting, as long as consistent testing and integration of the solutions are ensuring that their efforts are contributing toward their ultimate goal. In a similar way to how OEMs can embrace the chiplet trend to take more ownership of their supply chain, as explained in ABI Insight “Chipping Away at the Software-Defined Vehicle: How the Chiplet Trend Is Enabling Collaborative Innovation,” they can also slowly take control of software in their vehicles with these collaborations. For example, instead of using their own Central Processing Unit (CPU) designed in collaboration with a silicon vendor, an OEM could install its own non-safety-critical OS for infotainment applications.

Even with smaller steps toward their lofty goals, OEMs should not hesitate to partner for these software challenges with specialized help, or else risk losing valuable time to market to their competitors. For in-car connectivity and generative Artificial Intelligence (AI), CARIAD formed a joint lab in China with smartphone maker vivo, and for autonomous driving systems, it set up a joint venture with Horizon Robotics. BMW’s joint venture with Tata Technologies is the latest in a string of acknowledgements that OEMs’ best path to technology stack ownership and the SDV transition is taking advantage of existing technology leadership, rather than trying to take the reins entirely themselves.

In summary, to take software into their own hands, OEMs should:

• Utilize, participate in, and enable open-source development.
• Segment their software goals into smaller steps that can be independently tested and deployed.
• Engage in closely integrated partnerships with specialists in the software domain that they are expanding into to effectively deploy solutions that can scale across their entire range of vehicle lines and open up the potential of SDV development.