Mitigate chip shortage with new software architectures

The semiconductor chip shortage has slammed the auto industry. Automakers have trimmed their production plans across the globe by millions of vehicles, costing them more than $200 billion in lost sales in 2021.

How did we get here? It comes down to a convergence of factors, including the significant use of semiconductor chips in modern automobiles, surging demand for chips across various industries and the impact of the COVID-19 pandemic on automotive and semiconductor production and logistics.

At this point, you might ask: Why don't automakers just substitute other chips for the ones they can't get? For starters, many of the other automotive-certified chips are also in short supply. And, even if other chips were available — replacing one chip with another is often difficult.

Modern vehicles can contain over 1,000 chips in up to 150 electronic control units, which manage vehicle functions ranging from drivetrain to braking to power windows. Most of these control units are extremely cost-optimized, with just enough processor, memory and storage for their intended functions. The software running in these control units is optimized for the chips used — to minimize code size and CPU load. Updating this software to run on other chips and validating the solution can be difficult and time-consuming.

What is being done to address the problem? In the short term, automakers have been scrambling to procure more chips, and most have shifted production to higher-margin vehicles and to vehicles that require fewer chips, in order to maximize revenue and profit. Some have removed feature options and associated control units to reduce chip usage.

Automakers also are modifying their just-in-time supply chains to buffer critical components. But the reality is that the chip shortage will take a year or more to resolve fully because it will take time for chipmakers to increase capacity of their older production lines.

What automakers really need to minimize the impact of hardware supply disruptions is to develop hardware-agnostic software. Basically, they need to accelerate the evolution to software-defined vehicles.

• A key requirement of software-defined vehicles is that software must be portable. It must be abstracted from specific hardware that could pose a supply issue or technology limitation. This results in trusted software that can be reused easily across electronic control units, models and architectures, and is less susceptible to hardware functionality, quality or availability issues.
• Another key requirement is that software must be easily updatable. Software defects are inevitable in large code bases, and modern vehicles can contain 150 million lines of code (and growing). Automakers now realize that over-the-air software updates help ensure quality, safety and security — as well as enhance functionality — over the vehicle's lifetime. However, updating software has implications, such as development and test costs, time to market, transmission costs, installation time, failed updates, software interaction issues and potential security risks. So the smartphone model of constant updates isn't well suited for automotive. Which brings us to a third key requirement of software-defined vehicles.
• Software needs to be remotely monitorable and controllable. In enterprise networks, software-defined network devices (routers and switches) are designed to be easily and remotely controlled using policies or real-time commands. This allows end-point devices with less-complex software and cost-effective hardware to address a very broad range of use cases and situations.

Likewise, vehicles need to evolve to be fully software-defined in order to manage complexity and enable innovation. The key here is to develop software that can be easily, quickly and safely configured to change vehicle behavior without requiring new software.

This means embracing digital dynamics — speed, agility and control with digital technologies — to build the vehicles of the future.

In a software-defined world, many of the basic building blocks already exist in the software. This reduces the costs and time of developing new software, and reduces the number of software updates and their inherent risk. Vehicle software can also be easily and dynamically configured to work in specific, customized ways at scale. This will be increasingly important in the future, as users demand more personalized mobility experiences.

Going forward, the mantra for every automaker should be: Configure and automate when you can; update code when you must.

The chip shortage has made one fact crystal clear: Automakers need to change the way they make cars and they need to do it now. It's time to build mobility solutions that meet the evolving needs and expectations of the marketplace. And it's time to create vehicle features and functionality without having to write a single line of code. That's how automakers can gain competitive advantage and go further and faster on the digital road ahead.