Embedded development sheds its hardware tether

Automotive and embedded software teams have historically been tied to physical hardware. Developers kept test boards connected to their workstations or maintained in-house testing farms, largely because so much of the toolchain was proprietary. That model is now breaking down as processors grow more powerful, open standards emerge for the lower-level software stack, and open-source reuse becomes the norm across domains.

The friction is real: teams must deliver on application commitments while carrying the cost of maintaining their own development and test environments. The pressure only intensifies as the rate of foundational change accelerates. Cloud-based development promises to remove much of that overhead, provided the underlying infrastructure can support the specialized needs of embedded work.

What virtualization means for automotive targets

Virtualization has been a cornerstone of enterprise development for years, but embedded systems have adopted it cautiously. The diversity of processors, operating systems, and specialized software in this space makes a uniform approach impractical — embedded virtualization must accommodate far more than standard Linux and Windows on Intel architectures.

March 2024 marked a turning point with Arm’s announcement of its new Automotive Enhanced (AE) processors, designed to boost AI capabilities in the automotive sector while maintaining Instruction Set Architecture (ISA) compatibility. That compatibility is the key enabler: developers can write, build, and test code in the cloud, then run it in virtualized environments with digital twins that mirror their processor targets — even before those targets exist in silicon. Applications can be developed in the cloud and moved to vehicles or other edge devices without extensive reconfiguration.

CI/CD moves to the cloud

Continuous integration has been practiced by embedded teams for years, but managing development environments across assorted hardware configurations and deployment targets remains costly. Cloud-based CI/CD layers the usual cloud advantages onto that workflow:

  • Enhanced scalability. Teams allocate resources and optimize compute spend dynamically, running workloads in parallel across many hardware and software configurations. Developers can collaborate across geographies or even across supply-chain organizational boundaries.
  • Reduced complexity. Standardized cloud environments minimize setup and teardown time, and workflows can be shared across teams.
  • Improved quality. When compute is scarce or a CI environment is brittle, teams tend to optimize against too narrow a slice of the development process. Removing that friction and lengthening end-to-end feedback loops translates into better outcomes.

To make this practical for embedded developers, GitHub updated its infrastructure. In October 2023, native Arm64 support arrived for GitHub Actions, the hosted CI/CD workflow engine. Arm’s processor designs are central to much of the embedded and automotive world, which makes this platform support particularly significant — it frees developers from desktop-bound workflows by moving jobs into the cloud.

GitHub followed in April 2024 with the public beta of GPU-hosted runners, enabling teams building machine learning models to run complete application tests, including ML components, within GitHub Actions.

A broader industry shift

This convergence of cloud technology, virtualization, and new processor architectures represents a real change in how automotive software gets built. GitHub has joined SOAFEE.io (Scalable Open Architecture For Embedded Edge), an industry collaboration working to simplify vehicle software, and maintains its membership in the Connected Vehicle Systems Alliance (COVESA) while supporting Microsoft’s commitment to the Eclipse Software Defined Vehicle project.

The combination of GitHub Enterprise Cloud and Arm’s AE processors points toward a development model where testing transcends traditional hardware boundaries. Developers spend more time writing code and less time managing infrastructure, and applications transition from cloud to edge without extensive rework. The result is a meaningful acceleration in time-to-market for new automotive technologies — and a broader indicator that the cloud is becoming the default backbone for software engineering across industries that were once firmly hardware-bound.