Volvo Cars is moving its global Android Automotive software development environment onto an open-source cloud platform developed by Accenture and Google Cloud, shifting more of the engineering work behind connected vehicles away from dedicated physical development infrastructure. The Swedish carmaker has become the lead industry partner for Horizon, which combines cloud-native development tools, virtual test environments, remote access to device farms, and AI-assisted software workflows.

The agreement concerns the environment used to build and validate software rather than another consumer-facing feature inside the car. Engineers can create virtual workbenches, test software against simulated Android Automotive environments, and connect remotely to physical or virtual devices, reducing the amount of specialist hardware that has to sit beside every development team.

Volvo will also feed its engineering experience into future development of the platform, giving the arrangement a broader industrial role than a conventional customer deployment. Accenture and Google Cloud want Horizon to become reusable infrastructure for other vehicle manufacturers and suppliers rather than a custom environment built for one carmaker.

Supplier claims that the platform can accelerate testing or reduce development costs should be treated as performance targets rather than independent evidence of Volvo’s results, because the migration is still being implemented. The more durable change is architectural: more of the engineering pipeline is being treated as cloud infrastructure capable of being provisioned, shared, automated, and observed centrally.

Vehicle development becomes a software infrastructure problem

Cars have accumulated connected services, over-the-air updates, digital cockpits, driver interfaces, and software-managed functions, which means manufacturers now maintain software long after a vehicle leaves the factory. Engineering environments consequently need to build and test repeated software changes across different configurations without reproducing a conventional model-year development programme for every release.

Physical test benches remain necessary, although they can become a bottleneck when geographically dispersed teams need access to specific hardware. Virtualisation allows more testing to happen before code reaches a physical vehicle or dedicated rig, while cloud capacity makes it possible to run larger numbers of tests in parallel.

Remote device farms provide a bridge between those two worlds by letting teams connect to centrally managed physical and virtual environments instead of maintaining a local collection of specialist equipment. That can improve utilisation and make development environments easier to reproduce, particularly as software teams expand across locations.

The approach also changes onboarding and project mobility because a new developer can be given a standardised virtual environment instead of waiting for hardware to be delivered and configured. Those apparently mundane delays become expensive at scale when hundreds or thousands of engineers depend on specialised development equipment.

Cloud engineering moves upstream

Industrial cloud adoption is moving beyond corporate applications and customer-facing digital services into the engineering process itself. Builds, simulation, data analysis, testing, collaboration, and model development can all consume infrastructure dynamically, turning product development into another major source of cloud demand.

Automotive software is a demanding example because the final product combines software with physical components, safety requirements, long maintenance periods, and several technology suppliers. A developer can reproduce parts of an application environment virtually, but the vehicle ultimately has to behave predictably when those instructions meet processors, sensors, displays, networks, and other hardware.

Virtual testing is therefore most useful when it shortens the route to physical validation rather than pretending hardware has disappeared. Horizon’s design reflects that distinction by combining simulated Android environments with access to physical devices, allowing different stages of development to use different levels of virtualisation.

This division also creates clearer opportunities for automation because repetitive builds and tests can be run at cloud scale while scarcer physical resources are reserved for the stages that genuinely need them. The efficiency gain comes from changing the shape of the pipeline rather than eliminating hardware entirely.

AI enters the development pipeline

Artificial intelligence sits inside this model as one component of the engineering environment rather than the product itself. Horizon includes AI-assisted development and agentic testing, while Volvo’s experience is expected to influence how those functions are expanded for manufacturers and suppliers.

Models can help generate or analyse code, create tests, investigate failures, and move work between stages of a development process, although faster code production has limited value if builds, hardware access, validation, or release processes remain slow. The surrounding engineering system determines whether AI acceleration in one step creates real productivity or merely moves the bottleneck elsewhere.

Vehicle software also gives manufacturers reason to resist measuring success solely through development speed. Engineering teams need traceability around what changed, which tests ran, what failed, and how a release was validated before deployment, particularly when software will remain in service for years.

Volvo’s migration joins two technology shifts that are often discussed separately: vehicles are becoming more dependent on software throughout their lives, while the infrastructure used to produce that software is becoming more cloud-based, virtualised, and automated. The operational test will be whether that combination reduces engineering friction without weakening the physical validation and release controls that complex industrial products still require.