Automakers now increasingly lean on partners that do not build cars. In software-defined vehicles (SDVs), features are delivered and updated mainly through software. In a recent survey by AlixPartners on SDVs, when automakers were asked who their most critical partners were, 72% of US automakers picked hyperscalers and AI providers. Middleware/platform, or embedded software vendors were selected by 69% of US automakers. The industry's center of gravity is clearly shifting towards the technologies and partners that develop, connect, and operate vehicle software.
The car radio shows how supplier demands have evolved. Once a self-contained component, the car radio received AM and FM signals. It then evolved to accommodate cassette tapes and CDs. Each generation delivered a better listening experience, but the system remained largely within the car. Over time, it became an ecosystem play, with streaming services, phone integration, and voice assistants from different companies all delivering content through the same dashboard. Today, much of the in-car experience depends on software and services that include semiconductors, operating systems, applications, and the cloud. Delivering all the services reliably requires expertise that no single automaker or supplier is likely to possess across the entire stack.
The SDV stack typically has four distinct layers. In-vehicle computing hardware sits at the base, in the form of high-performance computers and zonal controllers. Operating systems and middleware manage hardware resources and enable software components to communicate. Applications deliver driver-facing and vehicle functions. Cloud-based development and operations provide the environment in which teams build, simulate, test, and deploy Over-the-Air (OTA) software updates.
Each layer has its own specialist firms, and the ecosystem extends beyond the car. Design tools, simulation environments, and manufacturing systems all play a part in taking a program from concept to production. The same survey by AlixPartners found that 68% of US automakers combine in-house development with partner purchases, with only 25% building primarily in-house.
All that expertise comes at a price. Partners bring their own tools, interfaces, and delivery processes. Making the layers work together remains the automaker’s responsibility. Finally, the consequences show up in release cycles. Among the automakers surveyed, 40% use more than half of their OTA releases for bug fixes, and 67% take over eight weeks to deploy new SDV applications internally. These figures indicate that integrating and validating software across multiple layers can often slow down the pace of delivery.
One ecosystem, but different priorities
The partners in an SDV ecosystem bring a variety of capabilities and commercial priorities. Automakers need to protect vehicle architecture, safety, and the features that make their brands unique. While cloud and AI providers bring powerful platforms, they need automotive domain knowledge to adapt them to vehicle requirements. Middleware and other layer specialists similarly offer expertise in particular technologies, but may not cover the entire stack.
Their interests align only when the interfaces, responsibilities, and outcomes are clearly defined. This is where ecosystem integration becomes a distinct engineering capability.
Organize partnerships around the layers of the stack
Mapping partners to layers changes how automakers buy and manage technology. Interfaces and responsibilities need to be agreed upon, before teams write code against them. Any gaps that exist can then turn up in architecture reviews well before they reach a test vehicle.
For partnerships like these to last, both sides must gain something from the arrangement. For a chipmaker, that might be automotive software built for its silicon, opening doors to new customers. A cloud provider can build automotive domain knowledge into its platform. A middleware specialist with deep in-vehicle expertise and less reach in the cloud gains partners who cover that area.
Share the software that does not differentiate
The most useful collaboration often happens between companies that compete. As a case in point, the Eclipse Foundation and the German Association of the Automotive Industry (VDA) used CES 2026 to announce that their open-source vehicle software pact, signed by 11 companies in 2025, now has 32 signatories, including automakers, suppliers, semiconductor companies, and cloud providers. The group aims to reduce development, integration, and maintenance efforts by up to 40% and achieve up to 30% faster time to market.
A good share of the underlying vehicle software performs common functions. Booting a controller, passing messages across the vehicle network, and running basic safety checks across brands are more or less alike. Building and maintaining these foundations collectively can free engineering resources for features that customers notice and pay for. Infosys joined the Eclipse SDV Working Group in 2026 to contribute to that shared work.
Give one partner end-to-end accountability
Engineering services firms are hearing a different request too. The traditional brief was often limited to one part of the work such as a module to design, or a function to test. However, SDV programs cut across multiple layers. So, automakers increasingly need partners that can coordinate between specialists and take responsibility for the outcomes. Call this role the SDV ecosystem orchestrator. It is a move from service provider to solution provider.
An adaptive high-beam feature illustrates the point. The driver pays for a subscription, and the feature is activated through an OTA update. Four things have to work together: the headlight controller, its middleware service, the feature app, and the cloud delivery pipeline. If the SDV ecosystem orchestrator has tested all four elements, the automaker knows whom to call when a fault appears. Individual specialists still remain responsible for their components.
This requires more than just project coordination. It requires a system integrator to have automotive domain knowledge, cross-layer engineering capabilities, and shared tools for managing software changes, validation, and compliance. While AI-assisted engineering can handle much of this workflow, it must still operate within the program's safety and governance requirements.
In a nutshell
Automakers’ expectations from their suppliers have largely shifted. They still want well-made components, but they also want an SDV ecosystem orchestrator who can get many specialists to work as one. In the era of SDVs, this level of engineering orchestration has become a core requirement.
That role completes a larger shift. SDVs need engineering workflows that AI can speed up without losing control, validation that keeps pace with every software release, and an ecosystem that performs as one. When all three come together, the vehicle becomes a platform that keeps improving long after it leaves the factory.
