How virtual validation is rewriting SDV test strategy
Vehicle software sits at roughly 300 million lines of code today and is expected to reach 600 million or more by 2027. New-entrant automakers already build Software-Defined Vehicles (SDVs), where vehicle functionality is delivered primarily through software rather than fixed hardware, in 24 to 30 months; legacy Original Equipment Manufacturer (OEM) cycles still take 48 to 60 months. SDVs could account for 90 percent of vehicle production by 2029, up from just 3.4 percent in 2021. Among other factors, what separates new entrants from legacy OEMs is validation speed: how fast an OEM can test the software it builds.
Physical validation, including hardware-in-the-loop (HIL) testing, test fleets, and proving grounds, was built for a different pace. Feedback is tied to hardware availability, so defects surface late, when fixes become very expensive. Physical coverage has a ceiling too: the rare, safety-critical edge cases advanced driver-assistance systems must handle can’t be reliably reproduced on a track or in a fleet.
Compliance is an important consideration as well. Over-the-air (OTA) updates let manufacturers ship new vehicle software at a cadence far faster than a traditional model-year release. But the evidence that each update meets standards like ISO 26262, the industry’s functional safety standard for road vehicles, is typically assembled only at a scheduled release gate. This means that if updates are pushed faster, that evidence has to be rebuilt from scratch more often.
Three capabilities make up a virtual validation stack, built one on top of the other rather than side by side. Virtual Electronic Control Units (vECUs) come first: software replicas of vehicle control units that let teams test production-intent code before physical hardware even exists. That foundation is what makes the next two possible. Cloud-scale scenario simulation reaches the long-tail and safety-critical conditions physical testing never will. Continuous traceability keeps compliance evidence current as work happens, instead of being reconstructed at a release gate.
Shift defect discovery left with virtual ECUs
A vECU runs production-intent code without the hardware needing to exist yet. vECUs also stand in for a physical device at the product conceptualization stage, before testing begins. Inserting vECUs into automated testing pipelines enables software-in-the-loop (SIL) testing from the earliest development stages, so defects that previously surfaced late in HIL testing show up at build time instead.
vECUs plug into AUTOSAR, the standard software architecture most automotive suppliers already build to, so virtual and physical components blend into one continuous process instead of running as separate phases. Teams build vECUs at different levels of detail, from simplified simulations to models that behave like the real hardware. In Infosys SDV work, virtualization has cut hardware requirements by 70 percent.
Cover the scenarios physical testing cannot reach
A test fleet covers a narrow slice of real-world conditions by logistics and economics alone. One recent study found that distributing a 200-scenario automated-driving test campaign across a cluster of Kubernetes — a system for running many computing jobs in parallel across a pool of machines — cut runtime from 475 minutes to under an hour, more than an eightfold speedup over running it sequentially. Cloud-orchestrated campaigns at that scale can reach the low-frequency, safety-critical edge cases, sensor glare, occlusion patterns, rare traffic behavior, that Safety Of The Intended Functionality (SOTIF) analysis requires. SOTIF is the standard that addresses safety risk from a vehicle behaving exactly as intended under unanticipated conditions.
Structured scenario libraries that blend requirement-derived tests with synthetic long-tail conditions generate the coverage metrics and residual risk arguments SOTIF analysis is built on.
Make compliance evidence continuous
ISO 26262 already requires traceability, which links requirements, design decisions, code changes, and test results into one continuous chain. SOTIF treats the same discipline as essential, even where it stops short of a formal mandate. UN Regulation 156 mandates a software update management system with traceability and auditability. Virtual validation changes the cadence. Instead of the evidence chain being reconstructed at scheduled release gates, the system stays current in near-real time, as each change happens.
A late change no longer forces teams to regenerate validation deliverables across SIL, HIL, and vehicle-level testing from scratch, because the evidence will be current all along. Such a capability matters most when programs can least afford delay, such as in safety recall investigations, regulatory audits, or OTA rollbacks, that need to show exactly what changed, and why.
Virtual validation rebalances the entire test stack. Coverage, defect discovery, and evidence generation shift into a scalable, software-first environment. Physical validation performs the important hardware confirmation. Today, such a rebalancing is a structural requirement for SDV programs competing on release cadence.