Insights from a cross-industry executive panel spanning automotive engineering and semiconductor

The automotive industry’s shift to software-defined vehicles (SDVs) is often framed as a technology evolution. In practice, it is a workflow re-architecture: a change in how product value is created, integrated, validated, and continuously improved over the vehicle’s lifetime.

As software content grows and compute architectures become heterogeneous, the late-in-the-cycle integration event has become the most expensive habit the industry is trying to break. Virtualization is emerging as the most pragmatic lever to do that—not because it replaces hardware, but because it stabilizes engineering throughput in a world where complexity and schedule pressure compound.

Watch the executive panel event recording and review the panelists' presentations

From tool choice to discipline

Virtualization used to be discussed as a toolset: helpful, sometimes transformative, often constrained by fidelity or adoption. That conversation is changing. Across the ecosystem, virtualization is increasingly treated as a discipline: a set of repeatable practices, interfaces, and validation approaches that must work across organizations that historically optimized for different timelines and incentives.

That is why this topic now sits at the intersection of automotive and semiconductors. Virtualization is being used to support early software development, integration, and validation while hardware platforms are still evolving, and that early start is now widely viewed as essential to delivering SDVs on time.

“Shift left” is no longer aspirational

“Shift left” has been a familiar phrase for years, but the meaning has sharpened: virtualization enables earlier detection and isolation of software defects, reduces expensive late-stage issues, and supports continuous integration and testing—capabilities that are increasingly treated as mandatory for SDV delivery.

What changes when a program truly shifts left is not only when defects are found, but how teams behave:

• Integration becomes continuous rather than episodic.
• Validation becomes scalable rather than capacity-limited by scarce physical assets.
• Architecture decisions are tested earlier, when change is still affordable.

The result is a more resilient development system: faster feedback loops, fewer surprises late in the program, and a better chance of maintaining SDV cadence without sacrificing quality.

The real question: what should be virtualized, and how far?

A mature discussion about virtualization quickly moves beyond “should we use it?” and toward “what exactly needs to be virtualized—and at what fidelity?” The answer depends on the use case: application development, integration, or system-level validation.

In other words, virtualization is not a single bet. It’s a portfolio. Some implementations prioritize speed and iteration; others prioritize fidelity. Treating virtualization as a discipline means building the portfolio intentionally—choosing the right depth for the right job, then evolving it as programs mature.

The industry is also moving toward more granular definitions of virtualization levels (for example, level-based models that describe increasing fidelity and scope), but a key gap remains: further standardization is needed to harmonize requirements and reduce friction between partners.

Standardization is the multiplier

If virtualization is now a cross-industry discipline, standardization becomes its multiplier. The ecosystem needs better alignment on interfaces and model standards to unlock the full value of virtualization and enable seamless collaboration across the supply chain.

Without standardization, every boundary between organizations risks becoming a translation tax: duplicated modeling effort, mismatched abstractions, and integration work that consume time without improving the  development flow. Standardization does not eliminate differentiation; it creates a stable base that allows each party to innovate without constantly re-building the same plumbing.

For SDVs, that plumbing matters more than most teams expect. The ability to plug components into a common validation flow—quickly, repeatedly, and with confidence—often determines what can ship, what can be updated, and what can be supported at scale.

Semiconductor and software are now co-scheduled

One reason virtualization is inherently cross-industry is that semiconductor roadmaps and automotive software roadmaps are now tightly coupled. Early access to virtual models can enable software development and feedback before hardware is available, supporting more robust system design and faster time-to-market.

From the automotive side of the equation, virtualization provides a path to system-level confidence earlier in the lifecycle—helping teams integrate sooner, validate more often, and reduce the risk of laste-stage issues.

The next wave: heterogeneous compute and AI accelerators

Vehicles are adopting increasingly heterogeneous architectures, including AI accelerators and specialized processing units. This trend increases the importance of virtual models and toolchains that simplify programming and integration.

Heterogeneity is both a capability unlock and a complexity multiplier. It increases the number of integration surfaces and the risk of late-stage “it works in isolation but not in the system” discoveries. Virtualization helps teams explore HW/SW architectures and validate integration behavior earlier, helping identify potential issues while changes are still feasible.

Beyond development: virtualization’s post-production value

Virtualization’s value does not end at SOP. It can support post-production activities such as root-cause analysis of field issues and continuous over-the-air updates, extending its relevance across the vehicle lifecycle.

This matters because SDVs are not finished products in the traditional sense. They evolve. And an organization that expects software to evolve safely needs repeatable environments to validate change, diagnose issues, and sustain quality over years—not just over a program milestone. Virtualization becomes a foundation for continuous improvement and faster innovation cycles, not merely a development accelerator.

Bringing the ecosystem together

In fast transitions, leadership is not only about having answers; it’s about convening the right questions across the organizations that must solve them together. That’s why MathWorks hosted a virtual executive panel on hardware virtualization to explore how automotive companies and semiconductor suppliers should prepare for the opportunities and challenges ahead.

The recurring themes were clear: virtualization is essential for managing SDV complexity, enabling earlier integration and validation, and accelerating time-to-market; standardization is critical to reduce duplication of effort and enable interoperability; and the right approach is often hybrid: choosing the right level of virtualization for the right use case.

Due to its importance, virtualization is increasingly treated as a discipline. It is becoming the common language that helps automotive and semiconductor ecosystems move faster together without breaking the safety and quality expectations that define the industry.

Watch the panel recording and continue the conversation

With sincere thanks to Ritesh Goyal (KPIT), Alessio Canepa (Iveco Group), Christopher Thibeault (Infineon), and Nicola Magistro (STMicroelectronics) for their time, preparation, and openness in sharing real‑world lessons—we truly appreciate their partnership and look forward to continuing the conversation together.

Watch the event recording and review the supporting materials from the session.

Where do you think is the biggest practical gap today—standards, fidelity, scaling validation throughput, or organizational adoption? Please share your thoughts in the comments below!