How India MATLAB Expo Grew with Auto’s Software Shift

India’s auto industry has spent the last 15 years racing from mechanical dominance to software overload. BS4 gave way to BS6, electric drivetrains arrived in force, and automated driving moved from conference slides to regulatory drafts. In that same window, one constant in engineering toolchains has quietly reshaped itself: MATLAB and Simulink, and with them the India MATLAB Expo that has grown into a key metric of the sector’s software ambitions.

From Combustion Controls to Software-defined Cars

When the first India MATLAB event ran in 2010, most automotive code written on the platform ended up in engines and transmissions. “Fifteen years ago, the automotive industry was majorly using electronics and software for their powertrain controls, for engine-related controls, for transmission controls, and so on. That mechanism is model-based design,” recalls Prasanna Deshpande, Manager, Application Engineering at MathWorks India. Engineers built mathematical models of engines and gearboxes and tuned controllers in simulation.

Today, the same tools sit much closer to the centre of the vehicle architecture conversation. As automakers pursue software-defined vehicles, where application software is decoupled from hardware and over-the-air updates are expected, MATLAB and Simulink are used to sketch system architectures, validate service-oriented designs, and coordinate software and data flows across domains. The journey mirrors the industry’s own evolution from a few standalone ECUs to fleets of controllers backed by high-performance compute platforms and GPUs running ADAS and automated driving algorithms.

Expo as a Mirror of the Auto’s Software Turn

The 15th India MATLAB Expo in Bangalore, still the event’s home base after occasional forays into multi-city formats, reflects how its remit has broadened. What began as a user conference for control engineers is now pitched as a platform where industry, academia, and startups compare notes on AI, electrification, virtualisation, and software-defined vehicles. “Technology is transforming at a very rapid pace, be it in the form of AI, electrification, automotive, aerospace… customers are continuously interested to know what are the recent developments and how it can benefit them,” says Vijayalayan R, Senior Manager, Application Engineering at MathWorks India.

The Expo’s audience has diversified, too. The core remains R&D and engineering design teams, but the personas now include system architects, software developers working on embedded controls and AI, and data engineers charged with extracting insights from vehicle and fleet data. For many OEMs, the event doubles as a forum to understand how peers are using virtual vehicles, virtual sensors, and digital twins to compress test cycles and avoid late-stage surprises.

Model-based Design Becomes a Digital Thread

Vijayalayan describes a shift from MBD as an isolated activity to models acting as a digital thread that stitches together requirements, design, testing, and field data. In a classic V-cycle, many software defects only surface during hardware-in-loop or full-vehicle testing, when fixing them is slow and costly. By contrast, “shift left” practices move as much verification as possible to the early stages, using models as virtual prototypes to check feasibility and behaviour while requirements are still being written. The aim, he says, is to “front-load most of the activities that actually happen in the test track to the desktops,” so teams can get it “right the first time.”

From Statistics to AI, and On to GenAI

MATLAB’s analytics stack has also expanded in lockstep with the data deluge hitting automakers. Early statistical and neural-network toolboxes have been re-cast as Statistics and Machine Learning Toolbox and Deep Learning Toolbox, reflecting the leap from basic data analysis to production AI deployed on vehicles and in the cloud. Those tools underpin virtual sensors that replace or augment physical devices – for example, using data-driven models instead of costly emission after-treatment sensors now mandated by BS6 and future BS7 norms.

On the newest frontier, MathWorks is weaving generative AI into its workflows, albeit with caution. The company launched MATLAB Copilot in 2025, followed by Simulink Copilot and Polyspace Copilot in April 2026, as embedded assistants that explain models, annotate scripts, and help track down bugs. “It’s like an interactive AI assistant that can help the engineers to explore, explain the models, explain the scripts, and debug everything,” says Vijayalayan. Deshpande stresses that in an industry full of safety-critical functions and tightly regulated user experiences, GenAI must remain a “productivity and efficiency multiplier” wrapped around “trusted tools” that perform the core work.

The newly introduced agentic toolkit extends that logic by allowing AI agents – running on whichever large language model a customer prefers, to call into MATLAB and Simulink through defined skills and plug-ins. “We are in the era of agentic workflows, where AI agents help in automating multiple-step tasks… Our goal is to enable the seamless integration of AI agents with MATLAB Simulink,” Vijayalayan says. In practice, that could mean agents that generate test benches, run design-of-experiments campaigns, or refactor model structures, leaving engineers to focus on architecture and interpretation.

Electrification: Batteries, Twins and Range Anxiety

If there is one technical area where MATLAB’s evolution is particularly visible, it is batteries. Indian operating conditions, heat, dust, and load cycles, have exposed the limits of naïve pack designs and shallow BMS strategies. Deshpande points out that engineers do not really “control” batteries; they manage them, working around electrochemistry that is highly sensitive to temperature and current. Keeping thousands of cells within safe thermal and electrical bounds, and preventing a single runaway cell from cascading, is the central BMS challenge.

Simscape Battery, a domain-specific library within the Simscape family, is MathWorks’ answer to that modelling burden. Engineers can pick from established cell types such as pouch and prismatic, assemble modules and packs, add cooling hardware, and quickly evaluate thermal performance and control strategies in one-dimensional simulations. On top of that, supervisory logic for detecting low voltage, high temperature, or other fault conditions is built in Stateflow, a graphical environment for state machines and decision logic. Because temperatures cannot be measured at every cell, MATLAB’s estimation algorithms, including Kalman filters, infer internal states from limited sensor data.

As EV fleets grow, those models evolve into digital twins, updated using operational data to reflect ageing and real-world stress. That enables more accurate range predictions, better charging recommendations, and, potentially, new services such as battery-as-a-service or cloud-based battery health offerings. For consumers, the payoff is less range anxiety; for OEMs and suppliers, it is a way to turn data and models into recurring revenue rather than one-time hardware sales.

Startups as Proof Points

One way to read MATLAB’s 15-year arc is through the lens of Indian startups that built their engineering pipelines around it. Ather Energy, now a mainstream electric two-wheeler OEM, first used MATLAB on complimentary licences while incubated at IIT Madras around 2013. “Then they shifted to our startup offerings, which is a very discounted offering. And then our technical teams were all through engaged with them,” says Deshpande.

Ather needed to make an electric scooter feel as familiar as an internal-combustion one while hitting performance and cost targets. Model-based design lets the team quickly prototype battery systems, motor controllers, and charging logic in simulation, benchmark against conventional vehicles, and cut the number of physical prototypes they have to build. According to Deshpande, Ather’s own user story describes reducing prototype counts from double digits to roughly a third of that, with strong designs emerging “even before the physical prototypes” were on the road.

Newer players such as Raptee, which is targeting high-voltage architectures for motorcycles, use MATLAB and Simulink to explore trade-offs between motors, battery packs, and vehicle dynamics entirely on the desktop. In three-wheelers, where price pressure is intense, and hardware must be heavily optimised, MathWorks has supported companies on motor controls and BMS that squeeze more control out of smaller ECUs.

Industry 4.0 and Beyond the Vehicle

The same modelling and AI tools are also moving deeper into the factory. As Indian auto hubs expand in regions such as Aurangabad and Nashik, machinery builders and plant engineers are using MATLAB and Simulink for virtual commissioning, modelling multi-domain production systems, and testing complex control algorithms in a virtual environment before installing them on expensive hardware. In parallel, predictive maintenance models help operators anticipate failures that could shut down lines, combining physics-based models with historical data to catch edge cases.

Computer-vision-based automated visual inspection is another emerging use case. Instead of relying on humans to spot cracks, paint defects, or assembly errors on fast-moving lines, camera systems feed images into AI models built and tested in MATLAB to flag anomalies at speed. For MathWorks, this extends its automotive exposure from vehicles in the field to the plants that build them.

Financials: Tools as a Growth Engine

All of this is showing up in MathWorks India’s financial trajectory. The company’s total income rose from Rs 923.6 crore in FY2024 to Rs 984.6 crore in FY2025, while net profit increased from Rs 54.0 crore to Rs 73.8 crore.

The directors, in their latest regulatory filings, say they are “continuously looking for avenues for future growth” and expect higher turnover and profit alongside broader economic expansion. For now, the combination of export-linked development work and rising domestic demand for licences and services, with automotive a key vertical, is turning what was once a specialist engineering tool into a visible profit engine.

The company faces intense competition from major software giants and specialised engineering simulation providers such as Dassault Systèmes and Altair, among others. Many companies, including startups, are also using open-source models.

According to a list prepared by TheirStack, Indian auto companies using MATLAB include KPIT, Tata Technologies, Ather, and Mercedes-Benz Research & Development India Pvt Ltd., among several others.

Next phase: Agents in a Billion-line World

Looking ahead, both Deshpande and Vijayalayan see the scale of automotive software as the main driver of change. A modern luxury car, they note, can already approach a billion lines of code, on par with aircraft, with around 100 ECUs gradually giving way to zones run by powerful processors. In that context, the industry’s move to software factories, where code flows through automated pipelines of modelling, testing, and deployment, looks inevitable.

If the past 15 years are any guide, the answer will be written both in the sessions at future India MATLAB Expos and in the software stacks of the vehicles rolling out of Indian plants.