Automotive industry transitioning to data- and intelligence-driven environment: Excelfore

Founded in Silicon Valley by industry veterans and serial entrepreneurs, Shrinath Acharya, Shrikant Acharya, and John Crosbie, Excelfore is a global leader in data connectivity solutions for software-defined vehicles, with more than 17 million vehicles on the road today. 

Backed by a robust portfolio of technology patents, Excelfore’s SDVconnect framework is widely deployed globally-- across Europe, US, China, Japan, and India, on major cloud hyperscalers, including AWS, Azure, Google, Baidu, and Tencent platforms.  

Excelfore is a founding member of the eSync Alliance, promoting open, collaborative standards for connected mobility. Excelfore was a launch partner for Arm’s CSS AI-defined vehicle platform, and the AWS Agentic AI marketplace.

Shrinath Acharya, Co-Founder and CEO, Excelfore Corp., tells us more. Excerpts from an interview: 

DQ: How are AI-defined vehicles changing the landscape of automotives?

Shrinath Acharya: The widespread use of smartphones and platforms such as Android Auto and Apple CarPlay has made seamless software updates essential, as vehicles must stay synchronized with frequent smartphone and OS upgrades. This demand has driven the evolution of vehicle hardware toward more powerful compute and memory resources, supporting advanced infotainment, connectivity, and automated driving features.

Despite these advancements, much of the vehicle network has long depended on traditional CAN-based ECUs. That is now changing as the automotive industry transitions to a data- and intelligence-driven environment. The rapid electrification of vehicles has further accelerated this shift. Electric drive motors, with their superior scalability and responsiveness compared to combustion engines, are enabling more precise and adaptive control systems.

This transformation requires both hardware and software architectures to evolve toward compute-centric designs capable of handling complex AI workloads. Over just a few years, the Software-Defined Vehicle (SDV) has evolved into what is now known as the AI-Defined Vehicle (AIDV). This new generation of vehicles is built around high-speed Ethernet and serial communication (SER) backbones to enable faster data flow and real-time decision-making.

The AI-Defined Vehicle represents a major leap forward in intelligence and adaptability. It continuously learns from data—understanding driver preferences, predicting road hazards through a network of sensors, and optimizing energy use based on driving patterns and environmental conditions. The result is a vehicle that not only updates its software but intelligently adapts to enhance performance, safety, efficiency, and user experience.

DQ: How is the auto industry is about to transform these new tech and lead to complete automation?

Shrinath Acharya: The first true automation in automotive came with Henry Ford's Model T assembly line, transforming cars from luxury items into affordable mass products. Back then, storied brands like Studebaker, Plymouth, and Packard dominated—many absorbed into GM, while Chrysler evolved through DaimlerChrysler to Stellantis. Ford remained uniquely family-controlled.

Early automation appeared not inside vehicles, but on factory floors—robots handling welding, painting, and assembly. Today, manufacturing is driven by scale, technology, energy efficiency, and personalization, powered by semiconductors. Modern vehicles contain up to 3,000 chips, demanding robotic precision for electronic integration.

China's automated manufacturing adds momentum—companies like Xiaomi launching competitive EVs from highly automated lines. Global automakers are pushed toward full automation, with standardized components like compute platforms, batteries, motors, and LiDAR. The primary differentiator remains exterior design and brand experience.

Vehicles now evolve into connected, intelligent systems—smartphones on wheels—continuously linked to centralized networks enabling real-time analytics, predictive maintenance, and autonomous intervention.

Yet, this brings a dark side: eroding privacy and a "cyborg effect." Vehicles monitoring every movement introduce new risks. As we embrace AI-defined mobility, regulatory frameworks must evolve to safeguard privacy and autonomy, ensuring innovation doesn't compromise humanity and personal freedom.

DQ: Can you explain how this is going to work out in the future?

Shrinath Acharya: AI-defined vehicles need a standards-based cloud-to-edge data pipeline down to the sensors being monitored. The primary communicator is the TCU (Telematics Control Unit)—a secure firewall/gateway that connects the vehicle to its cloud. 

Each car hosts a powerful central compute that fuses sensor inputs, produces curated insights about its dynamic environment, and shares them with the cloud. In the cloud, generative AI refines models; in the vehicle, an agentic AI applies those models, continuously calibrating and course-correcting. 

Telemetry can stream at 10-100 Hz over MQTT; configurations can update live, while full software updates are applied asynchronously when the vehicle is in a safe state.

A concrete example: range anxiety in an EV. I was driving a Porsche Taycan—nominally ~250 miles of range—in peak summer heat with air-conditioning on, and the usable range shrank to about 140 miles.  Last summer, while driving from Santa Barbara to Fremont, a distance of 295 miles, the travel required charging the vehicle twice at high-power charging stops even when we had started with the battery level at 96%.  When I reached home, it merely had 7% remaining- about 14 miles. 

The thrill of a high-performance car turned into a constant countdown, made worse by the scramble to find a fast charger as evening approached. With an AI-defined stack, the car would anticipate the deficit early, forecast the shortfall, adjust performance to stretch distance, and plan a single optimal stop—including charger availability and queue predictions.

Another practical case is delivery vans in stop-and-go traffic, where range suffers and unexpected drivetrain faults are common. Traditional architectures rely on fixed calibrations and scheduled diagnostics. An AI-defined stack predicts near-term energy demand from context, adapts torque/regen and HVAC in real time, flags early signs of component failure, and then improves the model fleet-wide via staged over-the-air (OTA) software updates—capabilities conventional ECUs simply can’t match.

DQ: What are the trends you are envisaging with this technological shift?

Shrinath Acharya: The AI-defined vehicle is a constantly connected data machine. Instead of trial-and-error on a test track, innovation starts with simulation-first: build and break ideas in an emulator, then graduate to a high-fidelity digital twin in the cloud that mirrors the car’s real responses. This unlocks edge-case coverage that no field-tracks can match and that in-turn feeds the large datasets AI needs.

Safety and comfort hinge on perception models. At highway speeds, the vehicle must correctly identify objects ahead (and behind while reversing). Cutting corners here leads to test failures and safety risk. The fix is a “left-shift” workflow: train and validate those models in the cloud, then run the exact behavior on the vehicle ECU. 

Open ecosystems like Autoware and ARM SOAFEE help teams pool talent and data, while cloud–hardware parity (e.g., AWS Graviton instances matched to in-car profiles) keeps what works in simulation working on the road.

All of this depends on a standardized OTA/data pipeline. Keep it robust, low-cost, and shared across the industry so teams focus on applications—not reinventing OTA. The eSync Alliance shows how standardization can scale securely across OEMs. 

When pipelines are common, updates are reliable, rollbacks are safe, and fleets learn together—avoiding a Tower of Babel where every AI program speaks a different dialect. So, the industry needs to differentiate on the vehicle features and the apps but align on the data plumbing.

DQ: Can you share implementation examples of how this is unfolding with your clients in India, and globally?

Shrinath Acharya: Excelfore's journey began by transforming the eSync open standard into a complete end-to-end pipeline. As customer demand accelerated, we evolved from robust implementation to modular components and microservices architecture.

eSync's distributed design, pushing complexity to cloud or edge devices, enabled clean separation of cloud and vehicle concerns. Open-sourcing in-vehicle device code gave OEMs and Tier-1 suppliers complete transparency, while our generic, scalable central orchestrator adapted seamlessly to any deployment environment, aggregating device registrations through respective agents.

As SDV requirements emerged, active AWS collaboration created a continuous learning loop: data aggregation, analytics, remote diagnostics, real-time fixes, and software updates. This became the foundational enabler for software-defined vehicles. Standard interfaces now connect vehicle data curation (Agentic AI) with cloud data lakes (Generative AI), completing Excelfore's evolution toward AI-defined vehicles.

China proved transformative! In 2018, OEMs demanded rapid, cost-effective OTA implementations without legacy constraints. Our first deployment launched in twelve months (March 2018 contract to April 2019 rollout). The next OEM compressed timelines to nine months. Today, Excelfore launches complex new vehicle programs in 4-8 months.