Agentic Artificial Intelligence (AI) is reimagining the in-vehicle experience. AI agents embedded in the digital cockpit enhance services that the automotive industry can monetize. Moreover, Agentic AI improves existing vehicle features that drivers seldom use. Mercedes-Benz’s recent integration of an AI work agent into premium models epitomizes the potential to use Agentic AI to strengthen customer loyalty and create new revenue streams.
While Agentic AI is a premium vehicle offering today, ABI Research forecasts widespread commercialization over the next 10 years as costs come down and consumer awareness expands. Our recent study indicates that vehicle shipments with Agentic AI will explode from 5 million in 2025 to almost 70 million in 2035.
But successfully commercializing and monetizing agentic systems will require careful consideration of the architectural backbone of vehicle compute. Both the edge and cloud have pros and cons, making hybrid architecture the go-to solution.
Key Takeaways:
- Agentic AI creates new revenue opportunities for automakers. It enables paid services in navigation, maintenance, productivity, and concierge features that strengthen customer loyalty.
- The vehicle’s compute architecture determines how Agentic AI can be monetized. Edge, cloud, and hybrid models each affect performance, cost, and pricing strategies.
- The industry is moving toward edge-first AI for speed, safety, and privacy. Local processing reduces latency, limits cloud dependency, and becomes more affordable as Neural Processing Units (NPUs) evolve.
- Cloud workloads will support Agentic AI but not lead it. Cloud compute will remain useful for updates and large language tasks, but Original Equipment Manufacturers (OEMs) want to avoid heavy cloud costs and uptime risks.
- Strong technology partnerships will be essential. Automakers must work closely with chipset vendors, hyperscalers, and AI developers to build seamless, scalable in-vehicle AI ecosystems.
Agentic AI Use Cases for Automotive
While basic AI personal assistants have been integrated into vehicles for years, Agentic AI offers more compelling driver experiences by contextualizing user behavior and communicating in a conversational manner. At the same time, consumers are not enthusiastic about paying for basic functionalities like remote starters. Instead, they favor value-add services that solve their problems. For example, 67% of connected car service subscribers tell Smartcar that they are willing to pay more for additional features. Things like maintenance reminders and advanced safety will help move the needle.
Agentic AI holds the key to automakers meeting these consumer expectations. Notable use cases that the automotive industry can commercialize include:
- Navigation and Exploration: Agentic vehicle systems can make personal recommendations based on driver behavior and advanced analytics. In natural language, the agent will suggest alternative routes, nearby charging stations, brand-preferred fueling partners, and more. Automakers should emphasize the safety benefits gained from reducing manual interaction from the driver.
- Vehicle Maintenance: Recent survey results from car seat manufacturer Diono indicate that just 7% of U.S. drivers are “extremely confident” they can correctly identify all vehicle warnings that appear on their dashboard. Drivers can ask the agent what a warning light means, if it’s safe to continue driving, or where to find a repair shop. This use case helps direct customers to have repairs done within the OEM’s service network.
- Digital Concierge: In-vehicle agents help drivers manage daily tasks through contextual awareness. The agent can do anything from booking a hotel reservation to adjusting entertainment preferences based on prior usage. OEMs must balance simplicity of the digital concierge with safety and reliability standards.
- In-Car Productivity: Modern agentic systems streamline vital communications on behalf of the driver. The agent can summarize calls, read and respond to critical messages, and modify calendar events. AI agents promote safety by allowing drivers to complete these tasks without having to look away and at their smartphones.
- Symbiotic Driving: The next phase of in-vehicle AI is a harmonious co-existence between agent and driver. While Advanced Driver-Assistance Systems (ADAS) already understand the vehicle’s surrounding environment, Agentic AI enables advanced driver input; for example, requesting the vehicle to drive in a different style or explaining why the ADAS responded in a certain way. Beyond being a hands-off experience, companies like Wayve and Qualcomm have developed solutions that enable the agent to explain why it takes certain actions. This builds transparency and encourages repetitive use.
Agentic AI Commercialization Hinges on the Automaker’s Underpinning Architecture
These use cases can be monetized through a variety of models, including revenue sharing, subscription-based access, or regional and behavior-based pricing. As an automotive OEM, selecting the right monetization model boils down to where the Agentic AI system runs tasks: at the edge, in the cloud, or both? Edge-first platforms tend to favor usage-driven or event-based pricing, while cloud-first platforms are typically tied to recurring costs (e.g., connectivity, storage, updates, etc.).
The vast majority of vehicles shipping with Agentic AI leverage a hybrid architecture, combining the best of both worlds. This allows vehicle OEMs to balance cost and performance. The three types of hybrid Agentic AI architectures that OEMs use are:
- Hybrid Edge-First: Agents perform most tasks locally on the vehicle’s chipset or chipset component. The cloud is reserved primarily for non-critical tasks like synchronization and Large Language Model (LLM) updates.
- Hybrid Balanced: Onboard compute and cloud computing share task execution. Whereas time or data-sensitive applications run at the edge, larger language or planning tasks are executed in the cloud.
- Hybrid Cloud-First: Most agentic and LLM workloads run in the cloud. Onboard compute manages sessions, fallback behavior, and safety buffers for intermittent connectivity.
Commercialization Is Increasingly Driven by In-Vehicle Edge
Hybrid balanced architecture is and will continue to be the most common choice for in-vehicle Agentic AI deployments through 2035. However, ABI Research expects the automotive industry to increasingly gravitate toward hybrid edge-first architecture over the next decade. This fact should be top-of-mind for OEMs devising Agentic AI monetization plans. The chosen underlying architecture has a direct impact on how customers interact with the digital cockpit and the cost to do so.
China is already ahead of the curve with edge AI vehicle deployments, driven by government preference for local processing and the country’s willingness to quickly embrace new chipset designs. In North America and Europe, OEMs are more cautious about running automotive tasks at the edge. They currently prefer to use hybrid balanced architecture to combine local autonomy with continuous cloud synchronization. Edge-first deployments are typically exclusive to premium vehicle models, although chipset innovations will eventually bring edge-first capabilities to mid-range vehicles.
While Western OEMs currently prioritize hybrid balance, we believe they will rapidly adopt edge-first AI after 2028. Behind this strategic shift is a growing need for low latencies, data privacy concerns, and Neural Processing Unit (NPU) cost reductions.
Edge-first Agentic AI-capable vehicles are also a preferred choice in regions with poor 5G coverage. Without low-latency connectivity to execute critical tasks, automakers are more likely to lean on edge processing.
At the other end of the spectrum, hybrid cloud-first architecture will only account for 10% of Agentic AI deployments in vehicles by 2035. In the future, the cloud will mostly be used to issue Over-the-Air (OTA) software updates. Moreover, OEMs want to avoid the complexities of supporting uninterrupted connectivity and hyperscaler inference costs.
Final Thoughts
Add-on services facilitated by Agentic AI, such as in-car shopping and productivity enhancements, create opportunities for automotive OEMs to increase their bottom line. Most of these new use cases will rely on both the edge and the cloud, with a growing reliance on the edge in the 2030s. The edge enables mission-critical use cases that can be further monetized. The cloud complements the edge by providing valuable OTA software updates, delivering new features, and other core functionalities.
Combining edge, cloud, and LLM intelligence into a unified ecosystem that can scale globally is essential to successful commercialization. OEMs must ensure that agentic assistants can work seamlessly alongside third-party ecosystems like navigation apps and workplace collaboration tools. This complicated process necessitates a restructuring of partnerships between automakers, semiconductors, hyperscalers, and AI developers.
Download the reports below for a deeper exploration of how Agentic AI is shaking up the automotive OEM playbook:
