AI in the automotive industry is critically important. Generative AI, Agentic AI, customer support, in-car assistants. Join Autotech Council members, OEMs, automotive suppliers, mobility startups and VCs as we focus on the non-self-driving use cases of AI, and bring the startups making waves.
Silicon Valley, California, March 13, 2026/Meeting Recap/ On March 12, 2026, the Autotech Council convened to discuss the rapid advancements surrounding the AI-Defined Car. The AI-DC marks an evolution from the traditional Software-Defined Vehicle (SDV) which has been a hot topic for a decade, leveraging that SDV architecture with a control layer of artificial intelligence, and lifting AI as the primary driver of the vehicle lifecycle. The scale of this shift is immense; it touches every phase of a car’s life, from design, to manufacturing, to car systems, to self-driving, to retail, to after-sales service, to the user experience. The automotive artificial intelligence market is projected to exceed 74 billion dollars by 2030 as manufacturers race to integrate smarter edge computing.
Our meeting was hosted by Amazon Web Services autolab (AWS) in the town of Santa Clara, drawing a crowd of over 100 attendees from across the mobility and technology sectors. The agenda highlighted several key contributors who are currently shaping the future of automotive intelligence. Cassidy Shell (Vice President, Research & Strategy, Mobility Impact Partners) opened with a presentation that evaluated where AI brings the most value, ranging from customer focused operations to deep research and development. Following this, a panel discussion focused on the multi-layered uses of AI in automotive planning, vehicle design, and manufacturing. This panel featured Ryan Goff (Automotive Technology Director, Elektrobit Americas Inc.), Kiran Cherukumilli (R&D Product Technical Manager, Valeo), and Stefano Marzani (Emerging Tech Global Leader, Automotive, Amazon Web Services). These experts shared insights on the navigation of physical AI and the integration of agentic systems to improve passenger safety and the overall user experience within the cockpit.
The insights shared during these sessions emphasized that customer expectations are not leveling off but are instead accelerating toward a demand for total digital fluidity. Ryan Goff noted that users now expect a more seamless integration for Large Language Models (LLMs) and increased productivity amplified by agentic AI to serve a growing list of needs. This vision treats the car as an intelligent partner rather than a passive tool. To meet these needs, the industry is seeing a push toward building powerful intelligent systems both on-board and in the cloud. Stefano Marzani issued a call to urgency for the automotive supply chain, asserting that AI agency must have a place in every aspect of the software lifecycle. This integration is predicted to facilitate a near-80% improvement in lowering cost and time expenditures compared to legacy systems. McKinsey research supports this trajectory, noting that generative AI has moved from pilot phases to practical applications that actively collaborate with human workflows.
Technically, the meeting tackled the difficult challenge of reconciling the output of deterministic systems with probabilistic models, particularly within the scope of Advanced Driver Assistance Systems (ADAS). Kiran Cherukumilli pointed out that in the context of constrained automotive electronic architectures, simple brute compute is rarely the most efficient answer. Instead, developers must find ways to balance the reliability of traditional logic with the predictive power of AI. Stefano Marzani suggested that off-board processing in the cloud can serve as the ultimate release valve for these systems, providing the heavy lifting required for complex reasoning that exceeds the local hardware capacity. This hybrid approach ensures that the vehicle remains a safe, deterministic machine while benefiting from the flexible, creative problem-solving capabilities of modern AI models.
The current importance of this technology lies in its ability to transform the vehicle into a proactive agent. We are moving beyond command-response interactions toward systems capable of reasoning and planning. In the automotive world, this translates to AI-driven coding, generative prototyping in the design studio, and hyper-personalized buyer journeys in retail. The use of Small Language Models (SLMs) trained on specific automotive datasets is also becoming a critical trend, allowing for high performance without the massive overhead of general-purpose models.
We enjoyed our usual segment of rapid-fire presentations from great startups in the sector The startups addressed topics such as:
The five-minute pitches provided a focused look at shovel-ready applications that are transforming the automotive value chain right now, and were the start of productive business engagements during the networking breaks and the lunch session. These face-to-face interactions are vital for the mobility sector, as they allow for the immediate vetting of new technologies and the exploration of potential partnerships. The goal of the Autotech Council is to foster these business relationships and to reduce the time it takes for members and innovators to get deals done.
We would like to thank the hosts, Amazon Web Services’ Autolab (AWS) for providing the venue and technical support for this gathering. We also thank the various presenters and panel speakers for their contributions to the discussion on automotive intelligence. All of the presentations from today's meeting will be available in the Autotech Council Member Library.