Automotive Computing Conference 2026
AI, chiplets and virtualisation reshape vehicle computing
At the Automotive Computing Conference 2026, AI-ready E/E architectures, virtualisation, chiplets and Physical AI will be discussed as building blocks for the next generation of vehicle computing.
Ultima Media
AI, chiplets, virtualisation and new interconnects are reshaping automotive computing. At ACC 2026 in Munich, experts will discuss how these technologies are changing future vehicle architectures, development processes and the path towards production-ready systems.
Software-defined vehicles have moved from strategic ambition towards implementation, but the next architectural challenge is already taking shape. Vehicles must support increasingly capable AI functions while keeping compute performance, energy consumption, cost, software complexity and development time under control. That tension sits at the heart of the Automotive Computing Conference 2026.
On 18 and 19 November, OEMs, semiconductor companies, software providers and research organisations will meet in Munich to examine the technologies behind the next generation of vehicle computing. The agenda spans AI-ready E/E architectures, generative AI, virtualisation and chiplets through to new interconnect technologies and Physical AI.
How will generative AI reshape vehicle architecture?
Mercedes-Benz opens the first conference day with Shivam Bahedia, AI & Data Science Specialist and Program Manager for AI-Defined Vehicles, and his keynote “AI in Motion: Reinventing the In-Car Experience with Generative AI”. The session looks beyond a conventional in-car assistant towards AI that is more deeply embedded in vehicle functions and software architecture. That raises the question of what an AI-defined vehicle architecture actually requires once context, orchestration and increasingly capable models become part of the vehicle itself.
A central design decision is where intelligence should run. Some functions require local execution for latency, privacy or availability, while more complex reasoning can draw on cloud resources. Finding the balance between cloud and edge intelligence becomes particularly important when personalised functions have to work across millions of vehicles without turning connectivity or compute costs into a bottleneck.
Lars Wemme, Senior Vice President ADAS, Chassis & EE Architecture of Automotive Microcontrollers at Infineon, then shifts the discussion towards Physical AI and functional integration. His contribution focuses on scalable software architectures and edge-AI capabilities within future E/E architectures.
The implications extend well beyond adding more AI to infotainment. As AI becomes involved in perception, decision-making and vehicle functions, the requirements placed on central computers, microcontrollers, data paths and functional distribution also change. Compute, software and the physical E/E architecture increasingly have to be designed as one system.
Where is AI already changing vehicle development?
The second major strand moves AI from the vehicle into the engineering process. Florian Haubner of Google Cloud will present Horizon, an agentic innovation platform for SDV development. Elektrobit, meanwhile, will examine how software sovereignty, open source, cost and responsibilities across the supply chain can be balanced as vehicle software platforms scale.
That discussion reflects a broader shift towards agentic AI in vehicle development. AI is moving into requirements work, software engineering, testing and validation, while the surrounding toolchain has to provide the structured interfaces, data and virtual environments that agents need in order to operate reliably.
Virtual development then becomes a recurring theme across the first day. Aumovio will show how Functional Twins can represent ECU behaviour, including PCB functionality, before all physical components are available. STMicroelectronics extends that approach with virtual MCUs and virtual eFuses, allowing ECU and power-supply concepts to be explored earlier in the development cycle.
IPG Automotive will address continuous virtual vehicle development across the engineering lifecycle. The objective is to connect simulation and verification more tightly, reducing the effort required for physical testing while moving integration and validation further upstream.
An international perspective comes from ECARX founder and CEO Ziyu Shen, who will explore how AI in engineering can accelerate the deployment of AI inside the vehicle. An Executive Fireside Chat with Volkswagen’s Markus Blum will then add the OEM perspective.
How could chiplets change automotive computing?
The second conference day moves more decisively towards hardware. Martin Tietze, Vice President New Technologies Digital Car at BMW, opens the “Future of Automotive Computing” block. His specific presentation topic had not yet been published in the agenda at the time of the original announcement.
Chiplets then take centre stage. Dr Bart Placklé, Senior Vice President Automotive Technologies at imec, will examine how chiplet architectures for automotive computing could support the performance requirements of automated and autonomous vehicles, and which technical and industrial conditions still have to be solved before such designs can scale.
The following sessions move from concept towards implementation. Synopsys will focus on UCIe compliance in heterogeneous automotive chiplets. Cadence will present a configurable reference platform for scalable compute SoCs, including interoperability, safety and security considerations. Socionext will discuss the path from chiplet concepts towards central vehicle computers, while Bosch will contribute the European CHASSIS project.
Together, these contributions address a fundamental question for future automotive high-performance computers: can compute cores, specialist accelerators and interfaces be combined more modularly, rather than forcing every new performance tier into an ever larger monolithic SoC? The attraction is greater reuse and flexibility, but qualification, packaging, interoperability and ownership of the overall system remain critical.
When does optical connectivity become relevant?
More centralised architectures and higher compute performance also increase the amount of data that must move through the vehicle. A dedicated conference block therefore looks at emerging interconnect technologies rather than treating bandwidth as a secondary consequence of compute architecture.
Dr Ali Safiei of Black Semiconductor will present integrated graphene photonics as a possible optical data path for zonal backbones and chiplet links. His work on optical PCIe and graphene photonics brings data rate, EMC, wiring complexity, energy efficiency and the scalability of centralised compute architectures into the same discussion.
Wireless connectivity forms the second part of the picture. Dr Francesc Fons of Huawei and Jacob Zheng of Eagle Drive Tech will use a 360-degree camera system to demonstrate an automotive application for the short-range wireless technology SparkLink, also known as NearLink.
The combination matters because a more powerful central computer is useful only if sensors, zonal controllers, memory and compute platforms can exchange growing data volumes with sufficiently low latency. Future architectures therefore have to treat data movement as a design parameter in its own right, rather than simply increasing processing performance.
What does Physical AI add to the discussion?
The final programme block widens the scope to Physical AI: AI systems that perceive and act in the physical world. In automotive applications, that includes automated driving and sensor processing, but the concept also reaches back into the development of electronic hardware itself.
Arm and covolv.ai will address virtual ISP models and end-to-end testing for L2++ driving functions. Priyanka Mathikshara, founder and CEO of Voltai, will take a different angle, examining AI agents that can design, validate and optimise vehicle electronics. Georg Olma of NXP will close the programme by asking which lessons from Physical AI and humanoid robotics can be transferred to the automotive industry.
The Automotive Computing Conference therefore connects two developments that are often discussed separately. AI is changing what the vehicle can do, while at the same time changing how software, electronics and complete vehicles are engineered. The two conference days will focus on the architectures required to bring those developments together and on which technologies are already moving towards production readiness.
