AI and the software-defined vehicle have become part of the same industry narrative. They are not the same transformation. SDV is an architectural, organizational, and business change; AI is a capability that runs across it. Treating them as one has produced a lot of announcements and comparatively few results.   This session focuses on what OEMs have actually deployed. Drawing on SBD Automotive's in-vehicle testing, benchmarking, and direct work with OEM programs worldwide, it uses specific examples across cockpit experiences and personalization, software development and validation, OTA operations, diagnostics, cybersecurity, and new digital services to show what shipped, what it changed, and what it took to get there. Alongside the successes, it covers where deployments have stalled and why: system complexity, AI validation, privacy, liability, regulation, and customer resistance.  Those examples are then placed on the SDV/AI flywheel, where vehicles generate data, platforms make it accessible, software and AI turn it into capabilities, and those capabilities create customer value and new data in return. SDV is the foundation and AI the accelerant, not the destination.  The session also touches on other critical questions like: who controls that flywheel, and whether the choices OEMs are making about what to own, source, or share are leading toward proprietary stacks or toward open interfaces and shared data models.  The session will challenge the COVESA community to consider where openness creates the greatest strategic value in this next phase of SDV: preserving choice, reducing integration complexity, enabling AI across heterogeneous vehicle platforms, and ensuring that innovation can happen across an ecosystem rather than within isolated technology stacks.