Ali Kani: Nvidia’s multipronged automotive strategy

In this episode of Shift, Molly Boygon interviews Ali Kani, VP of Automotive at NVIDIA, inside a moving Level 2 autonomous vehicle powered by NVIDIA's technology. The conversation dives deep into NVIDIA's three-computer automotive strategy—training, simulation, and in-car computing—emphasizing modularity and flexibility for partners. Kani explains that while some automakers like Tesla use only NVIDIA’s training infrastructure, others like Waymo, Uber, and Nissan leverage the full stack. He defends the power efficiency of GPU-based systems by highlighting performance-per-watt advantages and adaptive power scaling, countering concerns about compute intensity. The discussion also covers NVIDIA’s open-source AlphaMio model, its role in accelerating AV development, and the company’s strategy to remain relevant even if OEMs don’t use its chips. Kani addresses geopolitical challenges in China, noting strong partnerships despite export restrictions, and envisions a future where robotaxis lead the way to widespread autonomy, followed by passenger cars. He underscores that the core technological building blocks for full autonomy are now in place, with innovation now focused on faster, more intelligent simulation and testing. Key takeaways include: 1) NVIDIA’s automotive strategy is built on a modular, three-computer loop (training, simulation, in-car) that allows partners to pick and choose components; 2) The power draw of GPU-based systems is justified by superior performance-per-watt and adaptive scaling, especially for high-safety, high-capability autonomy; 3) Open-sourcing models like AlphaMio accelerates industry-wide progress while still benefiting NVIDIA’s core business in training and simulation; 4) Robotaxis are likely the first widespread deployment of autonomy, serving as a gateway for public adoption; 5) The real engineering challenge isn’t just the tech, but the human-AV interaction and regulatory balance between safety and cost. The tone is optimistic and forward-looking, with strong confidence in the trajectory of autonomous vehicle technology.