At AI Infra Summit 2026, Dr. Vinesh Sukumar, VP of Product Management at Qualcomm, explored the system innovations powering the next generation of on-device and hybrid AI. From model optimization and edge inference to agentic workflows and hybrid edge-cloud architectures, he shares how Qualcomm is helping scale AI across devices ranging from smartphones, PCs, wearables, IoT, and other edge platforms. • Why Qualcomm believes 95% of future AI tokens will be generated on-device, and the challenges of making edge AI practical at scale • How techniques like quantization, speculative decoding, distillation, and heterogeneous computing are enabling large language and reasoning models on resource-constrained devices • The future of hybrid AI, including personalized agents, dynamic edge-cloud routing, and distributed inference across phones, PCs, wearables, and beyond Learn more about Qualcomm’s approach to scalable AI infrastructure and the path toward intelligent, personalized experiences across the edge and the cloud: https://www.qualcomm.com/artificial-intelligence

 7703      130