Explore the evolution and future of TinyML in this 34-minute conference talk by Qualcomm's Sr. Director of Engineering, Liang Shen. Gain insights into how power-efficient AI engines have revolutionized smart devices, enabling AI deployment on billions of battery-powered gadgets. Delve into the emerging trend of always-on and long-continuous-run AI use cases, and understand the need for optimal minimum power solutions. Examine the details of ultra-low-power AI solutions and their impact on improving targeted use case quality. Conclude by discussing ongoing challenges and potential directions for intelligent algorithm evolution, including the role of the hexagon tensor accelerator and strategic partnerships in shaping the future of TinyML.