Speakers at Wednesday’s COMPUTEX forums said agentic AI is driving a shift from standalone tools to an interconnected ecosystem spanning people, devices and custom silicon
As AI rapidly evolves from an experimental technology into critical infrastructure, industry leaders at COMPUTEX 2026 said the next phase of growth will be driven by agentic AI, edge computing and increasingly interconnected devices.
Speaking at the AI Devices, IoT & Edge Intelligence forum, Kedar Kondap of Qualcomm said AI adoption remains in its early stages despite explosive growth, as only 13 percent of the global population uses AI today.
AI systems are already generating around 50 trillion tokens — “the new currency” of the AI era — per day, he said.
However, Kondap said the challenge is no longer generating demand, but delivering the computing power needed to support the next 50 trillion tokens.
Kondap said neither cloud-only nor device-only approaches are sufficient, arguing instead for a hybrid model that distributes workloads intelligently between cloud and edge devices, a capability supported by Qualcomm’s latest Snapdragon X series of AI PC chips.
Qualcomm envisions a future of seamless devices, where an AI-enabled car accesses the driver’s calendar from their phone and maps the best route to their next meeting, where a project started on mobile moves directly to a PC, he said.
“This is the moment where AI stops being an experiment, AI now becomes real infrastructure,” Kondap said.
Supporting that vision, Microsoft VP of Windows & Device Sales Mark Linton said Windows is “transforming to meet the needs of the agentic era,” with new tools designed to help developers build, test and deploy AI agents more efficiently, while ensuring security and policy controls.
With more than one billion daily active users on Windows 11, the company is now pushing more intelligence to the edge to run ever-more powerful AI models, he said, with PCs that can intelligently allocate computing bandwidth between workloads.
Powering on-device, hybrid AI will bring “unmetered intelligence” across Windows devices, he said.
Meanwhile, MIPS CEO Sameer Wasson said the rise of agentic and physical AI is reshaping semiconductor design, driving demand for more open and customizable silicon platforms.
“We believe we’re at the knee of the CPU evolution,” Wasson said, predicting that AI agents running on CPUs will redefine computing architectures and accelerate the move toward custom silicon.
The industry is moving “from a walled garden to an open orchard,” he said, as companies seek greater flexibility to design processors tailored to specific AI workloads.
Together, speakers painted a picture of an industry preparing for agentic and physical AI at unprecedented scale — one in which intelligent agents, connected devices and specialized hardware increasingly work together to power the next generation of computing.
