For decades, technological innovation has followed a similar pattern: computers become more powerful, hardware grows smaller, and new user interfaces are born, changing the way people interact with their devices. Each evolution has made technology more accessible and fueled innovation.
Take the mouse, which made computers accessible to everyone and paved the way for a graphical user interface. The browser brought the internet to the masses. Or the smartphone, which put a computer in our pockets and gave rise to the app economy. Each redefined computing for its era, expanding what people could do with technology. Today, computers are in cars, watches, smart glasses, earbuds, cameras, and more.
Now, we’re on the cusp of another inflection point: AI is becoming agentic, from responding to prompts to acting and coordinating on your behalf. I believe this evolution will not just be the defining transformation of the year, but of the next decade.
Going forward, AI agents will act increasingly independently. They will reason through multiple steps, draw on personal and environmental context, and take action—booking, coordinating, deciding, and adapting as conditions change. And they will do that by drawing on intelligence distributed across the device (or multiple devices), the network edge, and the cloud, depending on what the task requires and where it can be executed most efficiently.
Consider what this means in practice. Instead of opening multiple apps to plan a trip, an agent could handle the itinerary, check your calendar, book the flights, and make calls on your behalf to adjust a conflicting appointment. AI agents could also replace many of the apps you use if you grant the agent the right credentials and permissions. On a PC, an agent could work across your files and applications to complete multi-step tasks, assembling reports, extracting data, organizing documents, and optimizing workflows.
Agentic operating systems will extend these capabilities by unifying workflows, communications, and applications into a single interface that anticipates your needs before you ask.
This is poised to be one of the biggest transitions yet, and the pieces are starting to come together.
AI models are improving quickly. The agentic ecosystem is maturing, with on-device orchestrators like OpenClaw and Hermes, agentic assistants like Claude Desktop, and cloud platforms like Perplexity Computer all seeing rapid adoption. — and many already running on devices. Major tech ecosystems are building agents into their operating systems across phones and PCs, while major AI companies and new entrants are creating agentic operating systems from the ground up. Platforms and devices are being redesigned specifically for these new AI-based experiences, with new form factors emerging, such as smart glasses and other personal AI devices.
What many people still underestimate is that the hardware inside your device will matter more, not less, and their current devices were not designed for agentic AI and its evolution.
Today’s smartphones, PCs, vehicles, smart wearables, and more were built for a world centered on apps and human-directed interaction. Now, agents also need to operate efficiently on your device. They must run continuously in the background, fuse sensor data into context, and orchestrate multi-step tasks reliably and securely. This requires strong CPU performance for orchestration, power-efficient NPUs for local models, and greater contextual awareness and efficiency—all while preserving responsiveness and battery life. These demands will drive an upgrade cycle across silicon and software, spanning device categories.
Additionally, as agents become more pervasive, capable, and trusted, they will become the center of your digital life. Where your smartphone once took center stage, extending functionality to wearables such as smartwatches and earbuds, and pulling in their sensor data, the agent will act seamlessly across them. It will turn them into personal AI devices with new functionality, enabling you to interact with them independently.
Agents are also defining the true economics of AI. The adoption of AI agents is accelerating. Companies spent $1.5 trillion on AI globally in 2025, a figure expected to cross $2 trillion in 2026. Agents are a major reason why. They drive token consumption and computing costs exponentially, consuming 5 to 30 times more tokens than a simple chat interaction. As these systems improve, the demand for AI computing will be immense, requiring every available computing engine. But by distributing intelligence across cloud and edge—processing workloads where it’s most efficient—you can create an entirely new cost equation, and a far more sustainable one.
It’s a time of profound change, and one of incredible opportunity for companies that get it right.
There are few opportunities to be part of a generational technology transition. Like the internet and the smartphone, agentic AI will drive the next major paradigm shift. It will reshape business models, create entirely new categories of products and services, and make technology more intuitive than ever.
Qualcomm is one of the few companies with the technology assets across edge devices and the cloud to drive this next phase of AI. I know my team and I will be moving quickly to shape what comes next.
