Qualcomm CEO Calls 2026 the Breakout Year for AI Agents

Qualcomm CEO Cristiano Amon took the stage at Mobile World Congress in Barcelona with a statement that sets the tone for the industry for the year ahead: 2026 will be the year of AI agents. Not just chatbots, not just generative models, but fully autonomous systems able to act on a user's behalf in the digital world. Given Qualcomm's scale and its influence on the mobile ecosystem, these words deserve to be taken seriously.

MWC Barcelona traditionally serves as the venue where the biggest technology companies set their strategic priorities for the year ahead. If 2024 was dominated by the race to bring generative AI to smartphones, and 2025 by the first real results from local language models on mobile processors, Amon is now pointing to the next step in the evolution. AI agents are software entities that do not merely answer requests: they plan actions on their own, interact with apps and services, and make decisions within parameters set by the user. In essence, this is a shift from a tool to a digital assistant in the full sense of the word.

For Qualcomm, this forecast is not abstract futurology but a direct reflection of its business strategy. Over the past several years the company has methodically expanded the AI capabilities of its Snapdragon processors, betting on so-called edge AI: running models directly on the device, without going to the cloud. The Snapdragon 8 Elite and its server counterparts can already process models with billions of parameters locally. Agentic AI requires exactly this architecture. For a digital assistant to react instantly to context, work with personal data without sending it to remote servers, and keep functioning even on an unstable internet connection, the computation has to happen on the device.

Amon's statement carries even more weight when you look at what is happening around it. OpenAI, Google, Anthropic and dozens of startups are actively developing agent frameworks. Microsoft is integrating Copilot agents into its ecosystem. Apple is expanding Apple Intelligence. But all of these solutions need a hardware foundation, and this is where Qualcomm sees its key role. The company positions itself not as an AI model developer but as the builder of the platform on which those models and agents will run, from smartphones and laptops to cars and industrial devices. It is a carefully calculated position: instead of competing with the software giants, Qualcomm aims to become an indispensable link in the infrastructure.

It is important to understand that the move to agentic AI brings serious challenges. When a program acts autonomously, booking tickets, sending emails, making purchases, questions of security, privacy and accountability come to the fore. Who is responsible if an AI agent makes an erroneous transaction? How can you ensure an agent is not compromised by attackers? These questions do not yet have settled answers, and the industry will have to work them out while the technology is being rolled out. Qualcomm, for its part, emphasizes hardware security mechanisms built into its processors, but that may not be enough without matching software and regulatory frameworks.

For the market, Amon's forecast means faster competition on several fronts at once. Among chipmakers, Qualcomm, MediaTek, Apple and Intel will compete on AI workload performance on devices. Among ecosystems, Android and iOS will offer rival implementations of the agent experience. Between the cloud and the edge, the balance between local and server-side computing for AI agents will determine the architecture of the next generation of apps. Smartphone makers, in turn, get a new marketing narrative: where they once sold cameras and screens, the key selling point will now be the quality of the AI agent built into the device.

The Qualcomm chief's statement at MWC is not just corporate optimism. It signals that the infrastructure layer for agentic AI is approaching maturity. A technology that was discussed in the future tense only two years ago is taking on concrete hardware form. The main question remains: are users ready to trust autonomous programs with real actions in their digital lives? We will get the answer in 2026.