Qualcomm’s CEO says AI agents will be the new app
AI agents are poised to replace traditional smartphone applications as the primary interface for digital tasks, according to Qualcomm CEO Cristiano Amon. Speaking on CNBC’s The Tech Download podcast, Amon predicted that 2026 will be the “year of agents,” as software vendors integrate orchestrators capable of managing bank transfers, meeting schedules, and automated payments directly on behalf of users.
How will AI agents change the smartphone experience?
The transition shifts the smartphone from the center of the user’s digital ecosystem to a supporting role, acting as one of many “planets” orbiting the AI agent. Amon stated that while apps are not disappearing, they will evolve into back-end services that agents call upon to complete specific tasks. This model prioritizes the AI’s ability to pull context from various sources—such as emails, calendars, and real-time sensor data—to execute complex workflows without requiring the user to open individual applications.
What happens to hardware as AI moves to wearables?
Qualcomm is currently developing over 40 distinct AI-focused device designs, ranging from smart jewelry and pendants to earbuds equipped with cameras, according to Amon. The industry trend favors wearable form factors that can capture audio and video to provide agents with real-time environmental context. Amon specifically highlighted smart glasses as a major growth area, noting that annual shipments could rise from tens of millions today to hundreds of millions within two years, potentially challenging the 1.2 billion smartphones shipped annually.
Why are non-traditional firms entering the hardware market?
The push by AI companies to control hardware stems from a desire to own the “endpoint”—the physical device that facilitates the user’s interaction with the AI. OpenAI, which is collaborating with Qualcomm on agent-first phone technology, acquired Jony Ive’s hardware startup, io, for $6.4 billion last year. By owning the device, AI firms gain access to massive streams of high-fidelity video and audio data, which Amon noted is “exponentially larger” than the data currently used for model training. This data access creates a competitive advantage for building personalized services but raises significant privacy concerns regarding constant, ambient data collection.
Market Comparison: The Smartphone vs. The Agent
| Feature | Smartphone Model | Agent-Centric Model |
|---|---|---|
| Central Hub | The Phone | The AI Agent |
| Primary Interaction | Manual app navigation | Verbal or intent-based commands |
| Market Structure | Closed/Integrated (e.g., Apple) | Horizontal/Open ecosystem |
Will memory shortages stall the AI revolution?
The shift toward localized AI agents faces a significant supply chain hurdle: a global memory shortage. Amon described the current demand as exceeding supply, comparing the situation to the pandemic-era bottlenecks. He indicated that capacity may not stabilize until the latter half of 2027, with some industry projections extending into 2028. Despite these constraints, Qualcomm is aggressively expanding its footprint, including reports of potential acquisitions of AI-chip startup Tenstorrent, valued at $8 billion to $10 billion, to bolster its data-center chip roadmap.
Frequently Asked Questions
- Are apps going away entirely? No, Qualcomm’s CEO stated that apps will persist but will transition into services that AI agents utilize in the background.
- What is an “agent orchestrator”? It is a software layer within an operating system that manages and delegates tasks across different applications to satisfy a user’s request.
- Why are AI companies buying hardware firms? To own the “endpoint” and capture the rich, ambient data—audio and video—that these devices collect, which is necessary for training more advanced AI models.
Have you started using AI agents for your daily tasks, or do you still prefer manual app navigation? Share your thoughts in the comments below or subscribe to our newsletter for the latest updates on the future of mobile computing.