Qualcomm CEO Cristiano Amon is making one of the boldest claims in mobile computing: AI agents will become the next major interface and reduce the role of apps as the center of digital life.
The argument is not that apps disappear overnight. It is that people may stop thinking in app-by-app steps. Instead of opening a banking app, checking an account, switching to a calendar, opening a travel app, searching messages, and copying information across screens, a user would ask an agent to handle the task across services.
That is the shift Qualcomm wants to power. The company built its business around mobile processors, modems, wireless systems, Android flagships, cars, PCs, wearables, and extended-reality devices. Now Amon is positioning Qualcomm for a future where the phone is not only a screen filled with icons, but a local AI machine that understands context, acts across apps, and follows the user across devices.
In a recent CNBC interview and a TIME essay published this month, Amon described agentic AI as the next major computing transformation, comparing it to earlier platform shifts such as the internet and the smartphone. His view matches Qualcomm’s business interests, but it also captures a real question facing Apple, Google, Microsoft, Meta, and every major software platform: what happens when the assistant becomes more important than the app icon?
Apps May Become the Background Layer
The app model has defined smartphones since the first iPhone era. Users pick an icon, open a service, complete a task, close it, then move to another app. App stores, notifications, subscriptions, in-app purchases, widgets, and privacy permissions all grew around that behavior.
AI agents challenge that model because they start with intent instead of location. The user does not begin with “open this app.” The user begins with “book this,” “compare these,” “send that,” “summarize this,” “move this meeting,” “find the receipt,” or “prepare a trip.”
In that world, apps do not vanish. They become service layers. The agent may still need a bank, airline, calendar, messaging app, email provider, map, shopping account, or document editor. The difference is that the user may not open each one manually.
This is the platform risk for app makers. If agents become the front door, the app interface becomes less visible. A banking app could still process payments and expose account data, but the user might interact through an agent. A shopping app could still hold inventory and checkout, but the agent could compare prices and place an order. A travel app could still manage reservations, but the agent could assemble the itinerary.
That is why Amon’s comment is bigger than a prediction about phones. It is a prediction about control.
Qualcomm Wants the AI to Run Near the User
Qualcomm’s interest in agents is tied to hardware. The company wants AI to run on devices, not only in remote data centers.
That plays directly to Qualcomm’s strengths. Snapdragon chips already power smartphones, PCs, XR headsets, smart glasses, cars, industrial devices, and connected products. If agents need cameras, microphones, sensors, modems, neural processors, battery efficiency, and constant connectivity, Qualcomm has a strong reason to argue that the agent platform belongs at the edge.
A cloud-only agent can be powerful, but it may be slower, more expensive, less private, and less aware of local context. An on-device or hybrid agent can respond faster, use sensors directly, work with personal data more securely, and continue operating across hardware categories.
This is why Qualcomm keeps talking about the move beyond phones. Amon has pointed to smart glasses, wearable pins, camera-equipped earbuds, PCs, cars, and other AI-powered devices as part of the next wave. The agent is not only an app on a phone. It is a layer that can follow the user from one device to another.
That vision also explains Qualcomm’s work with Microsoft on agent-first computing concepts. Microsoft’s Project Solara, announced around Build 2026, is built around devices that place agents ahead of traditional app interfaces. Qualcomm and MediaTek are involved as silicon partners for reference designs, showing how chip companies want a role in defining this next hardware category.
Apple’s Answer Is Different
Apple is not using the same language as Qualcomm, but it is moving toward a similar destination from another direction.
WWDC26 showed Apple’s version of agent-first computing through Siri AI, Apple Intelligence, App Intents, Foundation Models, Private Cloud Compute, and deeper app actions. Apple is not saying apps are over. It is trying to make apps more available to Siri and system intelligence.
That distinction matters. Qualcomm talks about agents replacing apps as the main interface. Apple is trying to make the assistant work through apps without weakening the app ecosystem that defines iPhone.
App Intents is the bridge. Developers can expose actions and content to Siri, Shortcuts, Spotlight, widgets, and Apple Intelligence. A task app can expose creating a task. A travel app can expose itinerary details. A finance app can expose safe account actions. A photo app can expose editing tools. Siri AI can then act through those app-provided structures.
This is Apple’s safer version of the agent future. The app remains trusted. The developer remains visible. The system assistant becomes more capable, but Apple still controls permissions, privacy, and platform rules.
That may be less radical than Qualcomm’s framing, but it could be more realistic for iPhone users. Apple does not need to kill apps. It needs to reduce the number of times users must manually jump between them.
Google and Microsoft Are Moving Faster
Google and Microsoft are pushing aggressively into agentic computing. Google has Gemini across Android, Workspace, Search, Chrome, and developer tools. Microsoft is building Copilot into Windows, Microsoft 365, Azure, Teams, and new agent-first device concepts.
Microsoft’s Project Solara is especially relevant because it imagines devices built around agents from the start. Instead of treating AI as a feature inside an existing app grid, Solara points toward hardware where the agent shell becomes the main interface. Qualcomm’s participation gives Amon’s comments more weight because the company is not only talking about a theory. It is working on reference hardware and chip-to-cloud platforms that could support it.
Google’s position is different but just as strong. Android already reaches billions of devices, and Gemini gives Google a way to connect search, apps, personal data, cameras, and cloud models. If agents become the new front door, Google wants Gemini to own that layer on Android and beyond.
This puts Apple under pressure. The company has the most integrated consumer ecosystem, but it has moved more cautiously. Siri AI needs to prove it can do real work, not only answer more naturally. If Apple’s assistant cannot act reliably inside apps, users may turn to Gemini, ChatGPT, Claude, or other agents for more complex tasks.
The next interface battle may be less about which app a user opens and more about which agent the user trusts.
Why Apps Will Not Disappear Quickly
The “agents replace apps” idea is powerful, but apps are not going away soon.
Apps provide identity, branding, permissions, payments, customer support, data structure, visual interfaces, legal terms, and specialized workflows. A banking app is not only a screen. It is a regulated service. A health app is not only a data viewer. It handles sensitive information. A travel app manages bookings, loyalty accounts, refunds, and policies. A creative app gives users tools that require visual control.
Agents can simplify many tasks, but they still need reliable connections to services. They also need user trust. Letting an agent read a balance is different from letting it move money. Letting an agent suggest a flight is different from letting it book one. Letting an agent draft a message is different from letting it send the message without review.
That means apps may shift from daily destinations to permissioned service endpoints. Users may visit them less often, but the app and its backend still matter.
This is similar to how websites changed after apps arrived. Websites did not disappear. They became one layer in a larger system. Apps may face the same fate in the agent era.
The App Store Question Becomes Complicated
If agents become the main interface, app stores face a new challenge.
App stores are built around discovery, installation, updates, purchases, ratings, subscriptions, and developer relationships. If users start asking agents to complete tasks directly, the app store may become less central to daily behavior.
That could affect Apple more than Qualcomm. Apple’s services business benefits from the App Store economy, in-app purchases, subscriptions, and developer distribution. A world where agents handle tasks across services could weaken the importance of app discovery and individual app engagement.
Apple’s likely response is not to resist agents, but to make agent actions flow through its own platform rules. App Intents, Siri AI, Apple Intelligence, and the Foundation Models framework give Apple a way to keep apps inside the system while making them accessible through natural language.
The company can say: apps are still the trusted containers, but Siri can operate across them.
That may preserve the App Store’s role while making iPhone feel more agentic. It also gives Apple control over privacy prompts, permissions, purchases, and user consent.
Qualcomm’s Hardware Bet Needs Better Devices
Amon’s vision also depends on hardware that users actually want.
AI pins, smart glasses, camera earbuds, and ambient assistants sound promising, but the market has already seen failed or awkward attempts at replacing the phone. The smartphone remains dominant because it combines screen, camera, apps, payments, identity, messaging, maps, browser, and entertainment in one device people trust.
For agents to move beyond apps, the hardware must be useful, comfortable, affordable, private, and reliable. Smart glasses need all-day wearability. AI pins need a reason to exist beyond novelty. Earbuds with cameras raise social and privacy questions. Desk agents need to be more useful than a laptop, phone, or smart speaker.
Qualcomm can provide chips and connectivity, but the product experience will decide whether the agent platform becomes mainstream. The agent future needs hardware that feels natural, not intrusive.
This is where Apple still has an advantage. If Apple chooses to expand agentic AI through iPhone, Apple Watch, AirPods, Mac, Vision Pro, and future wearables, it can do so through devices people already use. Qualcomm’s ecosystem is broader but more fragmented, depending on OEMs to build products that consumers want.
Trust Will Decide the Agent Platform
AI agents require more trust than apps because they ask for more authority.
An app usually works inside a limited space. A banking app handles banking. A calendar app handles events. A messaging app handles communication. An agent may need to move across all of them, remember context, read private data, make decisions, and take action.
That creates new risks: wrong actions, hallucinated information, bad purchases, privacy leaks, account confusion, and unclear responsibility when something goes wrong.
Trust will decide which companies win. Apple will lean on privacy and device integration. Google will lean on search, Android, Gemini, and cloud intelligence. Microsoft will lean on enterprise workflows and productivity. Qualcomm will lean on chips, connectivity, and distributed AI hardware. OpenAI and Anthropic will lean on model quality and developer adoption.
The strongest agent platform will not only be the smartest. It will be the one users trust to act.
That is a harder problem than replacing app icons.
Apps Become Less Visible, Not Less Valuable
Amon’s prediction is best understood as a shift in visibility. Apps may become less visible to users, but more valuable behind the scenes.
An AI agent still needs services to act on. It needs inventory, maps, payment rails, identity systems, records, calendars, messages, photos, documents, health data, and media libraries. Apps and cloud services provide much of that structure.
The change is that the user may not care which app performed each step. They care whether the task got done correctly. That is uncomfortable for developers because brand loyalty could weaken when an agent abstracts the interface.
It also creates opportunity. Apps that expose useful actions to agents may become more valuable. Apps that refuse integration may become less convenient. The future app may be judged less by its home screen icon and more by how well it works with the user’s assistant.
That is already visible in Apple’s App Intents strategy, Google’s Gemini integrations, Microsoft’s Copilot connectors, and OpenAI’s agent tools.
Qualcomm Is Betting on the Layer Beneath the Interface
Qualcomm does not own iOS, Android, Windows, Gemini, Siri, ChatGPT, or Claude. Its bet is underneath them.
If agents become the next computing platform, they will need chips that can run models locally, handle sensors, protect personal data, connect across networks, and manage power efficiently. That is Qualcomm’s pitch. The company wants Snapdragon and its related platforms to become the hardware base for agentic devices.
This is also why Amon’s comments should be read as both prediction and business strategy. Qualcomm benefits if the world moves from app-first phones to AI-first devices, because that could create new categories of hardware beyond the smartphone replacement cycle.
Smart glasses, AI PCs, cars, wearables, home devices, industrial tools, and mixed-reality headsets all need processors. If each becomes an agent interface, Qualcomm has more places to sell chips.
The company does not need to own the agent. It needs to power the device where the agent runs.
The Next Platform War Starts at the Interface
The app era is not ending this year. It may not end this decade. But the interface is changing.
Users already ask AI systems to summarize, plan, write, search, organize, code, translate, compare, and troubleshoot. The next step is action. Once agents can reliably act across services, the app icon becomes less central.
Qualcomm’s CEO is saying the quiet part out loud: the next platform may not be an operating system, app store, browser, or phone screen. It may be the agent that understands what the user wants and coordinates the services needed to complete it.
Apple’s response will be Siri AI. Google’s will be Gemini. Microsoft’s will be Copilot and agent-first devices. Qualcomm’s will be the chips and hardware platforms that make those agents portable, local, and constant.
Apps will not vanish. They will move behind the assistant. The companies that control that assistant will control the next layer of computing.