Qualcomm CEO says AI agents will replace apps — as chip giant works on 40 new AI-powered devices

Qualcomm says it is working on more than 40 AI device designs, with CEO Cristiano Amon highlighting smart glasses, wearables, and other agent-enabled form factors as the next phase of consumer hardware. Amon said smart glasses shipments are already in the multiple tens of millions annually and could grow to hundreds of millions within a couple of years, potentially rivaling smartphones. The company is also upgrading its chip roadmap to support smaller, more power-efficient devices, signaling a strategic pivot toward AI-centric endpoints.

Analysis

The bigger implication is not “AI glasses” as a product category, but a shift in silicon demand from a single compute endpoint to a distributed ambient-compute stack. That is structurally favorable for Qualcomm because it monetizes every incremental device layer where latency, always-on sensing, and battery constraints matter more than raw cloud horsepower; it also raises the value of integrated modem + edge-accelerator packages versus commodity components. The second-order winner is likely the component ecosystem around low-power cameras, microphones, PMICs, and Wi-Fi/Bluetooth, while the loser is any platform that depends on the phone remaining the sole gateway to user attention.

For Apple, this is a medium-term strategic threat rather than an immediate earnings risk: the danger is not iPhone unit collapse, but margin dilution if the interaction layer migrates to wearables owned by third-party AI ecosystems. If agents become the primary interface, app-store economics weaken and the “default device” premium becomes less defensible, especially if smart glasses scale faster than expected over the next 18-36 months. Meta is better positioned than most because it already has distribution, social graph data, and willingness to subsidize hardware to seed usage.

The contrarian miss is that the market may be underestimating how slow consumer adoption will be outside of early adopters. Multiple form factors do not automatically translate into mass adoption; battery life, social acceptability, privacy, and killer-use-case clarity are still gating factors that can delay scale by 2-3 product cycles. That means the near-term trade is less about betting on a category breakout and more about owning the picks-and-shovels enablers while fading exuberance in hardware names whose valuation already discounts a rapid platform shift.

Key Decisions for Investors

• Long QCOM vs short AAPL into the next 3-6 months: QCOM has direct leverage to a multi-form-factor edge-compute cycle, while AAPL faces strategic risk if agents and wearables erode handset centrality; favor a modestly sized pair with the short capped if Apple doubles down on its own on-device AI roadmap.
• Add META on weakness over a 6-12 month horizon: if smart glasses are the first mass-adopted ambient AI endpoint, META benefits from distribution, data, and a willingness to sacrifice near-term margin for ecosystem control; risk/reward is attractive because the market already treats glasses as a side project rather than a core monetization vector.
• Buy QCOM call spreads 6-12 months out rather than outright stock: the thesis is a roadmap rerating as investors price in multi-device silicon demand, but the business is still exposed to handset cyclicality, so spreads capture upside from design-win optionality while limiting downside if consumer adoption stalls.
• Avoid chasing consumer hardware pure plays that need immediate breakout adoption; instead, own enablers such as analog/power and sensor suppliers on pullbacks. The trade has a better hit rate because the adoption curve can slip, but component attach rates can still compound as devices proliferate.
• Watch for a short-window reversal catalyst in Apple if it announces a convincing integrated agent + wearables strategy. If that happens, cover any AAPL short quickly; the bear case here is strategic, not numerical, and sentiment could snap back in days if the market regains confidence in ecosystem lock-in.