Qualcomm Is Remaking Itself Into an AI Company. Its Shares Look Dirt Cheap.

Qualcomm bought AI software company Modular in an all-stock deal valued at about $4 billion and secured a multi-year agreement with Meta for data center processors. Management now expects more than $15 billion in AI infrastructure revenue by fiscal 2029, up from essentially nothing today, while the stock trades at just 21x earnings versus a 44x tech-sector average. The article frames Qualcomm as an attractive, undervalued AI beneficiary, though execution on the Meta deal and Modular integration remains a key risk.

Analysis

The real second-order shift is that Qualcomm is moving from being a cyclical handset supplier to a claim on the compute stack that matters most for AI economics: inference at scale. That matters because inference is where capex discipline eventually shows up, and hyperscalers will keep looking for lower-TCO alternatives to GPU-heavy deployments if utilization and power become binding constraints. If this transition works, the market will re-rate Qualcomm less like a modem/IP vendor and more like a diversified edge-and-infrastructure compute platform.

The acquisition of Modular is strategically important less for near-term revenue and more for software lock-in. In AI hardware, the durable margin pool usually accrues to the layer that makes heterogeneous silicon easy to deploy; if Qualcomm can meaningfully simplify model portability across chips, it can reduce customer switching costs and make its hardware “good enough” rather than best-in-class. That is the competitive threat to incumbents: not a frontal assault on Nvidia’s training moat, but a wedge into heterogeneous inference where software abstraction can compress pricing power over time.

The main risk is timing mismatch. The stock is discounting an AI re-rating now, but the revenue ramp is likely lumpy over the next 6-18 months, with execution hinging on whether design wins convert into meaningful utilization and whether software integration actually improves developer adoption. Any delay in hyperscaler deployment, or evidence that workloads stay GPU-centric longer than expected, would likely trigger a sharp multiple reset because the valuation already reflects improved narrative quality.

Consensus appears to be underestimating how much of the upside is already in the price over a 3-month horizon, while still underestimating the optionality over 2-3 years. The setup is not “cheap stock with no story”; it is “credible story with a still-reasonable multiple,” which is usually fine unless the market starts demanding proof on a quarterly basis. That creates a window where the best expression may be a selective long versus more crowded AI names rather than an outright momentum chase.

Key Decisions for Investors

• Go long QCOM on a 3-6 month horizon, but size modestly: the setup offers re-rating potential if AI design wins start showing up, with downside cushioned by a still-reasonable multiple versus megacap AI peers.
• Pair trade: long QCOM / short NVDA into the next 1-2 quarters if the market continues rewarding inference and edge compute exposure over training exposure; this is a relative-value expression on AI capex diversification, not a bearish NVDA outright call.
• Buy QCOM Jan-2027 calls or call spreads on any 5-8% pullback: the asymmetry is that hardware/software integration can take quarters to prove, but if adoption is real, the stock can rerate before fundamentals fully appear.
• Hedge with a tight stop below the post-announcement support area: if hyperscaler traction or Modular integration commentary disappoints, the multiple can compress quickly because expectations are now elevated.
• For more risk-tolerant accounts, pair QCOM long with a small short basket of overextended AI beneficiaries that have no inference leverage; this captures the market’s likely rotation toward cheaper, more utilitarian AI exposure.