This AI Chip Stock Just Signed Massive Deals With 3 Hyperscalers, and It Still Looks Like a Great Buy Right Now (Hint: Not Nvidia or Intel)

Qualcomm expects its data center chip revenue to rise from “practically nothing” last year to $5B in fiscal 2027 and $15B by 2029, after signing three major hyperscaler deals (including $1B+ contracts for at least two hyperscalers and next-gen AI-agent CPUs for Meta starting H2 2028). Management targets EPS climbing to more than $18 by 2029 on ~18% annualized growth, arguing the stock is undervalued at <18x forward earnings despite the new AI CPU revenue stream. Competitive context is Intel’s current $5.1B data center/AI sales last quarter and Nvidia’s broader compute ecosystem, but Qualcomm highlights scaling leverage for earnings growth.

Analysis

This is better framed as an early signal that hyperscaler inference economics may shift from GPU-only stacks toward a more balanced CPU + custom-silicon architecture. That helps QCOM because its edge is power-efficient design wins and integration, but it is structurally negative for legacy server CPU share at INTC, which is already fighting for relevance in a market that increasingly prefers tailored silicon over off-the-shelf parts. The second-order effect is that some of the incremental AI spend migrates from the most expensive accelerators into orchestration and control layers, which can temper the multiple on NVDA if the market starts pricing a lower accelerator content per deployed agent.

The catch is timing: the meaningful revenue is back-end loaded, so the near-term stock response depends on whether investors believe these are repeatable platform wins or isolated engineering engagements. If QCOM does not show a visible pipeline conversion in the next 2-3 quarters, the story risks being dismissed as TAM theater, and the stock can fall back to handset-cycle valuation. The biggest falsifier is any slip in management commentary around data-center revenue contributions or a lack of backlog visibility by the next earnings cycle.

Consensus may be underestimating mix dilution. A chip business growing from a low base can lift reported EPS, but if the incremental revenue carries materially lower margin than QCOM’s licensing economics, the market may have to mark down the quality of earnings even as the top line improves. The right lens is not just absolute revenue by 2029, but whether QCOM can industrialize this into a reusable CPU/IP franchise rather than a handful of bespoke hyperscaler builds; if it can, the rerating could come before the revenue does.

Key Decisions for Investors

• Long QCOM / short INTC as a 6-12 month relative-value trade: the market is likely to award QCOM optionality for data-center share gains while INTC bears the more direct socket-loss risk; stop if INTC re-accelerates DC AI revenue or QCOM shows no backlog conversion by the next two quarters.
• Use pullbacks to accumulate QCOM rather than chase it: best entry is after the next earnings print or any post-rally fade, since the catalyst path is 2027-2029 and the stock may trade on skepticism before fundamentals catch up.
• Small tactical hedge: trim NVDA exposure on any further rally if management commentary across the sector confirms a heavier CPU mix for agentic workloads; this is a hedge against narrative compression, not a full short thesis.
• Watch META as a longer-dated beneficiary of lower inference cost, but do not overtrade it now: the 2028 ramp is too far out for near-term EPS impact, so it is more of a strategic watchlist item than an immediate position.