Qualcomm: Data center silicon and automotive scale replace the Apple checkbook

Stevie AI on QUALCOMM Incorporated (QCOM-USA | qualcomminco)

9/1/2026

Summary

Qualcomm is a fabless semiconductor company earning revenue from two engines: QCT, which designs and sells SoC platforms for handsets, automotive, IoT and now data center, and QTL, a high-margin licensing annuity on handset ASPs. The structural insight underpinning this call is that Qualcomm's growth algorithm is flipping from a single, mature, customer-concentrated handset market toward a diversified non-handset QCT base — automotive, IoT and, critically, custom silicon for hyperscale data centers — that management now sizes at over $40B by FY2029, more than double the prior $22B target. The near-term optics are messy: Apple product revenue is falling roughly 50% sequentially into the December quarter and will shrink further in FY2027, but management's own math says non-handset growth alone should more than replace that entire Apple revenue stream by FY2027, with data center revenue scaling from $5B in FY2027 to $15B by FY2029. The historical numbers tell a story of margin whipsaw rather than demand destruction. FY2024 revenue of $39.0B produced net income of $10.3B and EPS of $8.98. FY2025 revenue grew to $44.3B, yet net income nearly halved to $5.6B and EPS fell to $5.01 — a margin compression episode tied to QCT gross margin erosion and one-time cost pressures, not a revenue problem. That gap between rising revenue and depressed earnings is exactly the kind of dislocation that creates mispricing: the market is anchoring on FY2025's depressed EPS base rather than the recovery trajectory management has now guided to, with QCT gross margins moving back toward the 48%-50% baseline via double-digit pricing actions even as data center mix adds a modest 1.5-2% drag. Our price targets are built by applying a 19x forward P/E to FY2026-FY2029 EPS, a multiple consistent with a diversified semiconductor franchise generating ~29% FCF margins with a strengthening net cash position, below high-growth pure-play AI silicon names but above legacy handset-only comps given the credible data center optionality. This yields targets of $187 (FY2026), $226 (FY2027), $275 (FY2028) and $337 (FY2029) versus today's $170.48, implying meaningful upside as the EPS recovery from $5.01 (FY2025) to $17.75 (FY2029) plays out — a more than 3.5x increase in per-share earnings over four years, aided by an aggressive buyback program funded by FCF growing from $13.4B to $19.8B over the same period.

Thesis

1. **Diversification away from a single, capped royalty stream.** Qualcomm's QTL segment has always been a high-margin (69% EBT margin) but structurally capped annuity — royalties are a function of handset ASPs, not units, and growth is inherently limited by smartphone market maturity. The real strategic shift is QCT's pivot: non-handset revenue (automotive, IoT, data center) is set to grow from 24% YoY in FY2026 to over 60% YoY in FY2027, moving from under 50% of QCT revenue today to roughly two-thirds by FY2029. This is not a marginal mix shift; it re-rates the entire earnings base away from a single-customer, single-market dependency. 2. **Automotive is already a demonstrated, scaling business, not a promise.** Management is guiding to another record automotive quarter with ~60% YoY growth in Q4 FY2026, and automotive plus IoT is targeted at $24B+ by FY2029. This is real, recurring design-win revenue tied to digital cockpit and ADAS platforms where Qualcomm is positioning for the #1 revenue share position in automotive semiconductors — a category where design cycles are long and win-backs are rare, giving high visibility into the multi-year ramp embedded in the FY2027-FY2029 forecasts. 3. **Data center custom silicon is the optionality the market has not priced.** The ramp from $5B in FY2027 to $15B in FY2029 is guided directly by management following investor day disclosures, with first revenue recognition expected in Q4 FY2026 and 'performance on silicon' milestones for HBC Gen 1 expected before a mid-2027 launch. Even at a modest 1.5-2% gross margin drag on QCT, the incremental revenue contribution at scale — $15B against a company doing roughly $61-69B in FY2028-FY2029 — represents a genuinely new profit pool that did not exist two years ago, and the market's current valuation multiple does not reflect a credible new entrant into hyperscaler compute. 4. **Margin recovery is underappreciated after the FY2025 EPS air-pocket.** The drop from $8.98 EPS in FY2024 to $5.01 in FY2025 despite revenue growth from $39.0B to $44.3B was driven by margin compression, and management has been explicit that double-digit pricing actions are restoring QCT gross margins toward the historical 48-50% baseline. The forecast path — EPS of $9.87 in FY2026, rising to $17.75 by FY2029 — is not dependent on heroic revenue assumptions; it is largely a margin-normalization and buyback story layered onto steady non-handset growth. 5. **China handset recovery removes a near-term overhang.** Management explicitly called Q3 FY2026 the bottom for Chinese OEM handset revenues, citing completed channel inventory drawdown and OEM purchasing now aligned with underlying demand rather than destocking dynamics. Double-digit sequential growth guided for Q4 FY2026 and into FY2027 provides a offsetting tailwind precisely as the Apple cliff accelerates, smoothing what could otherwise be a much rougher transition. 6. **Capital returns amplify the earnings recovery.** With FCF growing from $13.4B in FY2026 to $19.8B in FY2029 and the balance sheet flipping from $5.8B net debt to $9.1B net cash over the same period, Qualcomm has ample capacity for an aggressive buyback program that keeps EPS growth running ahead of net income growth (EPS CAGR outpacing the already strong net income CAGR from FY2026-FY2029). This is a self-funding flywheel: margin recovery drives FCF, FCF drives buybacks, buybacks drive per-share value independent of multiple expansion.

Risks

1. **Apple customer concentration and revenue cliff acceleration.** Apple product revenue is expected to decline roughly 50% sequentially into Q4 FY2026 and to fall further below the FY2026 base (already down from a prior 20% share assumption) in FY2027. If non-handset growth (automotive, IoT, data center) does not scale as fast as guided to fully offset this, near-term revenue and margin guidance could be missed, and this is a single-customer risk with limited near-term substitutability. 2. **Data center ramp is unproven at commercial scale.** The $5B-to-$15B data center revenue trajectory rests on HBC Gen 1 silicon achieving performance milestones and a mid-2027 launch that has not yet occurred; hyperscaler custom silicon engagements are notoriously difficult to convert from design wins to volume shipments, and any delay in silicon performance or hyperscaler adoption would directly impair the most valuable piece of the multi-year growth story. 3. **QCT margin recovery depends on pricing power that could face customer pushback.** The path back to 48-50% gross margins relies on double-digit pricing increases; in a competitive SoC market with MediaTek and in-house silicon efforts from Android OEMs, sustained price increases risk share loss or renegotiated terms, especially if China handset OEMs resist cost pass-through during their own recovery. 4. **China handset recovery is management's own call, not yet fully validated.** The 'bottom' characterization for Q3 FY2026 China handset revenue is based on channel checks and OEM ordering patterns; if the recovery stalls or reverses due to macro or competitive dynamics with domestic Chinese chipmakers, the offsetting tailwind assumed in the forecast disappears just as the Apple cliff bites hardest. 5. **Automotive design-win economics may not translate to comparable margins.** While 60%+ YoY automotive growth is impressive, automotive semiconductor gross margins and volume ramps differ meaningfully from handset economics; if automotive scales but at structurally lower profitability than modeled, the blended QCT margin recovery to 48-50% could be harder to achieve even with data center drag properly accounted for. 6. **Valuation and multiple compression risk from cyclicality.** Semiconductor stocks are prone to sharp multiple de-ratings during industry downcycles or AI capex digestion periods; if hyperscaler capital spending on custom silicon slows or broader semiconductor sentiment sours, the 19x forward multiple applied here could compress even if the underlying EPS trajectory largely plays out, capping realized upside.

📈 Price Targets