Back to News
Market Impact: 0.55

Prediction: This AI Infrastructure Stock Could Be One of 2026's Biggest Winners

CLSITAVGOMRVLAMDINTCNFLXNVDANDAQ
Artificial IntelligenceTechnology & InnovationCorporate EarningsCorporate Guidance & OutlookCompany FundamentalsAnalyst InsightsInvestor Sentiment & PositioningMarket Technicals & Flows
Prediction: This AI Infrastructure Stock Could Be One of 2026's Biggest Winners

Celestica's AI-driven connectivity and cloud solutions (CCS) unit drove a 43% year-over-year increase in Q3 2025 revenue to $2.4 billion and now represents 76% of company sales. Management guides 2025 revenue of $12.2 billion (up 26%) and non-GAAP EPS of $5.90 (up 52%), expects ~31% top-line growth and $8.20 EPS in 2026, and reports its custom Ethernet switch share rose to 55% from 40% as hyperscaler demand ramps. Trading at about 3.2x sales versus a 5.5x Nasdaq average, Celestica argues that reaching $16 billion in revenue could imply an ~$88 billion market cap, forming the basis for a bullish investment case on continued AI infrastructure spending.

Analysis

Market structure: Celestica (CLS) is a direct beneficiary as hyperscalers accelerate AI data‑center builds — CCS is 76% of revenue and grew 43% YoY (Q3 2025), and CLS reports custom Ethernet share rose from 40%→55%. Upside flows to partners (AVGO, MRVL, AMD) that supply ASICs and switch silicon, while commodity EMS and legacy server OEMs face margin pressure as pricing shifts to customized, higher‑margin rack/network SKUs. Supply tightness will likely concentrate on high‑speed optics, switch ASIC capacity and advanced PCBs, creating pricing power for qualified EMS providers in 6–18 months. Risk assessment: Key tail risks are customer concentration (a single hyperscaler design win/start of production could drive +20–30% rev volatility), regulatory export controls on advanced chips (BIS actions within 30–90 days), and a sudden capex pause by hyperscalers (a 30% cut would likely remove >20% of CLS revenues given CCS weight). Time horizons: expect near‑term earnings volatility (days–weeks around reports), medium‑term validation when rack‑scale mass production begins in 2026, long‑term structural exposure if custom AI processors grow ~78% as CLS projects. Hidden dependencies include reliance on Broadcom/MRVL design cycles and contract pricing renegotiation risk. Trade implications: Tactical: establish a selective long in CLS (equity + LEAPS) to capture 2026 mass‑production upside while hedging with short dated call spreads to fund premium; consider pair trades long CLS / short INTC to express a shift toward custom networking ecosystems over CPU‑centric incumbents (6–12 month horizon). Macro knock‑ons: stronger AI capex momentum would steepen yield curves (pressure on long duration bonds) and lift risk assets; commodity demand (copper, optical components) should firm over 6–12 months. Contrarian angles: The bullish consensus underestimates two risks — hyperscaler insourcing/vertical integration and pricing clawbacks if customers consolidate suppliers; CLS’s current 3.2x sales multiple discounts execution risk but may underprice customer concentration. Historical parallels: networking cycles (2010s) rewarded design‑win leaders but punished firms that lost single large contracts, so require concrete design‑win confirmations (mass production notices) before full allocation. Unintended consequence: aggressive share gains can attract margin compression if multiple EMS competitors chase the same high‑margin rack projects.