Celestica: AI's Bottleneck Is Shifting

Celestica (CLS) is benefiting from an AI-driven shift in the infrastructure bottleneck toward networking, with a ramp of 800G switch deployments and planned 1.6T switches improving revenue and margin visibility. Rising compute intensity from agentic AI is increasing demand for high-speed switches, optical connectivity, and rack-scale infrastructure—areas where Celestica is expanding, including margin-accretive HPS programs. These product ramps and mix improvements support the company’s longer-term growth and earnings profile.

Analysis

Celestica’s differentiated value proposition is not just higher-speed switching assembly but orchestration of optics, power, thermal and rack-level integration — a win when hyperscalers move from siloed boxes to rack-scale, repeatable cell designs. That amplifies per-rack BOM for high-speed optics and test services, creating a multi-year, higher-margin revenue pool (think services + integration) that is stickier than one-off board builds and can support 100–300bp company-level margin expansion as mix shifts over 12–36 months.

Second-order winners include component test houses, optical-module suppliers and contract engineering partners that can scale validation for 800G/1.6T. The primary competitive threat is not another EMS copycat but architectural change (co-packaged optics, proprietary ASIC-to-ASIC fabrics) or hyperscaler vertical integration — each has discrete timelines (co-packaged optics seeding 24–48 months out) and differs in how it impacts external optics vs. system integration spend.

Tail risks are concentrated: a hyperscaler capex pause or a faster shift to co-packaged optics would compress demand and could reverse margin gains within 6–18 months. Near-term catalysts to watch are multi-customer program ramps, quotes for optical BOM increases, and design wins disclosed in quarterly calls — these will move conviction quickly and are must-watch data points for a 3–12 month trade horizon.

Consensus is underweight the durability of recurring rack-level services revenue but may be underestimating the technology risk from disaggregation/co-packaging longer term. The pragmatic trading stance is to capture near-term mix-driven upside while hedging the 24–48 month architectural risk that could move spend away from discrete high-speed optics and external integrators.

Key Decisions for Investors

• Long CLS equity (CLS) — 12 month horizon. Size initial position 2–4% portfolio. Target +35–45% total return if HPS program mix and optical BOM remain elevated; hard stop -18–22% on program delays or quarter-over-quarter booking misses.
• Call-spread to express asymmetric upside — buy 12–18 month CLS call spread (debit) to capture upside from program ramps while capping premium paid. Aim for 2–3x payoff if CLS wins additional rack-scale programs; max loss = premium (~1–2% position equivalent).
• Pair trade: long CLS / short FLEX (FLEX) — 9–12 month horizon. Size 1.5:1 (CLS:FLEX) to isolate premium for rack-scale integration vs broad EMS exposure. Target pair alpha 15–25% if CLS captures higher-margin HPS mix; unwind if FLEX reports >10% beat in next two quarters.
• Tactical exposure to optical-module suppliers (LITE) — 6–18 months. Buy small exposure to suppliers of 800G/1.6T optics to play optical-BOM tailwinds; take profits on 25–40% moves and hedge if market signals accelerated co-packaged optics adoption (technical roadmap updates or major hyperscaler disclosures).