Celestica May Be One Of The Cleanest AI Infrastructure Plays Left

Celestica was upgraded to Buy as it shifts from contract manufacturing to a key AI infrastructure co-designer and supplier benefiting from surging AI data center demand. Q1 2026 showed 53% revenue growth, and 2026 adjusted EPS guidance implies 68% growth, supported by operating leverage, margin expansion, and strong free cash flow. The company is also gaining from hyperscaler adoption of 800G and 1.6T Ethernet technologies and leadership in AI cluster switches.

Analysis

CLS is transitioning from a labor-arbitrage manufacturer to an embedded design-and-scale partner in the AI networking stack, which matters because that shifts it from low-margin pass-through revenue to higher-value content tied to architectural lock-in. The second-order winner is not just CLS equity holders: hyperscalers that standardize on high-bandwidth Ethernet get faster cost/performance improvement, while legacy interconnect vendors and slower-moving EMS peers risk losing socket share as customers consolidate around fewer, more scalable suppliers.

The bigger implication is that AI capex is moving down the stack from accelerators into the plumbing: switches, optics, and board-level integration. That tends to lengthen revenue visibility because once a cluster design is qualified, replacement cycles and incremental capacity ramps become stickier than one-off hardware orders. The main loser is any supplier dependent on proprietary or immature networking protocols; as Ethernet wins, differentiation migrates toward execution, yield, and thermal/power design rather than pure ASIC IP.

The bull case is still early in the cycle, but consensus may be underestimating how quickly margins can re-rate if mix keeps shifting toward co-designed systems. The risk is that AI infrastructure spend becomes more lumpy after the current buildout phase, or that customer concentration turns a few hyperscaler procurement decisions into abrupt growth deceleration; that would show up over months, not days. Near term, the stock likely trades on guidance revisions and booking momentum, but over 12-18 months the key watch item is whether free cash flow keeps compounding faster than revenue, which would confirm durable operating leverage rather than a temporary demand spike.

Key Decisions for Investors

• Go long CLS on any 3-5% post-rally pullback; target a 3-6 month hold for a continued multiple rerate if AI networking content keeps expanding. Risk/reward favors buying weakness because the market usually underprices mix shift before margin inflection shows up in reported numbers.
• Add a call spread in CLS (6-9 month tenor) to express upside with defined risk; structure for a 2:1 to 3:1 payoff if the next guidance update confirms sustained order visibility and operating leverage.
• Pair long CLS / short a slower-growth industrial or legacy EMS name over the next quarter to isolate AI-infrastructure share gain versus general manufacturing exposure. The thesis is relative multiple expansion for CLS as investors pay up for design-in durability.
• Set a stop on CLS if backlog commentary or hyperscaler capex signals soften for two consecutive months; the primary bear case is not valuation, but a pause in cluster deployments that could compress growth expectations quickly.
• Monitor 800G-to-1.6T adoption milestones over the next 2-4 quarters; if customer mix broadens beyond the top hyperscalers, add to the position because concentration risk should fall and the market will likely award a higher terminal multiple.