PDF
Celestica has crossed from commodity contract manufacturer to a hardware engineering partner sitting directly under the AI datacenter buildout, and the market has only recently woken up to it. It builds the switches, compute, and rack-scale systems the hyperscalers cannot build fast enough themselves, and it does so at margins no other large EMS player can match. The thesis is not whether the trend is real. It is whether the price already pays for it, and where a long-duration buyer should step in.
Celestica spans two layers of the buildout. Its differentiation and highest-value programs sit in connectivity, the 1.6T switching and co-packaged optics that move data across the AI datacenter, which is why we categorize it there. It also captures rack-scale compute demand through the AMD Helios collaboration, so a position in CLS is exposure to both the networking and compute layers in a single name. That breadth is part of the case, and part of why it concentrates AI-buildout exposure rather than diversifying it.
The demand is structural, not cyclical. Hyperscalers are spending because falling behind on AI is an existential risk, not a discretionary IT upgrade, and that spend flows straight into Celestica's Connectivity and Cloud Solutions segment. CCS grew 76% year over year in Q1 2026 and now drives roughly 80% of revenue. Management raised full-year 2026 guidance to about $19 billion, up from a $12.4 billion base in 2025, and said the 2027 outlook strengthened over the prior 90 days. The company is supply-constrained, not demand-constrained, which is the signature of an early-to-mid capital cycle rather than a top.
The margin story is the proof the transformation is real. Adjusted operating margin has roughly doubled since 2021, from 4.2% to 8.0% in Q1 2026, the company's stated profitability milestone, while GAAP operating margin reached 6.7%, up from 4.9% a year earlier. Adjusted gross margin ran 11.3% in Q1 2026 (10.8% GAAP), a step up from the high-single-digit levels of the early 2020s and a premium to commodity EMS peers. This is not labor arbitrage. The AMD Helios rack-scale collaboration, the 1.6T co-packaged optics switch award, and multiple 1.6T networking programs show Celestica winning at the engineering frontier, embedded in customer roadmaps years ahead of volume.
Switching costs are high, the balance sheet is near-unlevered at roughly 0.3x net-debt-to-EBITDA, and capital allocation is clean: buybacks, no dividend, no value-destroying M&A. The footprint outside China (Thailand, Malaysia, Canada, Mexico, Romania) turns US-China decoupling into a demand tailwind competitors with heavy PRC exposure cannot match.
The price now pays for perfection. At roughly 3.0x sales versus a decade-long range near 0.25x, and 30x EV/EBITDA against an EMS history closer to 10x, the multiple is borrowed from a platform business Celestica is not. Gross margin is roughly 11% (10.8% GAAP, 11.3% adjusted) and free-cash-flow margin is under 4%. Those are good manufacturing economics, not software economics, and the re-rating assumes the AI premium is permanent rather than cyclical.
The structural risk is customer concentration layered on hyperscaler in-housing. The three largest customers were roughly 35%, 15%, and 15% of Q1 revenue. The same hyperscalers designing their own silicon (Google TPU, Amazon Trainium, Microsoft Maia) have the capability and the incentive to route production to ODMs or bring it in-house once volume justifies it. EMS history is a graveyard of this exact pattern: Jabil and Apple, Flex and Motorola, and Celestica's own 95% collapse when the telecom capex cycle broke in 2001. A single top-three customer cutting capex or re-sourcing a program would compress both the estimate and the multiple at once, and at this valuation the downside on a 20-30% growth miss is 40-60%, far larger than the roughly 23% upside to consensus targets.
The bull case wins on the thesis and loses on the entry. On the questions that decide whether this is a durable franchise, the read is firmly bullish, and it holds up: the demand is structural, the margin expansion proves differentiation, and the balance sheet and footprint are assets, not liabilities.
The Skeptic's strongest point is not that the thesis is wrong but that the price leaves no cushion, and the Technician independently confirms it: the stock sits in no-man's-land below its 50-day moving average on declining volume, with the highest-probability long-term entry 10-18% lower at the $321 and $295-302 support zones. That is the tension, and it resolves cleanly. This is a name to own for years, bought with discipline rather than chased at a level the chart and the multiple both flag as premature. Conviction is a 7: high on the business, tempered by an entry that rewards patience and a concentration risk that is real even if not imminent.
| Metric | Value | Note |
|---|---|---|
| Price (Jul 10, 2026 close) | $359.85 | Settled NYSE close |
| Market Cap | ~$41.3B | 115M shares out |
| Revenue (TTM) | $13.79B | +36.7% YoY |
| FY2026E Revenue (guidance) | $19.0B | +53% YoY, from $12.4B FY2025 |
| EPS (TTM, GAAP diluted) | $8.27 | +131% YoY |
| FY2026E EPS (adj) | $10.15 | 2026 guidance |
| Gross Margin (Q1'26) | 10.8% GAAP / 11.3% adj | premium to commodity EMS peers |
| Operating Margin (Q1'26) | 6.7% GAAP / 8.0% adj | up from 4.9% / 7.1% in Q1'25 |
| CCS Segment Margin (Q1'26) | 8.6% | segment margin, up from 8.0%; not company op margin |
| ATS Segment Margin (Q1'26) | 6.0% | segment margin |
| FCF Margin (TTM) | 3.6% | Capex-compressed |
| Rule of 40 | ~45 (growth + op margin) | Pass |
| Net Debt | ~$394M | ~0.3x net debt/EBITDA |
| Interest Coverage | ~21.6x | Fortress |
| P/E (TTM) | ~43.6x | vs EMS history ~13-16x |
| P/S (TTM) | 3.0x | vs own 10yr avg ~0.25x |
| EV/EBITDA | 30.7x | vs EMS history ~8-12x |
| Beta (5Y) | 1.52 | High |
| Short Interest | 3.3% float | Benign |
| 200-day MA | $321 | +12% below price |
| Drawdown from ATH | -23.8% | ATH $472.40 (Jun 2, 2026) |
Build the position through DCA rather than a single entry, anchoring adds to the $321 (200-day MA) and $295-302 (year-open) support zones the Technician identifies. Cap at 6% because CLS does not diversify AI-buildout exposure, it concentrates it: it moves with NVDA, AVGO, SMCI, and the hyperscaler capex cycle as one correlated cluster. If a direct hardware/EMS peer is already held in size, CLS should replace rather than add to that sleeve. The worst-case zone assumes a reversion toward commodity-EMS multiples if the AI premium fully deflates.
Q2 2026 earnings after close Monday, July 27, 2026. First full quarter testing the acceleration priced into the $19.0B full-year guidance. A beat-and-raise likely resolves the consolidation upward; a guide-down confirms the Skeptic's asymmetry.
This report is produced and published by Otto Analytics, LLC ("the Company") and is provided for informational and educational purposes only. It is general in nature, does not constitute investment, financial, legal, or tax advice, and does not take into account the individual circumstances, objectives, financial situation, or risk tolerance of any reader.
No advisory relationship. Nothing in this report creates an investment-advisory, fiduciary, or other client relationship between you and the Company. The Company is not a registered investment adviser or broker-dealer and does not provide personalized investment advice or recommendations.
Not an offer or solicitation. This report is not an offer to sell, or a solicitation of an offer to buy, any security, and should not be relied upon in connection with any investment decision.
AI-generated content. This analysis is generated in whole or in part by an automated artificial-intelligence system based on a stated analytical objective. It may contain errors, omissions, or inaccuracies, and has not necessarily been independently verified by a human analyst. Data may be drawn from third-party sources believed to be reliable but not guaranteed to be accurate or current.
Risk of loss. All investing involves risk, including the possible loss of principal. Past performance is not indicative of future results. Forward-looking statements are inherently uncertain and should not be relied upon as predictions of actual future events.
Conflicts / positions. The Company, its owners, or affiliates may hold positions in the securities discussed herein. Any such positions are disclosed in the report where applicable.
Do your own due diligence. You are solely responsible for your own investment decisions and for executing your own trades. You should consult a licensed financial professional, attorney, and/or tax advisor before making any investment decision.
By accessing this report, you acknowledge and agree to the foregoing. This disclaimer is governed by the laws of the State of Oregon.