Celestica's Q1 Reaction: This Makes No Sense

Celestica posted 53% revenue growth in Q1 CY26, raised full-year guidance, and won its first likely 1.6T CPO Ethernet switch contract. Hyperscaler demand is accelerating for 800G/1.6T and CPO networking gear, with AMD Helios rack scale-up solutions progressing toward year-end availability. Shares fell more than 15% despite the strong print as investors worried about component shortages and margin pressure.

Analysis

The market is probably misreading this as a peak-growth print rather than an accelerating platform transition. The near-term beneficiaries are not just the obvious AI networking suppliers, but also the hyperscalers that can now de-risk rack-level deployment schedules by standardizing on higher-speed interconnects; that should pull capex forward and compress evaluation cycles for adjacent vendors. The first-order loser is anyone still exposed to legacy 400G/optics-heavy architectures, because each successful CPO win raises the switching cost of staying on the old bill of materials.

The second-order issue is that margin compression fears may actually be a feature, not a bug, if the company is using early-volume wins to secure design-in position before competitors can replicate the stack. In this part of the cycle, share gains often matter more than near-term gross margin, and the market usually over-discounts input scarcity until backlog visibility proves the bottleneck is demand, not supply. If the hyperscaler rollout stays on schedule into year-end, the relevant time horizon is months, not days: the equity can re-rate on order durability well before the revenue contribution is fully visible.

The consensus is likely missing that the selloff creates asymmetric entry because the stock is being priced as if component constraints will prevent conversion of design wins into shipments. That only holds if shortages persist across multiple quarters; if management’s guidance is conservative, then any stable supply cadence into the next two reporting periods should force a sharp revision higher in consensus estimates. For AMD, the catalyst is indirect but real: better rack availability improves the probability that scale-up deployments do not slip, supporting the broader AI server spend thesis.

Tail risk is execution: if CPO ramp complexity or optical supply tightens further, gross margin could lag for 1-2 quarters and keep sentiment suppressed. But that is a timing problem, not a thesis breaker, unless customer qualification slips beyond year-end. The key tell will be whether backlog converts without incremental working-capital strain; if it does, the current drawdown should look like a classic post-earnings dislocation.

Key Decisions for Investors

• Buy CLS on weakness over the next 1-2 sessions; use the post-earnings drawdown as entry, with a 3-6 month target for multiple re-rating if backlog converts and guidance stays intact.
• Express the view with a CLS call spread expiring after the next two earnings cycles; upside is driven by estimate revisions while premium paid is capped if supply issues persist.
• Pair trade: long CLS / short a legacy networking or optics-exposed peer over 1-3 months to isolate the CPO adoption trade rather than beta to AI capex.
• For AMD, look for incremental exposure through a small long into year-end rack rollout milestones; the upside is optionality from improved deployment cadence, but size should be limited because the direct read-through is indirect.
• If CLS stabilizes above the post-earnings low for 5-10 trading days, add to the long; if it loses that level on volume, treat it as a signal that supply-chain fears are becoming self-fulfilling.