CLS · Celestica, Inc.

Current Thesis

Celestica is the mid-cap EMS/ODM that builds the racks, 800G/1.6T switches, and optics hyperscalers buy by the data-center, and the fundamental story is still accelerating six weeks after a Q1 blowout but the tape just printed its first real momentum reset of the cycle. The 2026-04-27 Q1 result was a crush: revenue $4.05B (+53% YoY vs $2.65B), adjusted EPS $2.16 vs $1.20, and management lifted the FY2026 guide from $17B to $19B revenue and $8.75 to $10.15 EPS in a single move. The CCS segment where the hyperscaler dollars land grew 76% YoY to $3.24B at an expanding 8.6% margin. The sell-side chased after the number (CIBC to $480, BMO to $450), and price ran to a $474 high on 2026-06-02. It then lost 12.6% on 2026-06-05 to $371.71, roughly -21% off the high in three sessions, slicing through the $390 support shelf on a technical downgrade and reports of insider selling. Orders are real and the guide was raised; the price structure is mid-correction. The clean re-entry is a higher-low reclaim above $390, not a grab at the falling knife.

Bull Case

• Q1 2026 (2026-04-27) was an unambiguous beat-and-raise: revenue $4.05B, +53% YoY; adjusted EPS $2.16 vs $1.20; FY2026 guide raised to $19B revenue (from $17B) and $10.15 EPS (from $8.75). A raise of this size rarely coincides with a fundamental top.
• CCS segment +76% YoY to $3.24B (Q1 2026), margin 8.6% vs 8.0% prior: the hyperscaler-facing line is growing fast AND expanding margin, undercutting the usual "EMS growth at zero incremental margin" knock.
• HPS (Hardware Platform Solutions, the rack-scale line within CCS) ~$1.7B, +63% YoY (Q1 2026): the custom-rack integration dollars most levered to a single hyperscaler build cycle are still compounding off a large base.
• Alphabet's ~$80B capital raise (reported June 2026) maps to a $10–16B incremental networking opportunity for CLS, underwriting the CCS ~70% CY26 topline growth the bulls model.
• AMD Helios AI rack design win, cited by BofA at its $430 target diversifies demand beyond the Broadcom/Google ASIC-fabric work into the AMD GPU-rack ecosystem.
• Post-print PT escalation: CIBC $480 (from $425, June 2026), BMO $450 (from $370, June 2026), BofA $430. These raises came after the print confirmation on delivered results, with less of the speculative front-running that preceded the April number.

Bear Case

• -12.6% on 2026-06-05 to $371.71, ~-21% off the $474 (2026-06-02) high in three sessions: the first distribution event of the cycle, breaking the $390.11/$410.93 support shelf technicians flagged. Momentum names that shed 20% this fast usually base before they re-run.
• Technical downgrade to Hold plus ~$1.4M insider selling (early June 2026): a yellow flag that fast money saw stretched valuation near the highs, even if it does not break the thesis.
• multiple ceiling: EMS/ODM picks-and-shovels carry a structurally lower terminal multiple than the chip designers they serve. At ~$370 on ~$10.15 FY EPS the forward multiple is already rich for the category; any CCS decel compresses it quickly.
• Customer concentration: hyperscaler demand is top-heavy; a single program push-out (the kind that hit the name ~18% in a week in 2024) is a binary air-pocket the diversified guide cannot cushion in the moment.
• Consensus Q2 EPS $2.10 sits below the company's own $2.14–$2.34 guide midpoint a low bar, but it anchors expectations to a beat, so an in-line print disappoints a tape that touched $474.
• EMS gross-margin mean reversion: 2024-25 margins rode scarcity pricing; component normalization pressures the CCS margin-expansion leg the bulls lean on.

Setup & Price Structure

A corrective pullback inside a primary uptrend. After topping near $474 on 2026-06-02, price closed $371.71 on 2026-06-05, a 12.6% single-session drop that severed the $390.11 near-term support and is now testing whether the low-$370s holds. Analyst targets bracket wide: low $330, median ~$385, high $480 (CIBC), with a Strong Buy consensus across ~15 desks. With price now under the median PT, the chase-the-tape phase has paused. A constructive setup requires a daily close back above $390 (reclaim of the lost shelf), ideally after a higher low in the $340–$360 zone; a slide that loses ~$330 turns the correction into a trend break. RSI has almost certainly unwound from overbought after the three-day flush, so momentum is no longer the green light it was at the highs strength has to be re-earned before this is a clean accelerating-name buy again.

Catalyst Calendar (next 30 days)

• No hard dated binary inside the window (through ~2026-07-07). The next company gate is the Q2 2026 print, scheduled 2026-07-27 (consensus EPS $2.10 vs guide $2.14–$2.34; revenue guided ~$4.15–$4.45B) beyond 30 days but the next decision point.
• Hyperscaler capex commentary, ~2026-07-23 to 2026-07-31: GOOGL/META/AMZN/MSFT Q2 prints set the 2026 capex trajectory that drives CLS order flow; any 2026 capex trim is the read-through to watch.
• Ongoing (undated): incremental hyperscaler / chip-partner design-win headlines (AMD Helios, Broadcom fabric, optics) these moved the stock 7%+ intraday during June 2026 and remain the primary swing factor before the July print.

What Would Change Our Mind

• A daily close that fails to reclaim $390 and then loses $330 (the pre-May breakout base) converts the pullback from digestion into a structural top.
• Q2 print (2026-07-27) showing CCS YoY growth decelerating below ~50%, or any walk-back of the $19B FY2026 revenue / $10.15 EPS guide the entire re-rating rests on that raised guide holding.
• Any hyperscaler (GOOGL/META/AMZN/MSFT) trimming 2026 capex on its late-July call removes the demand engine and hits the whole AI-infra cohort, CLS included.
• Theme rotation from "design wins" to mainstream "is the AI capex sustainable" skepticism (the seed visible in the June 5 narrative) once that flips, further strength is distribution to sell into rather than a base to build on.

Correlation Notes

CLS trades as a high-beta expression of the same AI-capex impulse as ANET, AVGO, and DELL, tending to amplify their moves on both sides given the EMS operating leverage. It is most tightly tethered to hyperscaler capex headlines (GOOGL, META, AMZN, MSFT); a capex-guide cut from any of them is a shared-fate event. Within the picks-and-shovels cohort it screens as the under-owned, Canadian-listed name versus crowded ANET/AVGO the source of both the periodic mispricing and the sharper drawdowns when momentum funds rotate out, as the -21% three-day reset off the 2026-06-02 high showed. Optics and accelerator-interconnect peers (COHR, ALAB) are useful confirmation tells: cluster strength there validates the demand narrative, while divergence is an early warning that the CLS bid is thinning.