Celestica

Refresh delta (since 2026-06-20)

What changed

1. 2026 outlook raised twice; 2027 outlook issued. FY2026: $19.0B / $10.15 (Apr) → $20.5B / $11.30, adj. op. margin 8.4% (Jul) → $21.0B / $11.55, margin 8.4% unchanged (Oct). First 2027 outlook: revenue $40.0B (+90%), adjusted operating margin 9.0%, adjusted EPS $21.50, capex $1.5–2.0B, "supported by 2028 customer demand forecasts that are meaningfully higher than our 2027 revenue outlook" [Q2 release 2026-07-27; Q3 release 2026-10-26]. Q4 2026 guide: revenue $6.275–6.575B, adj. EPS $3.37–3.57.
2. Q3 revenue $5,829.7M (+83% y/y, +24% q/q), above the top of the $5.25–5.55B guide. CCS $4,872.3M (+102%), segment margin 8.9% (8.3% a year ago, 8.7% in Q2). Adjusted EPS $3.30 vs a $2.88–3.08 guide; GAAP EPS $3.26 [Q3 release].
3. Compute & Storage is now the fastest line; HPS's share of revenue slipped. Compute & Storage (renamed from Enterprise) $1,609.1M, +242% y/y, 28% of revenue; Networking (renamed from Communications) $3,263.2M, +68%. HPS ~$2.3B (+70%) was 40% of revenue vs 44% a year ago — the growth is outrunning the design-led business [10-q-2026-q3.md, MD&A].
4. Concentration rose. Three CCS customers were 35%, 17% and 14% of Q3 revenue (66% together; Q2 32/17/14 = 63%; Q1 35/15/15 = 65%; Q3 2025 30/15/14). Top ten 85% (80% a year ago) [10-q-2026-q3.md, note 3 and MD&A; 10-q-2026-q2.md]. The largest customer was ~$2.0B of Q3 revenue. The filing does not name them.
5. The ramp is being funded from outside the P&L. $3,391.3M net from 11.1M new shares at $310 (closed 2026-08-07; share count 114.9M → 126.1M, +9.7%); inventory $4,509.7M vs $2,188.0M at 31 Dec (raw materials $3,537.8M); a new A/R sales program with $700.8M sold at 30 Sep (nil at 30 Jun; limit raised to $1.0B in September); 2026 free-cash-flow outlook cut to $460M from $600M "driven by targeted working capital investments to support the accelerating growth in 2027"; and "we expect a net use of cash in 2027" [10-q-2026-q3.md, notes 4, 5, 9 and Liquidity; Q3 release].
6. Leadership reshuffle. CCS President Jason Phillips retiring at year-end, Steven Dorwart succeeded him 2026-07-06; CFO Mandeep Chawla moved to a new Group President, Global Markets role and Todd Ankenmann became CFO, both effective 2026-10-01. Rob Mionis is now Chair and CEO [Q3 release].

What held. CCS segment margin kept rising (8.7% Q2, 8.9% Q3). ATS revived: Q3 $957.4M, +23% y/y, margin 6.6% (5.5%), driven by Capital Equipment [10-q-2026-q3.md]. Capex is on the ~$1B 2026 plan ($767.4M through nine months) [10-QForm 10-QThe quarterly version of the annual report. Lighter, and not audited., Capital Expenditures]. Net cash, not net debt: cash $3,961.3M vs borrowings and finance leases $818.8M [10-Q balance sheet; estimate: arithmetic]. Supply chain (Lens 2) and moat read (Lens 3) unchanged.

Phase A — Understand the business

Company Overview — carried, with two notes

• End markets renamed effective Q3 2026: Communications → Networking, Enterprise → Compute & Storage (servers and storage). Segment structure and history unchanged [10-q-2026-q3.md, note 3].
• The release now describes customers as "leading hyperscalers and frontier AI labs" and the forward-looking statements mention pursuits "in data center power infrastructure" [Q3 release]. Neither is quantified in the filing.
• Coverage-bucket note stands: indexed robotics, actually AI data-center hardware.

Supply Chain — unchanged, see the previous dossier.

Moats — carried. The rise in concentration (Lens 4) is evidence on the bargaining-power side of that lens, not a change to the moat itself.

Segments

Nine months: CCS $11,923.6M (82% of revenue), CCS segment income $1,040.3M at 8.7% [10-q-2026-q3.md]. Q3 Compute & Storage growth is "hyperscaler AI compute volumes and strength in our storage programs"; Networking is switch programs [MD&A].

Phase B — Measure performance

Earnings Result (Q2 and Q3 2026)

Q2 2026 [Q2 release 2026-07-27; 10-q-2026-q2.md]: revenue $4,698.6M (+62%) vs $4.15–4.45B guide; GAAP op. margin 9.8%; adj. op. margin 8.2%; GAAP EPS $3.17 (includes +$0.90/share pre-tax TRS gain); adj. EPS $2.54 vs $2.14–2.34 guide.

Q3 2026 [Q3 release 2026-10-26; 10-q-2026-q3.md]:

• Revenue $5,829.7M (+83%); gross margin 11.0% (12.3% in Q2 — the drop is the TRS swing in cost of sales: $0.9M loss vs $48.4M gain).
• GAAP op. margin 7.9% (10.2% a year ago, also TRS-driven); adjusted op. margin 8.5%, a company high, vs 8.4% guided midpoint.
• Net earnings $397.2M; GAAP EPS $3.26; adj. EPS $3.30 (+109%). GAAP tax rate 13%, adjusted 18% (guided ~20%; "favorable profit mix and a discrete tax benefit"), helped by acceptance into a bilateral advance pricing arrangement.
• Interest income $19.7M (vs $1.0M) on offering cash.
• Operating cash flow $389.7M; capex $274.1M; non-GAAP FCFFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. $127.5M. Nine months: OCF $1,156.9M, FCF $412.5M.
• Balance sheet 30 Sep: cash $3,961.3M; A/R $3,763.1M; inventory $4,509.7M; A/P $5,040.6M; equity $6,214.0M (from $2,216.3M).
• Working-capital metrics: cash cycle 42 days (Q2 47, Q1 55, Q3 2025 65); A/R days 56 (Q2 63); inventory days 70 (Q2 67); A/P days 74 (Q4 2025 50). Customer cash deposits $560.7M.
• Consensus beat/miss: not sourced against a primary consensus. Beats are stated here only against the company's own guidance ranges.
• Price: $271.52, +10.1% on the day (IBKR, 2026-10-27 intraday, realtime); 52-week range $244.54–$474.00; 26-week high $474.00, 13-week high $393.56 [IBKR misc statistics]. The August offering priced at $310. At this price CLS trades at ~23.5x the 2026 adj. EPS outlook and ~12.6x the 2027 outlook — stated as arithmetic, not as a valuation view.

Earnings Calls — not re-run. The Q2 call (2026-07-28) and the Q3 call / Investor & Analyst Day (2026-10-27) transcripts were not read; the Investor Day presentation (8-K 2026-10-27, Ex. 99.1) was ingested as the 8-K cover only, and the exhibit was not read. Tone carried from June, which was web-sourced.

Comps — carried, stale. The June table's multiples are four months old and the stock has moved from ~$382 to $474 to $272 since; not refreshed.

Catalysts since June

• 2026-07-06: CCS president change (8-K).
• 2026-07-27: Q2 beat and raise to $20.5B / $11.30; "2027 growth to accelerate" beyond 65% [Q2 release].
• 2026-08-05 to 08-07: preliminary and final 424B5, offering of 11.1M shares at $310, $3.45B gross [EDGAR filing list; 10-Q note 9].
• 2026-09-11: CFO transition announced (8-K).
• 52-week high $474.00 inside the last 26 weeks, then a fall to a 13-week low of $246.24 before the print [IBKR]. The cause of the drawdown was not researched in this pass; no reason is asserted.
• 2026-10-26: Q3 beat, FY2026 raise, $40.0B 2027 outlook, new NCIB intention (up to 2.5% of public float, after the current NCIB expires 2026-11-02) [Q3 release]. Stock +10.1% the next session [IBKR].

Phase C — Judge people & books

Management — carried, with the changes in delta line 6. Two of the three most senior operating roles (CFO, CCS president) changed hands in a single quarter while the business nearly doubles; both successors are internal (Ankenmann joined 2018, SVP Finance since 2025).

Forensic Red Flags

• Receivable sales are new and material. $700.8M of A/R sold under the bank program at 30 Sep, nil at every prior quarter-end back to Q1 2025; program limit raised from $450M to $1.0B in September [10-q-2026-q3.md, note 4 and Liquidity table]. Because the sold balance went from nil to $700.8M within Q3, Q3 operating cash flow of $389.7M would have been roughly −$311M without it, and the A/R-days drop from 63 to 56 is partly this program. The program is uncommitted and the bank can end it on three months' notice.
• Inventory build. $4,509.7M, up $2,321.7M in nine months (cash-flow statement), raw materials 78% of it; inventory days 70, highest in the seven quarters shown. Partly offset by payables (+$3.2B) and $560.7M of customer deposits.
• FCF guide cut while earnings guide rose. 2026 adj. EPS outlook up to $11.55; FCF outlook down to $460M, implying ~$48M of Q4 FCF. The company guides to a net use of cash in 2027 with $1.5–2.0B capex.
• TRS noise persists. GAAP EPS includes TRS effects of −$0.03 (Q3) and +$0.71 (nine months); notional 1.25M shares, strike $288.87 — the stock is now below the strike, so further falls cost Celestica cash at settlement [10-q-2026-q3.md, Liquidity — TRS; Q3 release].
• Dilution. +11.1M shares (+9.7%) in August; the NCIB intention for Q4 would buy back at most 2.5% of public float.
• Tax. Thailand tax holidays expire in whole or part 2027–2029; Thailand VAT assessment ~$12M and Romanian assessment ~$7M remain in dispute [10-Q, Income taxes].
• Covenants: in compliance at 30 Sep. No restatement or material weakness disclosed in the 10-Q.
• Not re-checked this pass: SEC EFTS (LR/AAER) scan — prior regulatory/regulatory-findings.md stands; status of the February 2026 plaintiff-firm investigations (Pomerantz, BFA) — not re-searched, and the 10-Q's litigation note names no securities class action.

Phase D — Project & stress-test

Forward Projection — the June base case is void (built on the wrong $17.0B / $8.75 figure). Not re-modelled. The company's own path is now: 2025 adj. EPS $6.05 → 2026 outlook $11.55 → 2027 outlook $21.50 [10-K FY2025; Q3 release]. A $40.0B 2027 at 9.0% implies ~$3.6B adjusted operating income.

Bull vs Bear (short)

• Bull (what the evidence supports): guidance has been beaten and raised three quarters running; segment margins rose every quarter while revenue nearly doubled; the 2027 outlook comes with a stated 2028 customer-forecast backstop; the balance sheet moved to ~$3.1B net cash.
• Bear (what the evidence supports): 66% from three customers and 85% from ten, both rising; the growth now needs equity, receivable sales and a net cash outflow in 2027; Compute & Storage, the fastest line, is volume compute where HPS's share is falling; the stock fell from $474 to the mid-$240s inside 13 weeks before this print, a reminder that the tape treats the name as a hyperscaler-capex proxy.
• Pre-mortem: a 2027 program slip at the 35% customer leaves Celestica holding raw-material inventory bought for a $40B year, after the receivable program has already pulled cash forward.

Devil's Advocate (short)

The $40.0B 2027 number is a customer forecast relayed by management, not a contract: the 10-Q repeats that master agreements do not guarantee volume, and the forward-looking assumptions list "customer forecasts" and "close alignment with key customers on demand planning". A doubling business whose cash conversion fell (FCF outlook cut, A/R sold, equity raised) is the pattern a short would push on.

Management Questions — carried; add first:

1. What share of the $40.0B 2027 outlook is the largest customer, and how much of it is under firm purchase orders versus forecast?
2. Why sell receivables with $4.0B of cash on hand, and what is the planned steady-state use of the $1.0B program?
3. How much of the $3.5B raw-material inventory is covered by customer liability clauses or the $561M of deposits?
4. Compute & Storage grew 242% while HPS fell to 40% of revenue — what is the margin difference between the two, and which dominates 2027?