Celestica's Q3 2025: Contradictions Emerge on Customer Ramp Timelines, 800G Demand, Contract Visibility, Optical Wins, and Tariff Impact on Production Shifts

Date of Call: October 27, 2025

Financials Results

• Revenue: $3.19B, up 28% YOY (above high end of guidance)
• EPS: $1.58 per diluted share, up $0.54 or 52% YOY (adjusted EPS)
• Gross Margin: 11.7%, up 100 basis points YOY (adjusted gross margin)
• Operating Margin: 7.6% non-GAAP operating margin, up 80 basis points YOY (highest quarterly non-GAAP operating margin in company history)

Guidance:

• Q4 2025: Revenue $3.325B–$3.575B (midpoint +36% YOY); adjusted EPS $1.65–$1.81 (midpoint +56% YOY); non-GAAP operating margin ~7.6%; adjusted tax rate ~20%.
• 2025 updated outlook: Revenue $12.2B (raised from $11.55B; +26% YOY); adjusted EPS $5.90 (raised from $5.50); non-GAAP operating margin 7.4% unchanged; FCF $425M (raised).
• 2026 preliminary high-confidence outlook: Revenue $16.0B (+31% vs 2025); non-GAAP operating margin 7.8%; adjusted EPS $8.20; non-GAAP FCF target $500M; CapEx expected 2.0%–2.5% of revenue.

Business Commentary:

* Revenue and Financial Performance: - Celestica reported revenue of $3.19 billion for Q3 2025, up 28% year-over-year. - The growth was driven by strong demand in the CCS segment, particularly in the communications end market.

• Segment Performance and Margin Expansion:
• Revenue in the ATS segment was $781 million, down 4% year-over-year.

• ATS segment margin improved to 5.5%, supported by strategic portfolio reshaping and margin enhancement.

• Strong Demand in Data Center Networking:

• The CCS segment revenue was $2.41 billion, up 43% year-over-year.

• This growth was driven by strong demand in data center networking, particularly for 800G switch programs across large hyperscaler customers.

• Investment in R&D and Capacity Expansion:

• Celestica increased its R&D spend by more than 50% for the year, focusing on product roadmaps and design engineering capabilities.
• The company is expanding capacity in regions like Thailand and Texas to accommodate growing demand from hyperscaler and digital native customers.

Sentiment Analysis:

Overall Tone: Positive

• "Revenue of $3.19 billion was up 28%"; company delivered "the highest quarterly non-GAAP operating margin in the company’s history"; management said they are "on track to deliver our strongest performance on record" and provided a confident preliminary 2026 outlook of $16.0 billion revenue (+31%) and $8.20 adjusted EPS, signaling strong growth visibility and bullish tone across networking, compute, and CAPEX-backed expansion.

Q&A:

• Question from Mike Ng (Goldman Sachs): Can you expand on the R&D increases (50% next year) and capacity expansions through 2028—what key products justify these investments and are they grounded in new or existing customer commitments? Also, is the new 2026 storage platform win with an existing hyperscale AI/ML compute customer and how large is hyperscale storage opportunity?
  Response: R&D is focused on networking, AI/ML compute, rack-scale storage and software; investments are tied to existing customer engagements—storage win is with an existing customer; storage is a smaller but strategic adjacenty compared with networking.

• Question from Karl Ackerman (BNP Paribas Exane): You noted Thailand and Texas capacity could double from 2024–2027 while CapEx is only 2%–2.5% of sales—what assurances or commitments do customers provide to support that capacity? And on 800G/1.6T and liquid-cooled switch mix, how should we think about trajectory?
  Response: CapEx is disciplined and customer-tied ($300–$400M planned next year); we have 12–15 months of solid forecasts and some longer commitments per program; 800G now ~50% of shipments vs 400G, 1.6T ramps late 2026/into 2027 with liquid cooling integral to those designs.

• Question from Samik Chatterjee (JPMorgan): For the digital native customer expected to ramp in 2027, how should we think about magnitude relative to your enterprise business and do you have capacity to support that ramp?
  Response: Mass production is expected in 2027 (samples mid-2026); the opportunity could be multiple billions in the first year but is excluded from 2026 guidance; capacity and CapEx plans are being aligned and included in next year's expansion planning.

• Question from Ruben Roy (Stifel): You referenced potential additional operating leverage beyond 2026—how are you measuring that against rapid revenue growth and increased design activity? Also, how do you view the scale-up networking opportunity versus scale-out?
  Response: Management expects continued margin expansion (company-level ~7.8% in 2026 with upside beyond) driven by mix and productivity; Celestica is well positioned to capture scale-up networking due to incumbency, design leadership and advanced manufacturing capabilities.

• Question from David Vogt (UBS): How should we think about the CCS composition over the next several years—will compute and optical materially grow versus switching, and can you expand on the storage program referenced for 2026?
  Response: Networking (switching) remains the primary driver, but compute and optical are growing adjacencies; select hyperscaler storage programs exist and are strategic but smaller than the core networking ramps.

• Question from Thanos Moschopoulos (BMO Capital Markets): What growth are you expecting in CCS outside hyperscalers/digital natives (OEMs, others) in 2026 and beyond?
  Response: 2026 growth is primarily underpinned by hyperscalers; non-hyperscaler pockets (optical, OEM programs) also grow but are smaller contributors relative to hyperscaler-driven networking and compute ramps.

• Question from Todd Coupland (CIBC): With your largest customers on switching, are you typically single- or dual-sourced? Also, what does your 1.6T program win cadence look like through 2026–2027?
  Response: For new technologies we are often the preferred/exclusive supplier through development and early ramp; customers sometimes dual-source temporarily but often revert; currently ~10 active 1.6T programs with significant share wins and multiple ramps planned.

• Question from Jesse Pytlak (Cormark Securities): On Optical programs, what breadth of customers are you engaged with and are these commonly bundled with switching programs?
  Response: We have deep engagements with several primary optical customers; optical opportunities are correlated and sometimes bundled with networking, and optical capabilities will apply to future CPO deployments alongside plugable optics.

Contradiction Point 1

Customer Ramp Timeline and Demand Visibility

It involves differing expectations about the timing of a significant customer ramp and the visibility into future demand, which are crucial for strategic planning and investor confidence.

What is the impact of the 2027 digital native customer ramp on your outlook? - Samik Chatterjee (JPMorgan)

2025Q3: Initial silicon availability will dictate mass production timing. We expect significant revenue from this customer starting in 2027, with growth accelerating in the following years due to the magnitude of the opportunity. - Mandeep Chawla(CFO)

Given the 400G delay, when will 1.6T be available and what are your 2026 plans? - Ruben Roy (Stifel)

2025Q2: We'll start generating 1.6T revenue in the back half of 2026 and into 2027, paced by silicon availability. - Robert Andrew Mionis(CEO)

Contradiction Point 2

Demand and Growth Expectations for 800G

It highlights differing views on the pace of demand and growth for 800G, which is a critical product segment for Celestica's Connectivity and Cloud Solutions (CCS) business.

How are the other two 800G customers ramping in Q2 into Q3? - David Vogt (UBS)

2025Q3: We saw a 50-50 split between 400G and 800G volumes in Q2. 800G will ramp up faster than 400G in the back half. While one customer showed accelerated demand, the others are also ramping but at different paces. - Robert Andrew Mionis(CEO)

How many customers and platforms are driving your upward revised CCS outlook with 800-gig switch ports, and how does the breadth of design engagements for 800-gig and above compare to 400G? - Karl Ackerman (BNP Paribas Exane)

2025Q2: 400G demand has been strong, but 800G is now ramping and expected to continue. All top 3 hyperscalers have 800G programs, with one seeing significant acceleration in demand. - Mandeep Chawla(CFO)

Contradiction Point 3

Customer Contract Duration and Visibility

It involves differing statements about the duration and nature of customer contracts, which can impact revenue forecasting and strategic planning.

Can you discuss long-term customer visibility and changes in contract types? - Paul Treiber (RBC Capital Markets)

2025Q3: We have good visibility with long-term customer contracts, demonstrating strong engagement for future programs. This includes follow-on commitments even before silicon finalization. - Rob Mionis(CEO)

What is the current visibility with customers, especially in CCS, amid uncertainty? How should we interpret the re-acceleration in the second half of 2025 among enterprise clients, including existing and new customers? - David Vogt (UBS)

2025Q1: Conversations with customers validate forecasts, showing resilience in design work for future programs. - Mandeep Chawla(CFO)

Contradiction Point 4

Optical Transceiver Win and Market Position

It involves differing statements about the significance and strategic importance of an optical transceiver win, which can impact market perception and competitive positioning.

What are the growth expectations for CCS outside of hyperscalers and digital natives through 2026 and beyond? - Ruben Roy (Stifel)

2025Q3: Optical programs also contribute, but the mix of who is involved and in what way has evolved. It's a much larger pool of customers, and it's across more product types. - Mandeep Chawla(CFO)

How should we think about margins as enterprise performance improves? Can you discuss the optical transceiver win? - Ruben Roy (Stifel)

2025Q1: Optical transceiver win was a competitive bid for 800G transceivers, with significant volumes and additional optical proof points. - Rob Mionis(CEO)

Contradiction Point 5

Tariff Impact on Production Shifts

It involves differing expectations about the ability to shift production to the U.S. and Mexico in response to tariffs, which can impact operational decisions and costs.

What impact will the digital native customer growth beginning in 2027 have, and how will that affect your outlook? - Samik Chatterjee (JPMorgan)

2025Q3: With existing capacity in U.S. and Mexico, revenue can triple without additional space. Complexity of products affects shift speed, but capabilities in these regions are valuable. - Rob Mionis(CEO)

If tariffs are reintroduced, how quickly could you shift production to the U.S. and what would the CapEx costs be? - Thanos Moschopoulos (BMO Capital Markets)

2025Q1: With existing capacity in Mexico and USA, revenue can triple without additional space. Complexity of products affects the shift speed, but capabilities in these regions are valuable. - Rob Mionis(CEO)