The art of low margin mastery

Today, I’m digging into Celestica.

They build a large amount of the electronic hardware running the modern economy including servers, AI racks, aerospace systems, medical devices, networking gear, and more. They are the behind the scenes operator for hyperscalers (they don't disclose these customers but think the likes of OpenAI and Anthropic) and regulated industries that need reliable, high-complexity manufacturing at scale.

A few things stood out in this breakdown:

• Their shift into AI and cloud hardware (HPS) has meaningfully expanded margins.

• Operating discipline is their real moat. In a low margin business, they squeeze efficiency until it becomes a competitive advantage.

• Capex is surprisingly low for a manufacturing company because customers often fund or supply the specialized equipment.

• The big negative: customer concentration is high. Their top two customers make up almost 40 percent of revenue. That is real risk is if one of these hyperscalers slows (or pauses) spending.

Why this matters for operators:

• They are a strong example of how to climb the value chain instead of staying stuck as a commodity vendor. Review growth plus gross margin expansion can lead to non-linear growth in operating and net income (non-linear = good!)

• A low margin assembly business like this requires process discipline. If you're getting started down this route look: document your processes, hypothesize improvement opportunities, test, implement, and the iterate back through the loop. Their process power and global footprint show how execution, not heroics, is what scales a complex business.

• They position themselves directly in front of secular tailwinds like AI infrastructure instead of chasing every possible customer. If you're like me, you may not be well-positioned to participate in the infrastructure boom related to AI. But as AI enables more and more intelligence at our fingertips, there will be ample opportunities to leverage it in our businesses and careers.

As the days and weeks progress, just ask yourself what you're doing to give yourself an edge with the advances in technology that are being thrown at us daily.....

With that, I'll see you tomorrow!

-Nick

TL;DR

• Celestica builds complex hardware for hyperscalers, industrials, aerospace, and regulated markets.

• Its recent pivot into AI and cloud infrastructure materially boosted margins and revenue scale.

• Entrepreneurs can learn how process power, mix shift, and disciplined execution can turn a low margin business into a strong cash engine.

• Biggest watchout: heavy customer concentration and exposure to capex cycles (their customer's capex cycles).

• Key lesson: operational excellence is a moat if you apply it relentlessly.

The 30,000-Foot View

Celestica operates as a global design and manufacturing partner in two major segments: Advanced Technology Solutions (ATS) and Connectivity and Cloud Solutions (CCS). ATS serves aerospace, industrial, health tech, and energy customers. CCS delivers data center hardware, networking gear, and open compute platforms. The company wins business by pairing engineering with high volume production, offering design services, test, fulfillment, and end to end lifecycle support.

Revenue mix in 2024 was 33% ATS and 67%t CCS. Within CCS, Communications contributed 41% and Enterprise 26%. HPS, the high growth platform for cloud and AI hardware, made up 29% of total revenue. The business is concentrated, with the top ten customers accounting for 73% of revenue and the two largest customers representing 28% and 11%.

Key metrics:

• TTM Revenue: $11.28B

• TTM Gross Margin: 12.0%

• TTM net income: $716.7M

• Employee count: ~26,000

• Industry: electronics manufacturing services and hardware platforms

Company History

• 1990s: Celestica originates from IBM manufacturing operations. IPO in 1998 on TSX and NYSE.

• 2018: Acquires Impakt and Atrenne to expand ATS capabilities.

• 2021: Acquires PCI in Singapore to deepen supply chain and regulated market presence.

• 2023: Onex sells down ownership stake, removing governance and overhang complexity.

• 2024: Switches to US GAAP and expands AI hardware programs.

• 2025: Raises revenue outlook to $12.2B due to hyperscaler demand.

Show Me the Money

Standout Financial Features

• AI-focused mix shift raised margins and improved profitability.

• Operating margin expanded from 4.3% in 2023 to 8.2% TTM.

• Top ten customers remain 73% of revenue.

• Thailand represents over half of manufacturing revenue.

• Their capex was suprisingly low to me. But their model is high volume assembly, not capital-intensive fabrication

Financial Data

The N.O.O.B. Nine — Competitive Powers

The Nerd Out on Business Nine is made up of Hamliton Helmer's famous "7 Powers" of competitive advantage (Scale Economies, Network Economies, Counter-Positioning, Switching Costs, Branding, Cornered Resource, and Process Power) combined with two of my own (Data Flywheel and Distribution Advantage).

Average Score: 3.1/5 - Celestica has a solid moat built on execution and footprint strength.

Memorable Marketing

Celestica markets with quiet B2B credibility. Its strategy relies on technical proof, reliability, and highly specific case studies. The marketing narrative shifted heavily toward AI infrastructure and open hardware ecosystems.

Campaign Highlights

OCP Global Summit Demonstrations (2024 to 2025)

• Idea: Showcase real open hardware and AI rack systems.

• Channels: Industry events, LinkedIn, technical sheets.

• Why it worked: Met buyers in the right rooms and demonstrated deployable hardware.

AI Datacenter Positioning (2024)

• Idea: Reframe the company as a key supplier to AI platform builders.

• Channels: Investor decks, cloud events, solution briefs.

• Why it worked: Clear tie to a major secular demand wave.

Regulated Market Assurance

• Idea: Highlight compliance and reliability.

• Channels: Certifications, datasheets.

• Why it worked: Reduced risk perception in aerospace, health tech, and industrial markets.

Tactical Takeaways

• Turn complex engineering into simple visual explanations.

• Use live demos and events as year long content engines.

• Publish quality metrics to shorten sales cycles.

• Anchor your brand to a big secular tailwind.

AI Uses & Opportunities

Current usage includes supply chain forecasting, yield improvement, defect detection, and predictive quality algorithms.

Future opportunities:

• AI quoting assistant that reviews BOMs and generates cost optimized quotes.

• RL driven factory scheduling to reduce changeover time.

• Automated visual inspection with continuous model tuning.

• Supplier risk detection using real time signals from logistics, sanctions, and macro data.

• RMA analytics using LLMs to generate design rule updates.

Bumps in the Road

• Customer concentration creates dependency risk.

• Exposure to hyperscaler capex cycles makes revenue volatile.

• Low margin gravity is persistent without mix upgrades.

• Geopolitical risk from global supply chain positioning.

• Accounting noise from fair value items complicates comparisons.

Your Swipe File

• Climb the value chain rather than fight in commodity layers.

• Use objective performance data as marketing.

• Concentrate on profitable customers but hedge try to customer mix.

• Treat process like a product and continuously optimize it.

• Align your roadmap with powerful macro demand trends, if possible.