A company built on the single-layer / best-of-breed model deliberately narrows its scope to one function, one component, or one stage of the value chain — and then invests everything into becoming the undisputed leader at that one thing. The economic logic is simple: depth beats breadth when the function is critical enough that customers will pay a premium for excellence and cannot afford mediocrity.
Also called: Pure-play, Horizontal specialist, Component model
Section 1
How It Works
The single-layer model is a deliberate act of strategic subtraction. Instead of building an end-to-end solution that handles every step from raw input to finished output, the company identifies the one layer of the value chain where it can create disproportionate value — and ignores everything else. ARM doesn't manufacture chips. Visa doesn't issue cards. Corning doesn't build phones. Each company owns a single, critical layer and lets partners handle the rest.
The critical insight is that not all layers of a value chain are created equal. In any industry, there are commodity layers where competition drives margins to zero, and there are bottleneck layers where technical difficulty, regulatory barriers, or network effects concentrate value. The best-of-breed player identifies the bottleneck layer and builds an unassailable position there. ARM's instruction set architecture sits inside billions of devices not because ARM is the only chip designer, but because its architecture became the standard that an entire ecosystem of manufacturers, software developers, and toolchain vendors built around.
Monetization varies by layer. Licensing is common (ARM charges per-unit royalties of $0.01–$0.10 per chip plus upfront license fees).
Transaction fees work when the layer processes flow (Visa's ~0.13% network fee on each transaction). Component sales apply when the layer produces a physical product (Corning's Gorilla Glass, sold to OEMs). The unifying principle: the single-layer company captures value proportional to its criticality, not its visibility.
UpstreamInputs & PartnersRaw materials, adjacent components, integration partners
Feeds into→
Single LayerBest-of-Breed SpecialistOne function, executed at world-class level
Delivers to→
DownstreamCustomers & IntegratorsOEMs, platforms, end-users who embed the component
↑Monetizes via licensing, component sales, or per-transaction fees
The central tension of the model is dependency without control. You are, by design, a component inside someone else's system. Your customer's product decisions — which materials to use, which standards to adopt, which suppliers to dual-source — directly affect your revenue, and you have limited ability to influence those decisions. The best single-layer companies resolve this tension by making themselves so good, so deeply embedded, and so expensive to replace that the dependency runs both ways.
Section 2
When It Makes Sense
The single-layer model is not a default strategy. It's a specific bet that works under specific conditions. Choosing it when those conditions aren't present leads to a company that's narrow without being dominant — the worst of both worlds.
✓
Conditions for Best-of-Breed Success
| Condition | Why it matters |
|---|
| The layer has a steep performance curve | If the difference between "good enough" and "best" is marginal, customers won't pay a premium. The layer must reward excellence disproportionately — 10x better performance, not 10% better. |
| Modular industry architecture | The value chain must have clean interfaces between layers. If components are tightly coupled and interdependent, a single-layer player can't plug in without controlling adjacent layers too. |
| High switching costs once embedded | The layer must be sticky. If customers can swap you out for a competitor in a quarter, your specialization earns you nothing. The best single-layer positions involve deep integration — APIs, standards, toolchains, certifications — that make replacement painful. |
| Multiple downstream customers | A single-layer company that serves only one customer is a captive supplier, not a best-of-breed specialist. The model works when the same layer is needed by many different end products or platforms. |
| R&D intensity favors focus | When the layer requires deep, sustained R&D investment — years of materials science, algorithm development, or regulatory expertise — a focused company outspends generalists who must spread R&D across many layers. |
| Standards or ecosystems amplify reach | When an industry standard or developer ecosystem forms around your layer, your position compounds. ARM's architecture became more valuable as more software was compiled for it. Visa's network became more valuable as more merchants accepted it. |
| Customers prefer to assemble best-of-breed stacks | In markets where buyers actively resist vendor lock-in and prefer composable solutions — enterprise software, electronics manufacturing — the single-layer model has a natural buyer. |
The underlying logic is that specialization creates a virtuous cycle: focus enables superior performance, superior performance attracts more customers, more customers fund deeper R&D, deeper R&D widens the performance gap. The cycle only works if the layer is important enough that customers notice the difference and are willing to pay for it.
Section 3
When It Breaks Down
The single-layer model has specific, well-documented failure modes. Most of them stem from the same root cause: the company's fate is tied to a layer it controls, inside a system it does not.
| Failure mode | What happens | Example |
|---|
| Vertical integration by a customer | Your largest customer decides to build your layer in-house, eliminating you from the value chain entirely. The more successful you are, the more attractive this option becomes for them. | Apple designing its own chips (A-series, M-series), reducing reliance on Intel and Imagination Technologies. Imagination lost ~50% of its revenue when Apple dropped its GPU license. |
| Layer commoditization | What was once a performance bottleneck becomes "good enough" across multiple suppliers. Your premium evaporates as the layer shifts from differentiated to commodity. | DRAM manufacturing in the 2000s — once a high-margin specialty, it became a commodity where only massive scale (Samsung, SK Hynix) could sustain margins. |
| Architectural shift | The industry moves to a new architecture that eliminates or restructures your layer. The modular interfaces you depended on get redesigned, and your component no longer fits. | Kodak's film chemistry expertise became irrelevant when digital photography eliminated the film layer entirely. |
| Customer concentration risk |
The most dangerous failure mode is vertical integration by a customer, because it's the one you can see coming but often cannot prevent. When Apple announced its transition from Intel to Apple Silicon in 2020, Intel lost not just a customer but a signal to the entire market that its layer was no longer indispensable. The defense is to make your layer so technically complex, so deeply embedded in standards and ecosystems, and so continuously improving that building it in-house is more expensive than buying it from you. ARM has survived this threat for decades because the cost of designing a competing instruction set architecture — and rebuilding the entire software ecosystem around it — is prohibitive even for the largest companies.
Section 4
Key Metrics & Unit Economics
Single-layer businesses are evaluated differently depending on whether they monetize through licensing, component sales, or transaction fees. But the core question is always the same: how much value does your layer capture relative to the total value of the system it sits inside?
Value Capture Ratio
Your Revenue ÷ End-Product Revenue
What percentage of the final product's value accrues to your layer. ARM captures roughly 1–2% of the value of each smartphone sold. Visa captures ~0.13% of each transaction. Low ratios are sustainable if volume is enormous.
Gross Margin
(Revenue − COGS) ÷ Revenue
The signature metric of a well-positioned single-layer business. ARM operates at ~95% gross margins (licensing). Visa at ~80%. Corning's specialty glass at ~40%. Higher margins signal a layer that's harder to replicate.
R&D Intensity
R&D Spend ÷ Revenue
How much of revenue is reinvested in deepening the moat. ARM spends ~35–40% of revenue on R&D. Cadence Design Systems spends ~35%. This ratio must stay high to prevent commoditization — the moment you stop outrunning competitors, the premium disappears.
Customer Concentration
Top 3 Customers' Revenue ÷ Total Revenue
The vulnerability metric. Below 30% is healthy. Above 50% means you're one phone call away from crisis. Corning's top customer (Apple, for iPhone glass) reportedly accounts for a significant share of its specialty materials revenue.
Core Revenue Formula — Licensing ModelRevenue = (Number of Licensees × Avg Upfront License Fee) + (Units Shipped by Licensees × Per-Unit Royalty)
Margin = Revenue − R&D − Sales & Support (minimal COGS for IP-based models)
The key lever for a single-layer business is volume multiplication. You invest once in developing the layer, then collect revenue every time it's used across potentially billions of units. ARM's R&D budget is roughly $1–2 billion per year; the chips using its architecture generate hundreds of billions in downstream revenue. The economics only work if you can spread fixed R&D costs across a massive installed base — which is why single-layer companies obsess over market share and design wins.
Section 5
Competitive Dynamics
The primary source of competitive advantage in the single-layer model is technical depth compounded by ecosystem lock-in. Being the best at your layer is necessary but not sufficient. You must also make it prohibitively expensive for customers to switch — not through contracts or penalties, but through the accumulated weight of integrations, toolchains, certifications, and co-developed standards.
This model tends toward oligopoly or soft monopoly within a given layer, but not across the entire value chain. Visa and Mastercard together process over 80% of global card transactions outside China. ARM architectures power over 99% of smartphones. ASML holds a monopoly on extreme ultraviolet lithography machines. The pattern is consistent: within a specialized layer, one or two players dominate because the R&D investment required to compete is enormous and the switching costs for customers are high.
Competitors typically respond in one of three ways. Vertical integration — a large customer builds the capability in-house (Apple Silicon, Google's TPU chips). Alternative architecture — a challenger proposes a fundamentally different approach to the same layer (RISC-V as an open-source alternative to ARM's proprietary architecture). Bundling — an adjacent-layer player absorbs the function into a broader offering, eliminating the need for a standalone specialist.
The moat deepens over time through what might be called ecosystem gravity. Each new partner, developer, or toolchain vendor that builds on your layer adds weight to the ecosystem. ARM's moat isn't just its instruction set — it's the millions of lines of software compiled for ARM, the thousands of engineers trained on ARM tools, the hundreds of chip designers who've built custom cores on ARM's architecture. Replicating the technology is hard. Replicating the ecosystem is nearly impossible. This is why RISC-V, despite being technically capable and free to license, has taken over a decade to gain meaningful traction in markets ARM dominates.
Section 6
Industry Variations
The single-layer model appears across industries wherever value chains are modular enough to allow specialization and critical enough to reward it.
◎
Single-Layer Variations by Industry
| Industry | Layer | Key dynamics |
|---|
| Semiconductors | Chip design IP (ARM), fabrication equipment (ASML), EDA tools (Cadence, Synopsys) | Extreme R&D intensity. Multi-year design cycles. Ecosystem lock-in through toolchains and instruction sets. Gross margins of 60–95% for IP layers. Winner-take-most within each sub-layer. |
| Payments | Network processing (Visa, Mastercard), online checkout (PayPal), acquiring (Adyen) | Network effects at the transaction layer — more merchants accepting Visa makes Visa more valuable to cardholders. Tiny per-transaction fees (basis points) multiplied by trillions in volume. Regulatory moats in licensing and compliance. |
| Specialty materials | Advanced glass (Corning), specialty chemicals (BASF segments), advanced ceramics | Materials science R&D creates multi-year lead times. Qualification processes with OEMs take 12–24 months, creating natural switching costs. Lower margins (30–45%) than IP businesses but high barriers to entry. |
| Contract manufacturing | Electronics assembly (Foxconn), semiconductor fabrication (TSMC) |
Section 7
Transition Patterns
Evolves fromVertical integration / Full-stackFull-service / Integrated solutionDirect-to-consumer
→
Current modelSingle-layer / Best-of-breed
→
Evolves intoIngredient brandLicensingPlatform orchestrator / Aggregator
Coming from: Many single-layer companies were born from vertical disintegration — a fully integrated company that decided to spin off or open up its best layer. IBM's decision to use an open architecture for the PC in 1981 created the modular ecosystem that allowed Intel (chips) and Microsoft (OS) to emerge as dominant single-layer players. More recently, companies like Twilio started as internal tools (at Amazon, in Twilio's spiritual lineage) before being extracted and offered as standalone best-of-breed services. The pattern: a vertically integrated company proves a layer's value internally, then a specialist captures that layer for the broader market.
Going to: Successful single-layer companies face a strategic fork. Some evolve into ingredient brands — Intel Inside, Gorilla Glass, Dolby Atmos — where the component's brand becomes a selling point for the end product. Others expand into licensing models, where the IP is the product and manufacturing is left to partners (ARM's entire business). A few attempt to become platform orchestrators, using their layer as the foundation for a broader ecosystem (Stripe expanding from payments into Atlas, Radar, Treasury, and effectively becoming financial infrastructure).
Adjacent models: The single-layer model sits in productive tension with the full-stack / vertical integration model. The industry oscillates between these poles — periods of modularization (best-of-breed wins) followed by periods of integration (full-stack wins) — in what Clayton Christensen called the "law of conservation of attractive profits." When one layer commoditizes, value migrates to an adjacent layer, and the cycle repeats.
Section 8
Company Examples
Section 9
Analyst's Take
Faster Than Normal — Editorial ViewThe single-layer model is the most intellectually honest business strategy there is. It forces you to answer a question that most companies avoid: what are you actually the best in the world at? Not "pretty good at." Not "competitive in." The best. Because if you're going to stake your entire company on one layer, that layer had better be unassailable.
What most people get wrong about this model is thinking it's about simplicity. It's not. It's about concentrated complexity. ARM's business looks simple from the outside — license IP, collect royalties. But underneath that simplicity is one of the most sophisticated R&D operations in the semiconductor industry, producing architectures that must balance power efficiency, performance, security, and backward compatibility across billions of devices. The simplicity is the output. The complexity is the input.
The key insight that separates great single-layer companies from mediocre ones is ecosystem awareness. The best operators — ARM's leadership, Visa's network team, ASML's strategic planners — think obsessively not about their own product but about the health of the ecosystem around them. They invest in partner enablement, developer tools, and standards bodies. They understand that their revenue is a derivative of their ecosystem's success. A single-layer company that optimizes only for its own extraction rate is a single-layer company that's training its customers to build alternatives.
My honest read: this model is under more pressure today than at any point in the last thirty years. The trend toward vertical integration — Apple building its own chips, Google designing its own TPUs, Amazon building its own networking silicon — represents a fundamental challenge to the modular architecture that best-of-breed companies depend on. The largest technology companies have concluded that controlling critical layers in-house is worth the R&D cost, because it gives them differentiation their competitors can't match. This doesn't mean the single-layer model is dying. It means the bar for what constitutes "best-of-breed" is rising. You can no longer be merely better than the merchant silicon alternative. You must be so much better that even a company with Apple's resources concludes it's not worth trying to replicate you. ARM has cleared that bar. Many others have not.
The founders and executives I'd encourage to adopt this model are those with deep, defensible technical expertise in a layer that is becoming more critical, not less. If your layer is on the right side of the complexity curve — if AI, electrification, biotech, or another secular trend is making your function harder and more important — the single-layer model can produce extraordinary returns. If your layer is on the wrong side — if it's being commoditized, automated, or absorbed — no amount of focus will save you.
Section 10
Top 5 Resources
01BookThe foundational text on value chain analysis. Porter's framework for decomposing a business into discrete activities — and identifying which activities create disproportionate value — is the intellectual origin of the single-layer model. Chapter 2 on the value chain is essential reading for anyone deciding which layer to own.
02BookChristensen's "law of conservation of attractive profits" — the idea that when one layer commoditizes, value migrates to an adjacent layer — is the single most important framework for understanding when the single-layer model works and when it breaks. Chapters 5 and 6 on integration vs. modularity are directly applicable.
03Academic paperPorter's argument that strategy is about choosing what not to do is the philosophical foundation of best-of-breed positioning. The essay's distinction between operational effectiveness (doing the same things better) and strategic positioning (doing different things) explains why focus alone isn't a strategy — you must focus on the right layer.
04BookSlywotzky's framework for understanding how value shifts between layers, companies, and industries over time. Essential for any single-layer company trying to anticipate whether its layer will gain or lose value in the next decade. The Intel and Microsoft case studies are particularly relevant.
05EssayThompson's framework explains how the internet shifted power from suppliers to aggregators — and why single-layer companies that own demand aggregation (Google, Facebook) have different economics than those that own supply-side components (ARM, Corning). Understanding where your layer sits relative to the aggregation point is critical for pricing power and long-term defensibility.
