The Margin Migration: Why Software Profits Don’t Disappear—They Move

Software margins will compress. The question is where the money goes when they do.

How industries lose their margins

Industries follow a predictable arc. High margins exist when:

1. Scarcity is real (capital, expertise, regulation, distribution)

2. Differentiation is defensible

3. Switching costs are high or opaque

4. Price discovery is broken (buyers don’t know what “fair” is)

Margins flatten when supply explodes, capabilities standardize, buyers gain leverage, distribution becomes cheap, and the product becomes “good enough.”

This isn’t about technology versus non-technology. It’s about where value migrates when the core offering commoditizes.

Four industries that lived this cycle

Airlines: When deregulation killed pricing power

Before 1978, regulated routes created regional monopolies. Limited competition, opaque pricing, high barriers. Airlines made real money.

Deregulation opened routes to competition. Price transparency arrived. Carriers overbuilt capacity. Competition shifted entirely to price.

Airlines became a low-margin commodity.

Where margins went:

• Aircraft manufacturers (Boeing, Airbus) via duopoly control

• Loyalty programs (selling miles to credit cards)

• Airlines fly planes at break-even and profit elsewhere

When the core service commoditizes, profit pools migrate to whoever controls the new scarcity.

Long-distance telecom: When marginal cost hit zero

Before the 1990s, near-monopoly control and massive infrastructure moats generated enormous margins. $0.25/minute was normal.

Fiber overcapacity, regulatory pressure, and VoIP changed everything. Marginal cost collapsed toward zero.

Long-distance calling became essentially free.

Where margins went:

• Network effects and platforms (WhatsApp, Skype, Zoom)

• Services on top of the pipe, not the pipe itself

When marginal cost collapses and differentiation evaporates, pricing power dies. But demand doesn’t disappear—someone else captures it.

Personal computers: When standardization murdered integrators

Before the mid-1990s, proprietary systems (IBM, Compaq, DEC) commanded healthy integration margins.

Wintel standardization, contract manufacturing, and massive supply expansion changed that. Components became interchangeable.

PC hardware margins collapsed.

Where margins went:

• Operating system (Microsoft)

• Processors (Intel)

• Later: cloud services (AWS, Azure, Google Cloud)

Standardization kills integrator margins. Component owners and platform controllers win.

Asset management: When transparency killed fees

Before index funds, opaque performance claims justified 2% annual fees. Limited benchmarking. Distribution control through brokers.

Vanguard’s index fund revolution brought performance transparency. Automation reduced costs. Regulatory scrutiny increased.

Fee compression was brutal. Passive funds now charge 0.03%.

Where margins went:

• Scale players operating on volume (Vanguard, BlackRock)

• Truly differentiated alpha generators (Renaissance, Bridgewater)

• Everything in the middle got crushed

When outcomes become easily benchmarked and the “secret sauce” proves replicable, pricing collapses to near-cost.

The pattern

Airlines: profits moved from flying to aircraft + loyalty
Telecom: profits moved from infrastructure to platforms
PCs: profits moved from integration to components + services
Asset management: profits moved from active fees to scale + alpha

Margins don’t disappear. They migrate.

They move from product to platform, feature to workflow, tool to system, creation to distribution, execution to decision rights.

Where software margins are flattening now

1. CRUD SaaS and horizontal tools

Basic CRM, project management, note-taking apps, analytics dashboards, internal collaboration tools.

Why margins compress:

Feature parity is trivial. Building a competent project management tool isn’t hard. Notion, Coda, Obsidian, Roam converge toward the same features within months.

Switching costs are overstated. Data export exists. APIs exist. Migration is annoying but survivable.

AI accelerates commoditization. When AI can generate 80% of your features from a prompt, proprietary algorithms matter less. The “good enough” threshold arrives faster.

Buyers expect bundling. Microsoft and Google give away collaboration tools. Salesforce bundles analytics. Standalone point solutions face constant pricing pressure.

Vendors now compete on sales and distribution, not product quality.

2. Developer tools

Open source is everywhere. AI coding assistants reduce tool differentiation. Platforms bundle features—GitHub, GitLab, and cloud providers absorb standalone tools (CI/CD, monitoring, security scanning).

Standalone dev tools struggle unless they own a workflow, own unique data, or become platforms themselves.

3. AI wrappers

Short-term: huge margins. You built a ChatGPT wrapper with a specialized prompt and light UI.

Medium-term: brutal compression. Model access commoditizes. Wrapper differentiation evaporates.

This mirrors early hosting companies (before AWS) and CMS tools (before WordPress). Thin layers on top of commodity infrastructure get squeezed.

Where software margins stay high

1. System-of-record software

Core banking systems, payroll (Workday, ADP), ERP (SAP, Oracle), identity management, electronic health records.

Why margins persist:

Switching costs are actually real. Migrating your core banking system is a multi-year, bet-the-company project. The risk of failure dwarfs any cost savings.

Data gravity is immense. Years or decades of transaction history, customizations, integrations.

Regulatory and operational risk. Buyers optimize for reliability and vendor longevity, not price.

AI doesn’t kill this—it reinforces it. The company with your data builds better AI models for you.

2. Regulated or mission-critical workflows

Cybersecurity platforms, compliance management, payment processing, risk management systems, critical infrastructure software.

Failure cost dwarfs price. A security breach costs millions. Regulatory fines are existential. Buyers optimize for “won’t ruin us,” not cheapest.

Trust and vendor longevity matter. You’re buying insurance, not features.

Caveat: If the underlying technology standardizes (open-source security tools, automated compliance), margins compress even here. But the compression is slower and less severe.

3. Platforms that own demand aggregation

App stores (Apple, Google), marketplaces (Amazon, Shopify), ad platforms (Google, Meta), payment rails (Stripe, Visa).

They sit between buyer and seller. They control discovery. They extract rent not because they’re the best product, but because they’re the marketplace.

This mirrors what happened in media. Newspapers lost margin; Google and Facebook captured it.

4. Software with true network effects

Figma (multiplayer collaboration), Slack (team communication), GitHub (developer collaboration), Notion (shared workspaces).

Value grows with users, independent of feature quality. The 10th person on Slack makes Slack more valuable for the first nine.

Network effects resist commoditization even when features erode. You can build a Figma clone with better features. But you can’t clone Figma’s network. The designers are already there. The plugins exist. The templates are shared.

Facebook survived despite being technically inferior. Slack maintains pricing power despite Microsoft Teams being free.

Network effects are the closest thing to a permanent moat software has.

Where value migrates in software

Software margins aren’t disappearing. They’re migrating. The profit pool is shifting from “can you build it?” to “can you control the outcome?”

Building features is commoditizing. AI accelerates this. What matters now:

• Do you own the data?

• Do you own the workflow?

• Do you control distribution?

• Do you have network effects?

• Are you embedded in a regulated process where failure is unacceptable?

If yes: margins persist or grow.
If no: you’re in a race to the bottom.

AI’s dual nature

AI destroys differentiation in commodity tools. Your analytics dashboard, your CRM, your project management tool—AI makes building these trivial.

But AI also creates new moats for those positioned to capture them:

Data flywheels. The company with your data builds better AI models for you. Every interaction improves the product. This is compounding defensibility.

Workflow capture. AI that learns your specific processes, your team’s language, your customer patterns—that’s not replicable by a competitor, even with the same underlying model.

Multi-agent orchestration. Simple AI wrappers commoditize. But orchestrating multiple AI agents across complex workflows? That’s genuinely hard. The companies that nail this build durable advantages.

AI simultaneously crushes margins in simple tools and creates unprecedented moats for complex, data-intensive, workflow-embedded platforms.

Bundling reshapes the profit pool

Microsoft bundles Teams, OneDrive, and collaboration tools. Google bundles Workspace. Salesforce bundles analytics, AI, and platform features.

This doesn’t just compress margins for standalone tools. It creates new defensibility for the bundlers.

Why bundling wins:

Distribution advantage—easier to upsell existing customers than acquire new ones.

“Good enough” dynamics—customers tolerate inferior bundled features if switching costs are high enough.

Pricing opacity—bundling masks the true cost per feature, preserving margin.

The standalone tool might be better. But “free with what we already pay for” is brutal competition.

The margin pool concentrates. Large platforms capture more. Standalone tools need exceptional differentiation or vertical specialization to survive.

Strategic evaluation framework

When evaluating companies, investments, or career opportunities:

• What scarcity does this software control? Data, workflow, regulation, distribution, network effects—or just features?

• What happens if this product is 80% as good and free? Do customers still pay, or switch immediately?

• Is the moat structural or temporal? Structural moats: data gravity, network effects, regulatory embedding. Temporal moats: better features, faster execution, brand. Temporal moats erode.

• Who captures value if marginal cost goes to zero? If AI or open source can replicate the core offering, where does the profit go?

If the moat is features, worry.
If the moat is position, dig deeper.

The bottom line

Software margins aren’t disappearing. They’re migrating fast.

From features to position. From products to platforms. From tools to workflows. From creation to distribution.

The companies that survive aren’t the ones building the best features. They’re the ones controlling data, workflows, distribution, or network effects.

Airlines, telecom, PCs, asset management—all followed the same arc. Software is no exception.

Every margin compression creates a new profit pool somewhere else. When PCs commoditized, cloud services emerged. When telecom commoditized, platforms captured it. When software features commoditize, the winners own the layer above features.

Position when this happens determines everything.