The Second-Order Effects of AI on Business Strategy

McKinsey estimates AI could add $2.6T to $4.4T annually to the global economy. Every boardroom is asking how to capture that value. Almost none of them are asking what happens to their business model when their competitors capture it too.

First-order thinking about AI is straightforward: deploy the tool, reduce headcount, ship faster, cut costs. Second-order thinking is harder and more important. When everyone in your industry compresses the same costs, the competitive equilibrium shifts entirely. That shift is where strategy lives — and where most companies are not looking.

Effect 1: Headcount-Based Moats Are Evaporating

For decades, enterprise software companies built defensibility through operational scale — large implementation teams, armies of customer success managers, deep professional services benches. More headcount meant more installed customers meant more switching costs. That math is breaking.

Klarna replaced the equivalent of 700 customer service roles with AI in 2024, handling 2.3 million conversations at the same satisfaction scores. GitHub Copilot now contributes to more than 46% of code at some enterprise engineering orgs. The implication is not just cost savings — it is that competitors with 30 engineers can now execute what used to require 200. Scale as a moat is collapsing for operational functions.

What replaces it? Proprietary data, embedded workflows, and switching costs built into customer processes — not team size. If your defensibility depends on having more people than the next company, you are one deployment cycle away from losing it.

Effect 2: Pricing Power Compresses When Costs Approach Zero

When AI drives your marginal cost of delivery toward zero, your pricing model faces a structural problem: customers will eventually realize it too. The history of technology suggests this plays out over 5–10 years, but AI is compressing that timeline.

The strategic response is to move pricing from input (seats, hours, tokens) to outcomes (revenue generated, costs saved, decisions made correctly). Companies that reprice before the floor drops out will capture margin. Companies that wait will be negotiated down.

Effect 3: The Middle Management Layer Is Hollowing Out

Middle management exists to coordinate information and translate strategy into execution. Both of those functions are being automated. AI systems can synthesize data from 50 sources and surface the right decision point without a weekly status meeting.

This is not a headcount story — it is an organizational design story. The span-of-control math is changing. Where a VP could previously manage 6 direct reports effectively, AI-assisted tooling is pushing that to 12–15 for execution-heavy roles. The managers who own judgment, relationships, and accountability survive. The ones who own information flow do not.

• →Project management: AI tracks progress, flags blockers, and reallocates resources in real time
• →Data reporting: automated synthesis replaces the analyst whose job was pulling the weekly deck
• →Customer escalation routing: AI triages by severity and history before a human touches it
• →Hiring coordination: screening, scheduling, and scoring increasingly automated through the first two rounds

Companies redesigning their org structures now — not as a response to layoffs but as a proactive architectural decision — will have 30–40% lower G&A ratios than peers within three years. That difference compounds.

Effect 4: Competitive Clock Speed Is Accelerating

In 2015, a first-mover advantage in a software category lasted 3–5 years before a well-funded competitor could replicate core functionality. In 2026, that window is 6–18 months. AI-assisted development — GitHub Copilot, Cursor, Devin — compresses the time to build a functional product by 40–70% for standard features.

The strategic implication is not that execution speed matters more — it is that distribution and embedding matter more. When any team can build the product, the moat shifts entirely to who already has the customer. The companies that survive the next wave will be the ones that got deep into workflows before the next competitor shipped. Sequoia has been explicit: “distribution is the new product.”

What Strategy Actually Looks Like Now

I have been doing this across 65+ investments and three companies. The founders who are winning right now share one common thread: they are not competing on AI features. They are building AI into the structure of how customers run their operations, such that switching means disrupting the business process itself — not just switching software.