Chapter 02 · The New Calculus of Scale in Competition
The Value Creation Equations
One firm plateaus. The other compounds. The difference is not strategy, talent, or capital. It is structural math, and it is the reason up to 70% of today’s Fortune 500 are projected to disappear within two decades.
The two limits
Same equation. Opposite behavior at the limit.
Value creation responds very differently to scale depending on whether output is bounded by human coordination or by compute. Push each model toward its limit and the curves separate permanently.
The legacy firm
Plateaus, then approaches zero
As an organization scales by adding people, coordination complexity grows faster than human output. Every additional head adds communication paths faster than it adds throughput. Value creation flattens and then reverses. This is bureaucratic collapse, and it is the structural reason up to 70% of today’s Fortune 500 are projected to drop off the list within two decades.
The AI-Driven Enterprise
Scales exponentially
As execution costs fall toward zero, latency shrinks to a fraction of its former size and value creation compounds. The bottleneck moves. It is no longer headcount, capital, or coordination. It is a single variable: the magnitude and ambition of human intent.
The measurable ratios
Three numbers that tell you which curve you are on.
You can run all three against your own P&L this afternoon. None of them require a consultant.
ARR / FTE
Annual Recurring Revenue ÷ Full-Time Equivalents
The headline leverage ratio. It measures how effectively a company converts human capacity into value. A 1,000-person company generating $100M has an ARR/FTE of $100,000 and is structurally vulnerable to a 10-person competitor generating $20M.
Token-to-Salary Ratio
Inference and model spend ÷ Human payroll
On peak days at the most aggressive AI-native firms this ratio exceeds 1.0: the company spends more on tokens than on people. Capital is shifting from Time to Intelligence, and a flat token spend means you are not scaling.
Productive work per dollar of inference
Output delivered ÷ Silicon payroll
The correction to ARR/FTE. A company with ten employees and $500M revenue posts a $50M ARR/FTE, but if five of them earn $50M each, the efficiency is a mirage. Carbon payroll and silicon payroll both have to clear.
The benchmark shift
What $100M in ARR used to cost in people.
The 2000s · human-powered
600–900
Employees required to reach $100M ARR. LinkedIn and Shopify both landed in this band. Growth was headcount-intensive. You needed an army.
The 2020s · AI-powered
Under 50
Anysphere reached roughly $100M ARR with about 20 employees. ElevenLabs hit similar milestones with about 50. Same revenue. An order of magnitude fewer people.
The risk isn’t that AI will replace your people. The risk is that a company using AI will replace your company.
Run it on your own numbers
Take this further.
Chapter 02 downloads
The discussion guide for The New Calculus of Scale, plus Paul’s explainer video.
AI Decisions & Roadmap Workbook
Eight pages that turn these ratios into a sequenced plan for your organization.
Framework diagrams
Every figure from the book in high resolution, cleared for your own decks.
Chapter 2 goes deeper on all of it.
The mortality table, the headcount trap, the efficiency illusion, and the binary choice facing every established enterprise.
Get the Book
About Paul Cheek
Paul Cheek is a global expert on AI-driven enterprises and enterprise innovation. He is a Senior Lecturer at the MIT Sloan School of Management and Senior Advisor for Entrepreneurship & Artificial Intelligence at the Martin Trust Center for MIT Entrepreneurship. As founder of the AI-Driven Enterprise Institute and Entonomy, he develops data systems and software that power AI agent-run businesses. A Forbes 30 Under 30 honoree, bestselling author of Disciplined Entrepreneurship: Startup Tactics, and recipient of MIT’s Monosson Prize for impact on entrepreneurship education, Paul has advised and built ventures from seed to scale, with his work featured in Bloomberg, CNBC, Forbes, CNN, Inc., Entrepreneur, and more.
Read Paul’s full bio →The weekly newsletter
Stay close to the work.
What Paul is seeing in AIDE strategy, board-level AI, and PE/VC moves. Short. Tactical.