The Compounding Nobody Is Talking About

There's a number I keep coming back to.

Three percent. That's the rough average rate at which global GDP has grown annually for most of modern economic history. It's the number that underlies every retirement plan, every infrastructure budget, every national growth forecast ever written. It's been the drumbeat of human progress — steady, reliable, and for most of us, invisible.

Three percent is what human ingenuity looks like when it operates at human speed.

I've spent thirty years at the intersection of technology and business strategy — at Logitech, RAPT, Afero, and CyberProtonics — watching technologies come and go, some that lived up to their promise and many that didn't. I've learned to be skeptical of hype and attentive to what actually compounds.

AI is different. Not because of what it does today, but because of what compounding looks like when the underlying rate changes.

At 3%, the world doubles in 24 years. At 6%, it doubles in 12. At 9%, it doubles in 8. The math is simple. What changes everything is whether AI can sustainably shift the rate — not just for one sector, but across all of them simultaneously.

It can. Here's why.

A Horizontal Accelerant

Every major technology wave before AI was vertical. The industrial revolution transformed manufacturing. The internet transformed information distribution. Both were extraordinary. Both were largely domain-specific in their initial impact, spreading to adjacent sectors over decades.

AI is structurally different: it accelerates compounding across all domains at once. It doesn't improve one sector. It improves the rate of improvement in every sector that runs on knowledge, decision-making, and pattern recognition — which is, increasingly, all of them.

Most people think of compounding as financial — interest on interest. But compounding is really just what happens when gains become the foundation for the next round of gains. Better decisions produce better outcomes. Better outcomes fund better tools. Better tools enable better decisions. The cycle shortens with every turn.

AI is the first technology in history capable of accelerating that cycle simultaneously in medicine, education, business, and beyond — not over decades, but now.

Medicine: Compressing the Calendar

Start with the domain where the stakes are highest.

Harvard Medical School's Dr. Nick Nassikas put it plainly: what we could have accomplished in medicine over the next 50 to 100 years, we will now accomplish in 5 to 10. Drug discovery, disease treatment, diagnostics — the constraint was never ambition, it was speed.

Microsoft's AI Diagnostic Orchestrator solved complex medical cases with 85.5% accuracy in 2025 — against a 20% average for experienced physicians. DeepMind's AlphaFold solved the 50-year protein folding problem and earned a Nobel Prize. That single breakthrough opens the door to drug design at a scale that would have required decades of manual laboratory work.

The compounding effect in medicine runs like this: better disease modeling produces better clinical trials, which produce better treatments, which free up resources for prevention, which feeds better models. The cycle already exists. AI shortens every interval in it.

Education: The End of One-Size

Mass education was designed for industrial efficiency — standardized content, standardized pacing, standardized assessment. It was the best system we could build when human teachers were the only available resource. It left enormous human potential on the table.

AI changes the fundamental constraint. For the first time, every student can have a learning experience calibrated to their pace, their strengths, and their gaps — in real time, at scale, at a cost approaching zero.

This one is personal for me. My son is 14, and I've spent real time this year building a practical AI framework for him — the do's and don'ts, the optimizations for note-taking, studying, and getting work done. Not using AI to shortcut the thinking, but to deepen it. How to use it to stress-test an argument before turning in an essay. How to use it to reconstruct and reinforce notes after class rather than transcribe during it. How to ask better questions rather than just get answers. The difference between a student who learns with AI and one who simply delegates to it is enormous — and it's a distinction that will define academic and professional trajectories for this entire generation.

What strikes me is how quickly a 14-year-old, once given the right framework, adapts. The learning curve isn't the technology. It's the intellectual discipline to use it well. That's teachable. And the students who develop it now will compound that advantage across every year of school and every year of their careers.

The compounding here runs across generations. A student who learns more effectively doesn't just get a better job. They become a more capable parent, a more informed citizen, a more productive contributor to every institution they touch. Better education compounds into better everything else. AI compresses the generational lag.

Business: From Quarterly Intuition to Real-Time Intelligence

Every business decision made on incomplete information is a tax on growth — invisible on the income statement, very real in outcomes. The market not entered. The product not killed early enough. The hire made on gut feel that didn't work out.

The traditional decision cycle: gather data, wait for a report, convene a meeting, debate, decide, implement, observe, repeat — quarterly if you're disciplined, annually if you're not.

AI compresses that cycle toward real-time. Not just faster reports, but genuine decision intelligence: pattern recognition across thousands of analogous situations, simulation of downstream effects, early warning on emerging risks. McKinsey estimates AI could generate $2.6 to $4.4 trillion in annual economic value — equivalent to adding another Germany to the global economy every year. The mechanism isn't magic. It's the elimination of the decision tax, at scale, across millions of businesses.

The companies moving fastest aren't just growing faster. They're building decision infrastructure that will widen the gap with every passing quarter.

What the Numbers Say

The skeptic's case deserves a direct answer. Goldman Sachs chief economist Jan Hatzius noted that AI investment contributed essentially nothing to U.S. GDP growth in 2025 — primarily because most AI infrastructure spending flows to imported semiconductors rather than into domestic economic activity. Goldman forecasts measurable GDP impact beginning in 2027.

That's a lagging indicator behaving exactly as lagging indicators do. Goldman's own analysts calculated that AI has already contributed roughly $160 billion in real U.S. economic activity since 2022 — but only about $45 billion has shown up in official statistics. The value is being created faster than the measurement systems can capture it.

The infrastructure is being laid. The organizational habits are being built. PwC projects AI's contribution at $15.7 trillion by 2030 — roughly 14% of global GDP. Goldman sees productivity growth accelerating into the early 2030s as integration deepens. The question isn't whether this compounds. It's whether you're positioned to compound with it.

The Strategic Implication

Across thirty years of watching technology create and destroy competitive advantage, the pattern is consistent: the leaders in any transition aren't the ones who moved first on the technology itself. They're the ones who moved first on the organizational capability to use it.

The internet didn't reward companies that bought the most servers. It rewarded companies that rebuilt their operations around digital-native workflows.

AI won't reward companies that spend the most on subscriptions. It will reward companies that rebuild their decision-making, their knowledge management, and their customer intelligence around AI-native workflows — before their competitors do.

The compounding effect starts the moment you start. The gap widens with every quarter you don't.

Three percent has been the ceiling for so long that we've stopped questioning it.

What if it isn't a ceiling? What if it's a floor — the baseline of human ingenuity operating without amplification?

The structural conditions for a step-change in the rate of human progress are now present in a way they have never been. Multiple domains compounding simultaneously. Decision quality improving at scale across millions of organizations. The cycle shortening with every turn.

That's not a productivity story. That's a compounding story. And if you understand compounding, you know the most important question isn't how big the number is today.

It's how long the runway is.

The runway has never been longer.

Related Reading