Key Takeaways

  • The current lack of broad productivity gains from AI isn't a failure, but a classic "Solow Paradox," echoing how electricity's early adoption didn't immediately boost output.
  • After electricity became available in Manhattan, productivity growth barely budged for decades because economies needed to fully "digest" and reorganize processes, like factories redesigning layouts from steam to electric.
  • Stripe's internal data already shows early indicators of AI-driven re-platforming, including "collapsing business sizes" and the rise of "agentic commerce."
  • John Collison believes AI is in this critical digestion phase, but expects it to reorganize the economy far quicker than electricity's 30-year timeline.

The Productivity Paradox Isn't New

Founders and builders often scratch their heads: we have AI everywhere, yet official productivity numbers aren't exploding. Where's the promised boom? John Collison, co-founder of Stripe, says this isn't a bug; it's a feature of truly disruptive technology. It’s a replay of the "productivity paradox" observed with electricity. Collison paints a clear picture: “Finally, electricity in Manhattan. And for decades after, even as electricity adoption grew, productivity growth barely budged or even slowed down.” People got excited about electrified factories, but the immediate output gains were minimal. Why?

The problem, Collison explains, wasn't the technology itself. “The electricity did work.” The real challenge was that economies needed time to adapt. Factories designed around a central steam engine – with all machines clustered for belt drives – couldn't just swap in electric motors and call it a day. They had to be completely redesigned, sometimes tearing down walls, moving production lines, and rethinking workflows from the ground up. This "digestion" period, where processes and organizations adjust to the new tech, is messy and time-consuming. Collison emphasizes, "Transformative technology looks for a long time like it's not doing much."

AI's Digestion Phase: Shorter, More Intense

Collison argues we're witnessing this digestion phase with AI right now. While the headlines focus on large language models, the real impact is quietly reshaping the core infrastructure of business. He points to early, tangible signs visible in Stripe's vast transaction data: a rise in "solopreneurs and global-first startups," alongside "collapsing business sizes" and the emergence of "agentic commerce." These aren't just buzzwords; they represent a fundamental re-platforming. Companies are getting smaller, more distributed, and increasingly relying on autonomous AI agents to handle tasks that once required human teams.

Historically, the shift from steam to electricity took nearly 30 years for the economy to fully reorganize. Collison sees a similar, yet accelerated, trajectory for AI. He suspects we won't have to wait nearly as long. “Electrification took 30 years to reorganize the economy. But I suspect we won't need to wait anywhere as long as that for AI.” The pace of change is simply faster today. This implies that while the current productivity numbers might feel flat, the groundwork for exponential growth is being laid with incredible speed. Businesses are fundamentally changing how they operate, how they structure teams, and how they interact with customers – often before the economic models catch up.

What to Do With This

Forget waiting for a universal AI productivity boom. Your move is to actively redesign your business's core processes now, anticipating a world where AI agents and smaller teams are the norm. Pull apart a key workflow this week – sales outreach, customer support, content creation – and identify which 80% could be handled by an AI agent if you re-architected the process entirely, rather than just adding AI as a bolt-on.