The Lazarus Effect — America's Productivity Miracle

American productivity has risen from the dead. After a decade of ~1% annual growth, non-farm business output per hour has climbed to ~2% per year over the past five years — the fastest rate in two decades. The Fed has raised its long-run GDP forecast. Jerome Powell said he never thought he'd see this many years of really high productivity. And AI had almost nothing to do with it.

Key Quotes

"I never thought I'd see this many years of really high productivity."

— Jerome Powell, outgoing Fed chair, at a press conference. The man who spent a career watching economic statistics sound stunned is a better signal than any chart.

"It is too early to credit artificial intelligence for the resurrection."

— The Economist. Productivity began picking up in the early 2020s. LLMs entered real commercial use only in the past year or so. The main macroeconomically discernible impact of AI so far is on business investment in data centres, not output-per-worker statistics. This is a useful corrective to AI triumphalism: the real economy is moving, and AI hasn't yet left a footprint.

"America's suits have at last taken full advantage of the signature innovations of the 2010s: smartphones, cloud computing, videoconferencing and the like."

— The biggest productivity jumps came not from the information sector (steady at ~6%, same as 2000-19) but from professional services, management, and real estate — nearly 24% of the economy. These businesses don't produce technology; they consume it. The 2010s built the tools. The 2020s finally learned to use them.

"Electricity is an input into just about everything and Americans pay half as much for it, on average, as Europeans and a third less than the Japanese."

— The shale-fracking revolution turned America into a net energy exporter. LNG plants now ship gas to Europe and Asia at premium prices. But the indirect effect matters more: cheap, abundant electricity means workers and machinery don't have to think about energy use. Energy-intensive industries survived in America that collapsed in Europe.

Key Themes

#productivity #economics #energy #adaptability

The AI non-cause

Productivity revived before AI went commercial. This isn't an AI story — or rather, it's a story about how much economic transformation happens without AI. The AI boom's main visible effect is capex (data centres), not output. When AI does show up in the statistics, it'll be layering onto an already-healthy trend. The counterfactual matters: America didn't need AI to resurrect productivity.

Technology adoption has a long tail

The signature innovations of the 2010s — smartphones, cloud, videoconferencing — took a decade to show up in productivity statistics for the sectors that merely use technology. The information sector was already efficient. The gains came when everyone else figured out how to absorb what had already been built. This has uncomfortable implications for AI timelines: even if the models are ready, the organizational learning might take another ten years.

Energy as economic substrate

Cheap electricity isn't just cheaper manufacturing. It's a permission structure: when energy is abundant and inexpensive, you don't optimize for it. You produce more. The energy-intensive industries that collapsed in Europe survived in America. This is the least-discussed driver of American economic exceptionalism, and it's structural, not cyclical.

Flexibility as the meta-advantage

America handed out cash during COVID instead of tying workers to existing jobs. When hiring resumed, people flowed to more efficient firms. The economy absorbed tariffs, deportations, and a war in Iran without breaking its productivity trend. Adaptability isn't just a nice property — it's a measurable economic input that compounds.

Critical Analysis

This is a genuinely contrarian piece disguised as a victory lap. The headline finding — productivity is back — is interesting. The real argument is in what didn't cause it: not AI, not tech sector growth, not luck. The Economist builds a case for three structural drivers (tech adoption lag, energy abundance, economic flexibility) and one contingent one (cash-not-job-attachment pandemic policy). It's a coherent story.

What's missing: the distribution question. A 2% productivity growth rate that accrues entirely to capital looks different from one that shows up in wages. The article is silent on who captured the gains. Also absent: the sustainability of energy advantage as the world decarbonizes. Cheap natural gas is a competitive moat only as long as carbon isn't priced.

The most provocative implication is the AI one. If the 2010s' innovations took a decade to show up in non-tech productivity, and AI is only now entering commercial use, then AI's productivity impact might not register until the 2030s. That means the current AI investment boom is a bet with a very long feedback loop — and the productivity statistics won't tell anyone they're right or wrong for years.

The Powell quote is the lede buried in paragraph three. A Fed chair expressing genuine surprise at sustained high productivity is about as close as central bankers get to saying "we don't understand what's happening." That's worth more attention than it gets in the piece.

Sources: The Lazarus effect — America is experiencing a productivity miracle Via: razorgirl (@razorgirl.diy) on Bluesky Last updated: 2026-05-15