Productivity Divergence: Decoding Jim Paulsen's Thesis on the AI Capex Cycle

Why the innovation cycle runs on different rules than the business cycle, and what that means for every American's financial future.

What to Know

• Information sector profits rose nearly 90% from October 2022 to present while the rest of the S&P 500 rose only 10%
• Since 1998, information sector productivity grew 322% versus just 32% outside the technology sector
• Paulsen's Main Street Meter sits lower than 80% of readings since 1952, signaling extreme anxiety that must resolve before the bull continues
• M2 money supply ran negative year-on-year for 16 consecutive months, the first time since records began in 1960
• Nonfarm business productivity rose 4.9% in Q3 2025, but whether AI or layoffs drove it remains unresolved

Michael Burry is short the AI trade. He has called AI infrastructure catastrophically overbuilt, taken a significant leveraged position against the market, and launched a Substack called Cassandra Unchained after shutting down Scion Asset Management in November 2025. His argument is straightforward: too much capital is chasing a technology whose financial returns have not materialized at the scale the market is pricing in.

Jim Paulsen, Author, Paulsen Perspectives

Jim Paulsen looks at the same landscape and reaches a more tactical conclusion. He agrees tech valuations are wildly disconnected from the real economy, but argues this historic divergence makes the market overdue for a significant corrective phase, not a collapse, to allow the rest of the economy to catch up.

Tech Has Left the Business Cycle Behind

The core of Paulsen's argument, laid out across his November 2025 podcast appearance and his March 2026 Paulsen Perspectives Substack, rests on one structural claim: tech has decoupled from the business cycle. It now runs on an innovation cycle, and those two cycles have entirely different kill switches.

In every post-war recession before 2022, when the broader economy contracted, the technology sector contracted with it. That changed. From October 2022 onward, information sector profits rose nearly 90% while profits across the rest of the S&P 500 rose only about 10%. Total U.S. NIPA profits remain 5 to 10% below their long-run trend line. S&P reported profits are 18 to 20% above it. Both measures historically moved together until 2022. The divergence is structural, not cyclical.

Paulsen's explanation is that tech has gotten large enough and rate-insensitive enough to operate independently of cyclical forces. New era companies now account for more than 20% of GDP change, up from roughly 8% in the 1980s. If Apple invents something everyone needs, it grows regardless of whether the Fed funds rate is 2% or 6%. That immunity is what allowed tech to accelerate during a period when most of the economy quietly entered recession.

Tech profits diverge from full economy at 2022 inflection. Created via Gemini.

And that recession was real, Paulsen argues, even if it never showed up in headline GDP. From 2022 onward, Russell 2000 fundamentals collapsed and never recovered. Employment and earnings outside the information sector stalled. Small business owners have been saying conditions are bad for years. They were right. The Fed kept tightening because the S&P headline looked fine. It did not look fine underneath.

The Productivity Numbers Tell a Complicated Story

This is where the productivity debate becomes complicated. Paulsen's March 2026 Substack piece lays out a warning Wall Street has largely missed. Since 1998, information sector productivity has grown at an annualized pace of 5.48% compared to only 1.21% in the rest of the economy. Information sector productivity is up 322% over that period while productivity outside the information sector is up just 32%. Innovation has not transferred across the economy. It has stayed contained inside the sector that generates it.

The productivity spikes that excite markets tend to arrive during weak labor periods, not strong ones. Information sector productivity rose nearly 12% in 2025 while information sector employment declined 1.8% and job growth in the broader economy was essentially zero. Cutting staff looks like productivity on paper. It is not sustained efficiency. The same pattern appeared in 2002 to 2003, 2010, and 2020 to 2021, each a recession or near-recession. Nonfarm business productivity rose 4.9% in Q3 2025, but whether AI drove that or layoffs did remains unresolved.

Sentiment Says the Bull Is Younger Than Anyone Thinks

So why is Paulsen still constructive? Because the bull has room to run precisely because it has been so unloved. His Main Street Meter, which divides consumer confidence by the unemployment rate, currently sits lower than 80% of all readings since 1952. At the dot-com peak, both confidence and the meter were near all-time highs. Today they sit near historic lows while the S&P approaches records. Sentiment this pessimistic near market highs is historically a buying signal, not a warning sign.

However, Paulsen issues a critical warning that Wall Street is missing: this innovation cycle cannot run independent of the business cycle forever. For the market to sustain itself, the productivity gains locked inside the information sector must finally "catch" and spread to the other 80% of the old economy. If they don't, tech giants will eventually run out of buyers with the actual capital to fund their massive AI infrastructure, forcing a harsh market reality check.

Wrap Up

What this means for ordinary Americans is a market environment unlike 1999 in one critical way. Animal spirits have never been allowed to percolate in this bull. M2 money supply ran negative year-on-year for 16 consecutive months, the first time since records began in 1960. Real money supply growth sits near 1%. The yield curve was inverted longer than at any point in post-war history. None of those conditions existed before the dot-com bust.

If policy normalizes and the rest of the economy catches its breath, workers, small business borrowers, and households told a boom is happening while feeling none of it may finally get their turn.