Jim Paulsen has been watching economic cycles for four decades. He’s seen tech manias, credit busts, and everything in between. And right now, the veteran strategist and author of the Paulsen Perspectives newsletter believes Wall Street is making a familiar mistake — underestimating the second-order effects of a technological transformation that’s already underway.
His argument is deceptively simple. Artificial intelligence isn’t just another shiny object for Silicon Valley to monetize. It’s a productivity accelerant that could reshape the American labor market, compress costs across industries, and extend the current economic expansion well beyond what most forecasters expect. The kicker: almost none of this is reflected in consensus estimates.
“People are focused on the wrong part of the AI story,” Paulsen told Business Insider. The obsession with which companies will “win” the AI race — the chipmakers, the cloud providers, the model builders — has distracted investors from a far larger economic phenomenon. Productivity growth. Real, measurable, broad-based productivity growth of the kind the U.S. hasn’t experienced since the late 1990s.
That matters enormously.
Productivity — output per hour worked — is the single most important variable in long-run economic growth. When it rises, companies can pay workers more without raising prices. Profit margins expand without squeezing consumers. The Federal Reserve gets room to keep rates lower for longer. Government tax receipts climb, easing fiscal pressures. It is, in the language of economics, the closest thing to a free lunch that exists.
And Paulsen thinks we’re at the front end of a sustained productivity surge. Not a quarter or two of better numbers. A multi-year acceleration driven by AI adoption across sectors that have barely begun to integrate the technology — healthcare, logistics, legal services, financial planning, manufacturing, agriculture. The list is long.
He points to early data that supports this thesis. U.S. nonfarm productivity growth came in at 2.3% annualized in the fourth quarter of 2025, according to the Bureau of Labor Statistics — well above the 1.4% average that prevailed from 2005 to 2019. That’s not a one-off. Productivity growth has been running above trend for several consecutive quarters now, a pattern that Paulsen argues is consistent with the early stages of a technology-driven structural shift.
The parallels to the mid-1990s are striking and intentional. Back then, widespread adoption of the internet and enterprise software drove a productivity boom that confounded skeptics for years. Economists kept waiting for the gains to fade. They didn’t — not until the dot-com bust and the subsequent recession. In the interim, the economy grew faster than expected, inflation stayed lower than expected, and the stock market delivered returns that seemed almost absurd in real time.
Paulsen isn’t predicting a repeat of the 1990s bubble. But he is saying that the productivity dynamics could rhyme. And that most economic models, which assume productivity growth will revert to its sluggish post-2008 trend, are systematically underestimating what’s coming.
So what does this mean for jobs?
This is where the conversation gets uncomfortable. The prevailing narrative — amplified by breathless headlines about AI replacing white-collar workers — assumes a zero-sum outcome. AI gets smarter, humans get fired. Paulsen pushes back hard on this framing. History, he argues, shows that productivity-enhancing technologies create more jobs than they destroy, though the transition can be painful and uneven.
The mechanism is straightforward. Higher productivity lowers costs. Lower costs lead to lower prices or higher margins, often both. Lower prices stimulate demand. Higher margins fund expansion and investment. New industries emerge. Old ones restructure. The net effect, over time, is more employment, not less — though the composition of that employment changes dramatically.
Consider what happened with ATMs. When automated teller machines were introduced in the 1970s and 1980s, everyone assumed bank teller jobs would vanish. They didn’t. The number of bank tellers in the United States actually increased for decades after ATMs were deployed, because the machines lowered the cost of operating a branch, which led banks to open more branches, which required more tellers for customer service and sales functions that machines couldn’t handle.
Paulsen sees a similar dynamic playing out with AI. Yes, certain tasks will be automated. Yes, some roles will be eliminated. But the broader effect will be an expansion of economic activity that creates demand for workers in ways that are difficult to predict today. The jobs of 2030 will look different from the jobs of 2024. Many of them don’t exist yet.
Not everyone agrees.
Daron Acemoglu, the MIT economist who won the Nobel Prize in 2024, has been far more cautious about AI’s productivity potential. His research suggests that the economic gains from AI will be “modest” — perhaps adding 0.5 to 1 percentage point to total factor productivity over the next decade, far less than the techno-optimists predict. Acemoglu’s concern is that AI will be deployed primarily to automate existing tasks rather than to create genuinely new capabilities, and that the distributional consequences — who benefits and who doesn’t — could be severe.
Goldman Sachs has taken a middle position. The bank’s economists estimated in a widely cited 2023 research note that generative AI could raise global GDP by 7% over a ten-year period, equivalent to roughly $7 trillion. That’s significant but not transformative on an annual basis. More recent updates from Goldman’s team have nudged the estimates higher as adoption has accelerated faster than initially modeled.
The debate, in other words, isn’t about whether AI will boost productivity. It’s about how much, how fast, and how broadly.
Paulsen sits firmly on the optimistic end of that spectrum. His reasoning draws on a pattern he’s observed across multiple technology cycles: the biggest economic benefits arrive not when a technology is invented, but when it becomes cheap and ubiquitous enough for ordinary businesses to deploy it. The PC didn’t transform productivity when it sat on a few desks at IBM. It transformed productivity when every small business in America had one. The internet didn’t change commerce when it was a novelty for academics. It changed commerce when grandmothers started buying books on Amazon.
AI is approaching that inflection point now. The cost of running large language models has dropped by roughly 90% since early 2023, according to estimates from multiple research firms. Open-source models have proliferated. API access has become trivially easy. Small and mid-sized businesses — the backbone of U.S. employment — are beginning to integrate AI tools into their daily operations in ways that don’t make headlines but do show up in productivity statistics.
A regional insurance company using AI to process claims 40% faster. A law firm drafting contracts in hours instead of days. A manufacturing plant using predictive maintenance algorithms to reduce downtime by 25%. None of these are front-page stories. Collectively, they represent an economic force that Paulsen believes is being dramatically underpriced by financial markets.
The investment implications are significant. If Paulsen is right — if productivity growth sustains at 2% or above for the next several years — then corporate earnings estimates are too low, interest rates can stay lower than the bond market expects, and the current economic expansion has considerably more room to run. Equity valuations that look stretched on backward-looking metrics might actually be reasonable on a forward-looking basis.
That’s a big “if.” But it’s not a crazy one.
Recent data from the Federal Reserve’s Beige Book, released in March 2026, noted that businesses across multiple districts reported using AI tools to offset labor shortages and improve operational efficiency. The language was measured, as Fed communications tend to be, but the pattern was clear: AI adoption is no longer confined to the tech sector. It’s spreading into healthcare systems, retail chains, agricultural operations, and financial services firms of all sizes.
The labor market data tells a complementary story. Despite widespread fears of AI-driven layoffs, the U.S. unemployment rate remains below 4%. Job openings, while down from their 2022 peak, are still elevated by historical standards. Wage growth has moderated but remains positive in real terms. This is not what a labor market being hollowed out by automation looks like. It’s what a labor market in transition looks like — one where productivity gains are being absorbed through higher output rather than lower headcount.
Paulsen acknowledges the risks. A trade war, a geopolitical shock, a financial accident — any of these could derail the productivity story. And the distributional question that Acemoglu raises is real. If AI’s benefits accrue primarily to capital owners and highly skilled workers, the political and social consequences could be severe, regardless of what the aggregate numbers show.
But his core thesis remains: the market is fighting the last war. Investors spent the post-2008 era conditioned to expect slow growth, low productivity, and secular stagnation. That mental model is being disrupted by a technology that is fundamentally different from anything the economy has absorbed since the internet. And the adjustment in expectations — when it comes — could be substantial.
“The productivity story is the story,” Paulsen said in his Business Insider interview. “Everything else — the Fed, the deficit, the trade situation — those are subplots. If productivity is accelerating, a lot of problems that look intractable start to become manageable.”
He’s not wrong about the math. A sustained increase in productivity growth of even half a percentage point per year would, over a decade, add trillions of dollars to U.S. GDP, generate hundreds of billions in additional tax revenue, and meaningfully improve the long-term fiscal outlook. It would ease the burden of an aging population, reduce the inflationary pressure from tight labor markets, and create space for both corporate investment and consumer spending to grow simultaneously.
The question isn’t whether that outcome is possible. The question is whether it’s probable. And on that point, reasonable people can — and do — disagree sharply.
What’s harder to dispute is that something is changing. The productivity data, the adoption curves, the corporate investment patterns — they all point in the same direction. Whether this turns into a 1990s-style boom or something more modest will depend on factors that are impossible to forecast with precision: the pace of regulatory action, the trajectory of global trade, the speed at which the workforce adapts to new tools and new roles.
Paulsen is betting on the boom. And for investors who’ve spent the last 15 years anchored to a low-growth worldview, his argument deserves serious consideration — not because he’s necessarily right, but because the cost of being wrong about productivity, in either direction, is enormous.
The market has priced in the AI winners. It hasn’t priced in what happens when AI makes everyone else more productive too.
WebProNews is an iEntry Publication