Everyone has been telling you that artificial intelligence is turbocharging the American economy. The charts, the TED talks, the breathless CNBC segments, the venture capitalists explaining it to you like you're a golden retriever. There's just one problem: a provocative new analysis suggests it's not actually true. At least not yet.
The Boom Is Real. The Explanation Is Wrong.
Let's be clear about what's actually happening, because the underlying economic story is genuinely good. The U.S. has experienced a real, measurable surge in economy-wide productivity over the last couple of years. Output per person-hour of labor is up. Companies are getting more done with the same number of workers. That part everyone agrees on.
What Axios is reporting, based on a new analysis, is that the reason for this boom is not what the tech industry has been loudly and expensively telling us. AI advances may be producing gains at the micro level, in specific sectors, specific tasks, specific workflows. But at the macro level, the economy-wide level, AI is not the driver of this trend. Not right now. Not yet.
So what is driving it? According to the analysis, companies are getting more output per labor hour because they are making better use of existing capital. The machines and infrastructure and systems they already have. The stuff they bought before the AI hype cycle hit warp speed. Boring, unglamorous, utterly un-venture-fundable capital efficiency.
What 'Better Use of Existing Capital' Actually Means
This framing deserves more attention than it's going to get, because the financial press is going to bury it under seventeen more AI earnings call transcripts by Thursday afternoon. When economists talk about making better use of existing capital, they mean companies have figured out how to squeeze more productivity out of assets they already own. Better management practices. Smarter logistics. Tighter operations. Less waste.
Think of it this way: you don't need a superintelligence to realize that your warehouse has been running its routes in the wrong order for six years, or that half your workforce is in back-to-back meetings that produce nothing, or that your procurement process has three redundant approval layers that exist purely because someone built them in 2011 and nobody ever questioned them. Fixing those things produces real, measurable productivity gains. It just doesn't make for a good pitch deck.
The analysis that Axios is reporting on is being described as provocative, and it is, specifically because it runs directly against the narrative that Silicon Valley, Wall Street, and most financial media have been aggressively selling for the last three years.
The AI Hype Machine's Uncomfortable Morning
Here is the part where we pause and appreciate the specific awkwardness of this moment. Corporations have spent, collectively, hundreds of billions of dollars on AI infrastructure, AI subscriptions, AI consultants, and AI strategy retreats at tasteful boutique hotels. CEOs have told shareholders, repeatedly and with great confidence, that AI is transforming their businesses and driving productivity gains that will justify all of it.
And now a serious economic analysis drops and says: actually, the productivity surge is real, but AI isn't the reason for it at a macro level. The economy got more productive because of unsexy capital utilization improvements. You did not need to blow the budget on GPU clusters for that.
This does not mean AI will never matter at scale. The analysis acknowledges that AI is producing genuine micro-level gains in some sectors, which is real, which matters. But there is a very large gap between 'producing real gains in some specific applications' and 'driving the most important macroeconomic productivity trend of the past two years,' which is what the tech industry has been implying, loudly, every single day.
Why This Is Hard to Talk About Honestly
The problem with findings like this one is the incentive structure around them. Every major technology company, every AI startup, every venture firm with money in AI, and every C-suite executive who has publicly tied their strategic vision to AI adoption has a material interest in the narrative that AI is already transforming productivity at scale. The counter-narrative, that maybe we're still in a 'wait and see' period, is not something these people are eager to amplify.
And so you get a strange situation where the economic data tells one story and the financial media, largely funded by the advertising and attention of the same tech industry, tells another. It doesn't require a conspiracy. It just requires a lot of people nodding along to the story that benefits them.
The Axios report is careful to note that AI may still become a major macro driver. This could still be the early innings. The gains could be front-loaded in micro applications before they show up in aggregate statistics. All of that is possible. But 'possible in the future' is doing a lot of work in a lot of conversations that are presenting it as an established present-tense fact.
The Dingo Take
Here is what this analysis is really telling us, stripped of the careful economist hedging: the single most hyped technological development in a generation, the thing every earnings call has been crediting for every positive business outcome since late 2022, is not yet showing up as the driver of the economy's most significant recent productivity improvement. The productivity improvement is real. AI's starring role in it, at the macro level, is not confirmed. Those are just the facts as Axios is reporting them.
None of this means the AI skeptics were right about everything, or that the technology has no future. What it means is that the gap between 'this thing is genuinely useful in specific contexts' and 'this thing is revolutionizing the entire economy right now' is enormous, and a lot of very expensive investment decisions, a lot of layoffs justified by AI-driven efficiency claims, and a lot of policy debates have been conducted as if that gap doesn't exist.
Somebody should probably go tell the shareholders.
