There are two ways you might frame preliminary productivity and cost data released on Thursday by the BLS.
One spin says productivity gains decelerated and wage pressures increased across the US economy at the end of 2025, not what you’d expect at a time when human capital’s going the way of the Dodo.
Another spin on the same figures says the increase in productivity, despite falling short of the prior quarter’s outsized gain, was nevertheless robust and thereby constitutes more evidence that America is in fact experiencing an efficiency renaissance.
I’d be inclined to go with the latter interpretation, although I wouldn’t use the word “renaissance” just yet. It’s early days, and even if we do see the kind of productivity boom AI enthusiasts are fond of predicting, “renaissance” will feel like a cruel misnomer given the read-across for human job losses. Hold that thought.
The headline productivity print, 2.8%, was easily ahead of the 1.9% consensus, and the prior quarter’s already brisk growth rate was revised even higher to reflect a 5.2% gain. As a reminder, productivity is just output per hour.
For all the fanfare, recent gains don’t exactly leap off the page. Even Q3’s 5.2% advance, the briskest in half a decade, isn’t suggestive of an epoch.
Unlike the prior two quarters, Q4’s productivity gain wasn’t accompanied by a drop in unit labor costs. ULC printed a 2.8% gain as well, ahead of the 2% consensus.
The unit labor cost calculation is compensation deflated by productivity, which means hourly comp growth was nearly 6%. (Real comp rose more than 3%.)
Overall, it’s fair to say Thursday’s release kept the dream alive, so to speak, as it relates to the tech-enabled productivity boom thesis.
But at the risk of resorting to an editorial cliché, there are still far more questions than answers, not least of which are i) will efficiency gains ultimately justify the stratospheric cost of the AI buildout and, more provocatively, ii) what’s the cost to society and human well-being of a scenario where labor costs go from being the largest business expense to a line item so trivial it’s barely worth mentioning?
Cost for the productivity gain is unknown. Biggest concern is that the ai increased productivity will eliminate on balance jobs. What if this unknown plays out as causing more jobs than we had before?
I think it is easy to be pessimistic but often these types of doom and gloom narratives are just simply wrong.
Since the advent of mass production people have been predicting the end of jobs and depression due to factory innovation and automation. Time and again developments proved this pessimistic view wrong. I personally think this narrative will be wrong as well. Now there will be winners and losers. The winners will be those that adapt to using these tools. The losers will be those the pessimistic narrative impacts so hard they refuse to learn about it.
Therefore by promoting a pessimistic narrative people are creating conditions where people are deluded into not learning about a new technology. Don Quixote tilting at the the windmills of our age. Are the pessimists actually trying to help us or is this part of a mass psychosis promoted by the Kremlin to delude us into ignoring our greatest strength?
100% agree. I’m in healthcare, I welcome the AI overlord that will do my charting.
Every few years, there’s a layoff at my health care system. These are unfortunately necessary because management teams will never self-police. Everybody knows at least five people who don’t actually do anything at work. But in my 25+ years, I’ve never seen a layoff that changed anything in any meaningful way. Somehow the number of managers, executive assistants, and underperformers is a mathematical constant
In the software industry we see the emergence of Service as Software replacing traditional Software as a Service. The key difference being SaS uses AI to automate tasks and workflows whereas traditional SaaS digitized what we used to do with paper and phone calls. SaaS made people more efficient, SaS is intentionally focused on replacing people with AI agents.
I’m an AI optimist. The glass is half-full until it breaks.