How AI Will Actually Change American Manufacturing

Every time a major technology arrives, people worry it will wipe out jobs. We have been here before. The computer and the internet sparked the same fears, and what actually happened tells us a lot about where AI is taking manufacturing now.

The fear makes sense. But history keeps pointing the other way. We don't end up with fewer jobs forever. We end up with different jobs, and usually more of them. Manufacturing looks like the place where a lot of that new work is about to get created.

The computer already taught us how this goes

In 1987, economist Robert Solow joked that you could see the computer age everywhere except in the productivity numbers. Companies had spent two decades buying computers, and the payoff still hadn't arrived.

It finally showed up around 1995. Big technologies take years to deliver, because companies have to rebuild how they work before the gains ever show up.

When those gains came, they reshaped the workforce. MIT economist David Autor showed that computers were good at automating routine office and production tasks. Those were the middle-skill jobs that once anchored the working class, and they shrank.

That part is worth sitting with, because the transition is genuinely painful for the people caught in it. When your role gets automated away, "the economy creates new jobs" is cold comfort if you're the one out of work now, and the new jobs rarely show up in the same town or the same week you lose the old one. Any honest version of this story has to hold both things at once: the disruption is real for individuals, and the long-run trend still creates more work than it destroys.

That second part is the one that gets forgotten. Autor found something else. About 60% of the jobs Americans hold today didn't exist in 1940. We have never protected employment by freezing old jobs in place. We create new ones, and over time we create more than we lose.

That track record is the reason to be optimistic. What's different this time is where AI is hitting first. For forty years, the service economy absorbed the workers who lost manufacturing and clerical jobs. Now AI is putting pressure on those same white-collar roles. That is exactly why attention is swinging back to making things in America.

Why manufacturing, and why now

Manufacturing is ready for it. Factories already produce a record volume of goods with a workforce that companies still can't fully staff.

Deloitte and The Manufacturing Institute project that U.S. manufacturing will need 3.8 million net new employees between 2024 and 2033. Roughly half of those roles could go unfilled if the skills and applicant gaps don't get solved. The hardest to fill are the skilled technical roles that run and maintain equipment.

Reshoring is already bringing work home. In 2024 alone, companies announced roughly 244,000 manufacturing jobs coming back to the U.S. through reshoring and foreign direct investment, and 88% of those jobs were in high-tech or medium-high-tech sectors.

The real barrier was never just labor cost

What makes this moment different is that the real barrier to manufacturing in America was never just wages. It was complexity.

Most factory software projects miss their goals. Unplanned downtime costs U.S. manufacturers an estimated $50 billion a year. Decades of knowledge walk out the door when a veteran retires.

AI goes straight at those problems. Fewer failed systems. Machines that flag a failure before they break. Less paper, smoother scheduling, and your best people's knowledge captured in software instead of lost when they leave.

It's already showing up in the numbers

This isn't theoretical. Manufacturers investing in smart production and AI are reporting real gains right now, not someday.

Deloitte's 2025 Smart Manufacturing Survey covered 600 executives at large U.S. manufacturers. After putting smart manufacturing initiatives in place, they reported average improvements of 10 to 20% in production output, 7 to 20% in employee productivity, and 10 to 15% in unlocked capacity.

Predictive maintenance is one of the clearest wins. Deloitte reports that these programs can increase equipment uptime by 10 to 20%, and industry studies across multiple sectors cite 40 to 70% reductions in unplanned downtime once a program matures.

When a plant gets that much cheaper and simpler to run, starting one becomes more realistic. That makes building in America a stronger bet than going overseas. Rising wages abroad keep shrinking the old cost gap, and the work it creates pays well.

The payoff takes time, and early movers pull ahead

The same pattern from the computer era is playing out again. Most companies investing in AI today say they haven't seen major returns yet. McKinsey's 2025 State of AI survey found that only 39% of organizations attribute any enterprise-wide EBIT impact to AI so far.

The gains are real. They just go to the companies that rebuild how they work, not the ones running a pilot and waiting to see what happens.

McKinsey and MIT's MIMO program found that the highest-performing companies using machine intelligence improved KPIs by 9.5% on average across 21 metrics, against 3.5% for everyone else. That's close to three times the impact, and a follow-up study found the gap widening to 3.8x as the leaders pulled further ahead.

Every wave of technology so far has ended the same way, with us building something bigger on the far side of the disruption. For the first time in a generation, manufacturing is lined up to be where a lot of that new work gets created.

If you want to figure out where AI fits in your operation, whether that's predictive maintenance, scheduling, knowledge capture, or something else entirely, get in touch. I'll tell you exactly where I'd start.

For related reading: What to Automate First and Why Your Business Data Isn't Being Used.