The automation-labor paradox: Why US manufacturers are losing the reshoring race they started

American manufacturers announced plans to bring back approximately 240,000 jobs by 2025, which is about a 7% decrease from 2024¹. During that same period, factory employment decreased by over 70,000 workers, reaching 12.69 million—the lowest level since March 2022². Investment is surging. However, headcount is shrinking. This discrepancy is not a forecasting error; rather, it is the decade’s defining operational problem. At the heart of this issue lies the automation-labor paradox: the belief that robots and AI will solve the manufacturing labor shortage is conflicting with evidence that poorly implemented automation is exacerbating the problem. This article argues that manufacturers who win the reshoring race will not be the ones who buy the most automation. Rather, they will be the ones who develop the process discipline necessary to make automation work.

The automation bet – and why it’s underperforming

Among 214 surveyed US manufacturers, 92% call automation essential to long-term competitiveness, yet only 37% report having significant or full automation in place. That gap between strategic conviction and operational reality is the clearest single-number proof that adoption is not a question of intent. It is a question of capability.

The automation project failure rate tells the rest of the story. When projects miss expectations, manufacturers cite three causes most often: 50% point to the wrong technology being identified, 39% to a lack of internal expertise, and 32% to budget overruns³. None of those are robotics problems. Each describes a process and capability failure. Robotics in manufacturing does not select itself, install itself, or sustain itself. It inherits whatever discipline or whatever disorder already exists on the floor.

The root cause: Automating the wrong thing

The instinct, when labor is scarce, is to buy the machine that replaces the worker. The result, when the underlying process is unstable, is to lock waste in at machine speed. Automating inefficient processes does not eliminate the inefficiency. It encodes it, accelerates it, and makes it harder to see.

This is what lean manufacturing principles have always insisted on, and what lean automation integration formalizes: standardize the work first, then automate the standard. Skip the first step, and you are not investing in productivity; you are digitizing waste. MIT Sloan’s March 2026 expert panel reached the same conclusion from a different angle: risk-averse manufacturers investing to maintain the status quo rather than transform operations are the primary cause of America’s productivity gap, and a shortage of automation and software expertise compounds the dynamic⁴. The technology is available. Organizational readiness is the binding constraint.

What standardized work and TPM actually solve

Standardized work is not bureaucracy. It is the operational baseline that makes every other improvement measurable. Without standard work instructions, a 30-second cycle-time gain cannot be verified, a defect cannot be traced to a cause, and an automation cell cannot be designed against a stable target. With them, every variation becomes a signal.

Total Productive Maintenance (TPM) addresses the parallel problem on the equipment side. TPM lean practice transfers ownership of routine machine condition from a small maintenance team to the operators who run the equipment every shift. The metric that proves it works is Overall Equipment Effectiveness (OEE), the product of availability, performance, and quality. OEE improvement is the most direct readout of whether process discipline is real or theatrical. Value stream mapping reveals where losses concentrate before any capital is committed; in most plants, the chokepoints are not where dashboards indicate. They are in the handoffs, the changeovers, and the unplanned stops that no single department fully owns.

Capability as infrastructure

The labor side of the paradox is more structural than most boardrooms acknowledge. Roughly 26% of the US manufacturing workforce, about 3.9 million workers, is currently eligible for retirement, with 1.5 to 2 million roles projected to remain unfilled by the early 2030s⁵. The manufacturing skills gap is not a hiring problem. It is a capability transfer problem on a generational scale.

The economics make external hiring even less viable. The manufacturing wage premium, once roughly 40% in the 1960s and 1970s, has collapsed to about 2% nationally and is negative in coastal regions, where workers earn more at Amazon or Target than in entry-level manufacturing (MIT Sloan Management Review, 2026). Reshoring leaders confirm the bind: surveyed OEMs say a sufficiently skilled US workforce would prompt them to reshore 30% of currently offshored products — outranking tariffs, tax cuts, or any other lever⁶.

US manufacturers spent $32 billion on workforce training in 2026, a 22% increase from 2019⁷, and 44.7% still cite workforce quality as a top business challenge⁸. The spending is rising. The outcomes are not, because most of it is reactive and fragmented rather than systematic. Workforce transformation requires the same discipline as process improvement: structured, repeated, and observed at the gemba.

Where to start: From process stability to scaled automation

There is a sequence that works, and reversing it is the most expensive mistake in the room. Begin with a gemba walk to see the actual process, not the documented one. Use value stream mapping to expose where time, inventory, and effort accumulate. Establish standardized work where variation is most detrimental, and ratify it through short, structured kaizen events. Build Daily KAIZEN™ rhythms so frontline teams own the standards they wrote. Then, and only then, automate what is already stable.

Continuous improvement in manufacturing is not a slogan. It is the operating system that makes capital investment compound rather than evaporate. The companies winning the automation race are not the ones with the most robots. They are the ones who built operational excellence in manufacturing, relying on gemba kaizen as a daily habit, capability transfer as an explicit discipline, and automation as the last step rather than the first. The retirement cliff and the reshoring surge will not wait. The manufacturers who treat process stability and operator capability as infrastructure, not initiatives, will define the next decade. The rest will keep buying machines they cannot run.

To bridge this gap between automation ambition and operational reality, manufacturers must combine process discipline with advanced digital capabilities. At the Kaizen Institute, we support this transformation through our AI Solutions for Manufacturing, delivering customized digital innovations tailored to real production challenges. By integrating advanced analytics, artificial intelligence, and lean principles, we help organizations enhance quality control, enable predictive maintenance, optimize energy and resources, and improve demand forecasting and inventory management. The objective is not simply to automate, but to ensure that automation is built on stable, high-performing processes—empowering manufacturers to achieve sustainable productivity gains and operational excellence in an increasingly complex industrial landscape.

References

1. AMT. (2026). Tech-charged reshoring fuels skilled workforce. ↩︎
2. Manufacturing Tomorrow. (2026, January 29). Report: U.S. manufacturing faces hiring crisis as 26% of workforce nears retirement. ↩︎
3. Robotics Tomorrow. (2025, December 15). As 2026 approaches, U.S. manufacturers call automation critical: Yet most still lag in adoption, new report finds. ↩︎
4. MIT Sloan School of Management. (2026). Future of manufacturing: How to solve the U.S. productivity paradox. ↩︎
5. MIE Solutions. (2026). U.S. manufacturing labor shortages and hiring pressures in 2026. ↩︎
6. AMT. (2025). 2025 reshoring priorities. ↩︎
7. Manufacturing Dive. (2026). Manufacturers spent about $32B training workers: MI survey. ↩︎
8. National Association of Manufacturers. (2026). Manufacturers invest billions in workforce training. ↩︎