TL;DR:
Two current stories about the state of U.S. manufacturing seem to clash. One says “look forward to growth through 2026.” Another reports that manufacturing activity in December hit the lowest point of 2025 and is still in contraction. Both can be true, depending on time horizon and what you look at. Against that backdrop, AI and labor in manufacturing are not about a simple replace-people narrative. The structural reality is a long-term labor shortage and rising labor costs. AI can help manufacturing labor shortages by taking on repetitive, attention-heavy tasks, so scarce people can focus on higher-value work.
If you only read headlines, the status of U.S. manufacturing in 2026 looks really confusing.
On one side, a recent Quality Magazine piece quotes ITR Economics’ Taylor St. Germain saying the manufacturing outlook for 2026 is positive. His message: “Be prepared for manufacturing growth.” He expects manufacturing and industrial demand to accelerate through 2026, with manufacturing “just crossed into positive territory” after being negative for most of 2024 and early 2025.
This sounds like the party is about to start!
On the other side, a new Manufacturing Dive article on the December 2025 Institute for Supply Management (ISM) Purchasing Managers’ Index reports something different: ISM’s manufacturing index came in at 47.9% in December, well below 50 for the tenth month in a row, signaling continued contraction. Demand components such as new orders and backlogs improved slightly, but stayed in contraction. ISM’s Susan Spence explicitly asks whether this is the start of a turnaround or “just another blip,” and notes that manufacturing contracted for most of 2025 even as the overall economy grew.
So, are we still in the ditch after all?
So manufacturing sits on top of two apparently conflicting foundations: forecasts of future growth and data showing current weakness. The first step is to make sense of that, a second step will be to relate this to AI and labor in manufacturing – a key issue in all reports.
How both stories can be true
The two perspectives are not actually mutually exclusive, let’s unpack them.
Quality Magazine and ITR are talking about the next several years:
- Their analysis relies on industrial production, manufacturing indices, and a set of leading indicators that point to accelerating industrial and manufacturing demand “through really the entirety of 2026.”
- The article acknowledges that the past 12–18 months have been tough, but argues that the data now point to a turn and that manufacturers should prepare for growth.
Manufacturing Dive, by contrast, reports where the sector stands right now:
- ISM’s PMI is a monthly diffusion index based on survey responses from purchasing managers. A reading below 50 means more respondents report contraction than growth. December’s 47.9 was the lowest reading of 2025 and marked the tenth straight month below 50.
- Comments from ISM and S&P emphasize soft demand, tariff uncertainty, and the risk that current production levels may not be sustainable without better orders.
Deloitte’s 2026 manufacturing outlook sits between these two views. It notes that the PMI remained below 50 for much of 2025 and that manufacturers faced price and demand pressure, but also finds that 80% of surveyed executives plan to invest at least 20% of their improvement budgets in smart manufacturing in 2026 to improve competitiveness and resilience.
Taken together, a more coherent picture emerges:
- Short term: as of late 2025, the factory sector is still struggling. Orders and sentiment remain weak.
- Medium term: several forecasters see potential for stronger manufacturing and industrial demand through 2026, driven by regional investment, policy changes, and ongoing reshoring trends.
That means the “party is starting” and “we’re still in the ditch” stories are looking at different points on the same timeline. For manufacturing leaders, the more important question is what stays true across both states of the world.
The structural labor story does not change
The most stable piece of this puzzle is not the demand side, but the workforce.
Several data points line up:
- Deloitte and The Manufacturing Institute both estimate that the U.S. manufacturing skills gap could leave up to 2.1 million jobs unfilled by 2030, with a potential cost of about $1 trillion in that year alone.
- Deloitte’s more recent analysis notes that, as of September 2025, there were about 12.7 million people on U.S. manufacturing payrolls, well below the 17.2 million in 2000, even after post-pandemic gains. Any workforce growth is expected to concentrate in higher value-added, capital-intensive segments.
- The Quality Magazine article calls labor the biggest challenge facing manufacturers in 2026. They see a roughly a 28% increase in manufacturers’ labor costs between now and the end of the decade, driven by strict immigration, baby-boomer retirements, and a smaller cohort of younger workers.
The implication is clear: whether demand is weak or strong in any given quarter, the long-term picture is a structurally tight labor market, rising labor costs, and difficulty finding enough people with the right skills willing to work in manufacturing.
That is the backdrop for thinking about AI and labor in manufacturing. The story is not just about weathering a slump or catching a wave. It is about operating in an environment where people are scarce and expensive for years to come.
How AI can help manufacturing labor shortages in both scenarios
Against that structural background, AI becomes less about a one-off cost-cutting tool and more about a way to allocate scarce human capacity intelligently. The use cases that matter most tend to share a few traits: they are repetitive, attention-intensive, and highly sensitive to mistakes.
Visual inspection and quality checks
Surveys of smart manufacturing projects consistently show quality control as a leading application area for AI, specifically machine learning.
Visual inspection is difficult to staff and scale using manual methods. Inspectors get tired; sampling leaves gaps; criteria vary by operator and shift. However, AI-based visual inspection systems can:
- Inspect 100 percent of the products at line speed
- Apply the same quality criteria across lines, plants, and shifts
- Escalate only those images or cases that need human review
When the sector is “still in the ditch,” this helps manufacturers maintain quality and throughput with fewer errors and less rework, even when open inspector positions are hard to fill. If the “party” does start and demand increases, the same systems allow plants to accept more orders without requiring a proportional increase in inspection headcount.
Inbound and outbound shipment verification
Dock operations are another area where AI can help manufacturing labor shortages without removing people altogether.
Cameras and AI models can assist with:
- Checking pallet counts and basic configuration against the order
- Reading and validating labels or barcodes
- Flagging likely mis-ships or documentation mismatches
This reduces the time warehouse staff spend on routine and often tedious counting and checking, allowing them to focus on exceptions, claims, and process improvements. Again, this is useful in a slow year and essential in a busy one.
Maintenance and reliability engineering
AI-based predictive maintenance does not eliminate the need for technicians. It does change the way their time is used.
By continuously monitoring sensor data, AI models can:
- Detect early signs of abnormal behavior
- Prioritize which assets need attention first
- Support more planned interventions during scheduled downtime
In an environment where experienced technicians are retiring faster than replacements can be trained, this is one way AI and labor in manufacturing can work together: the system handles the continuous monitoring; humans decide what to do with the insights.
Planning when the macro picture is muddy
If the demand outlook is genuinely uncertain, one economist says “prepare to party,” and another still signals contraction, how should manufacturing leaders think about AI investments?
A few principles tend to hold across cycles:
- Favor use cases where AI removes repetitive, attention-heavy effort from people, not just headcount. That usually means quality checks, routine verification, and data scanning rather than entire jobs.
- Treat labor redeployment as part of the business case. If AI takes over certain tasks, where will those people add more value?
- Evaluate projects on their ability to improve output per person over several years, not just in the next quarter. That aligns with the structural labor data rather than this month’s PMI.
- Be explicit about governance and validation, for the trams on the shop floor trust in AI systems will depend on clear problem definitions, sensible performance metrics, and ongoing monitoring, rather than marketing claims.
If you want to talk to real practitioners who have implemented AI solutions on the shop floor and discuss your use cases, please get in touch.
Here you can find more information about the Accella Quality Box(TM) and Accella Dock Check(TM), also check our our use cases that provide tangible, real-life examples of AI implementations in manufacturing.