The “silver tsunami” in U.S. manufacturing is the wave of retirements among older industrial workers, roughly 27% of the manufacturing workforce is now over 55. According to a 2024 study by Deloitte and the Manufacturing Institute, the industry may need 3.8 million new workers over the next decade, with as many as 1.9 million of those jobs going unfilled. The deeper risk isn’t just headcount but the decades of operational knowledge held only in the heads of senior technicians and employees, the kind of knowledge loss that translates directly into longer onboarding, higher unplanned downtime, more quality incidents, and in the worst cases, safety failures.
The National Association of Manufacturers’ quarterly outlook survey has had “attracting and retaining talent” as the top-cited business challenge for more than seven years running, with 65% of manufacturers calling it their primary concern in 2024.
What is actually leaving the building
The retirements are the visible event, however, the asset walking out the door is something most plants have never put on a balance sheet. It is the deep operational knowledge built by repetition, mistakes, and accumulated judgment.
Broadly speaking, this breaks into four buckets:
- Process knowledge: The actual sequence of how a job gets done well, including the small deviations from the official SOP that experienced operators have learned matter for yield or quality.
- Troubleshooting knowledge: The pattern recognition that lets a senior maintenance tech walk up to a machine, listen for three seconds, and know it’s the upstream bearing rather than the obvious vibration alarm.
- Sensory and contextual knowledge: How the resin smells when it’s about to flash, what “normal” looks like on a vision-system output at 3 a.m. when the line has been running hot for six hours, when to override an interlock and when absolutely not to.
- Relational knowledge: Which person at the supplier actually picks up the phone, which parts are best ordered from which distributor, the workarounds for the legacy PLC that the original integrator never documented.
Standard operating procedures, when they exist at all, usually capture only the first bucket and often capture it imperfectly. The other three live in people’s heads.
The cost shows up in three places
(1) Unplanned downtime and longer MTTR
When the operator who could fix the bagging-line jam in four minutes retires, the replacement takes forty. That doesn’t show up as a knowledge problem in the line-status dashboard and only shows up as a rising mean-time-to-repair number (MTTR) and a falling first-time-fix rate. Operations leaders often spend months chasing the wrong root cause like inspecting equipment, blaming suppliers before realizing the variable that changed was the person who left.
(2) Onboarding time and ramp
Mid-sized manufacturers consistently report 6 to 18 months to bring a new hire to full productivity on a complex line. In a market with sub-1% manufacturing unemployment in many regions and an aging cohort heading for the door, every extra month of ramp is a month of constrained output. The Manufacturing Institute’s data on apprenticeships suggests most plants want the structure but lack the senior bandwidth to teach the way they used to.
(3) Safety incidents
This is the bucket no production unit wants to be in. On September 20, 2022, the BP-Husky refinery in Toledo erupted in flames, killing two operators and causing more than $600 million in damage. The US Chemical Safety Board investigation pointed to operating conditions that an experienced operator would have recognized as dangerous but the team on shift did not have the depth of experience to recognize the warning signs in time. Most manufacturers will never face anything that catastrophic. But every plant has near-misses that experienced operators avert without ever filing a report, and every retirement thins that protective layer.
Why traditional knowledge-transfer programs fail
Many plants are not well equipped to quantify the cost of knowledge-transfer the problem.
The retire-and-document sprint. When a key operator gives notice, leadership tasks an engineer or HR partner to “document what they know” in the four weeks before they leave. The output is usually a Word document of generic steps that anyone with a manual already knew. The actual knowledge that is the judgment, the workarounds and the sensory cues are just not captured in an interview because often the expert operator doesn’t consciously know they’re using it.
Shadowing without structure. Pair a junior tech with the senior for a few weeks and hope the knowledge transfers. It can work, but it is slow, capacity-constrained (your most experienced people are also your most valuable producers), and it transfers knowledge to one person. When that one person leaves, you’re back to square one.
The video library. A plant decides to film senior operators doing their work. Within six months, the result is a SharePoint folder of 200 unindexed videos that nobody can search and nobody watches. The knowledge is technically captured but operationally inaccessible which, for a frontline worker debugging a problem at 2 a.m., is the same as not being captured.
What’s changing: knowledge capture as something that happens by default
The shift underway in 2026 is from knowledge transfer as a project (something a plant does in a panic when someone gives notice) to knowledge capture as a default state of operations.
Three things are making that possible. First, head-mounted cameras have become light enough, cheap enough, and rugged enough to be worn through a full shift without operators noticing. Second, AI models are now good enough to watch hundreds of hours of egocentric video and extract structured procedures, identify decision points, and surface answers to natural-language questions like “how did Tony deal with the bearing alignment on line 3 last quarter?” Third, the marginal cost of storing, indexing and analyzing all of that data has collapsed massively.
The practical effect is that a plant doesn’t need to wait for someone to retire to start capturing. Operators work normally, the capture happens passively, and the knowledge accumulates as a queryable asset rather than a series of unfinished documentation projects.
This is the category Myto operates in. Our system uses AI smart glasses such that the capture is egocentric and non-intrusive: it records what the operator sees, including the small adjustments, the hesitations and the workarounds. That is the layer traditional documentation misses, and it is exactly the layer that walks out the door at retirement.
Myto works with mid-sized U.S. manufacturers to capture operator knowledge as it happens no documentation projects, no on-site burden. Book a 30-minute call.
What plant leaders should do this quarter
Three steps, in order, that don’t require a tooling decision to start:
- Inventory the risk. List the 10 to 20 roles where a single departure would meaningfully impact output, quality, or safety. Note the eligibility-to-retire window for each. This is your knowledge-loss heat map and it usually surprises the management teams. Most plants have more single points of failure than an org chart suggests.
- Triage by recoverability. For each high-risk role, ask: if this person left in 30 days, what knowledge would be irrecoverable, what is recoverable with effort, and what is already documented? Most plants find the irrecoverable bucket is alarmingly large.
- Start capturing before you need to. Whatever method you pick: structured interviews, video capture or AI-powered platforms, begin with the highest-risk roles and start while the experts are still working normally, not in their last two weeks. Knowledge captured under normal operating conditions is dramatically richer than knowledge captured under deadline pressure.
The shift in mindset that matters most
The plants that come through the next ten years in the best shape will be the ones that stop treating institutional knowledge as a soft asset that lives in HR conversations, and start treating it as an operational asset that gets maintained the same way as preventive maintenance: continuously, on a schedule, with clear ownership.
The U.S. manufacturing silver tsunami is not really a workforce story. It is an information story.
The plants that figure out how to keep the information when the people leave will be the ones that fill their open positions. They will be able to ramp new hires faster, because their lines run more reliably, and because the quietly accumulated knowledge that made their operation work is well documented for everyone across the production workforce to use.
If your plant is staring down a retirement wave and the documentation isn’t where it needs to be, that is exactly the problem Myto is built for. We work with mid-sized U.S. manufacturers to capture operator knowledge as it happens no documentation projects, no on-site burden. Book a 30-minute call →
Frequently asked questions
How many manufacturing workers are over 55?
Roughly 27% of the US manufacturing workforce is 55 or older, according to the Manufacturing Institute. That is significantly higher than the cross-industry average and the primary driver of the so-called “silver tsunami.”
How many manufacturing jobs are projected to be unfilled by 2033?
Deloitte and the Manufacturing Institute project a net need of 3.8 million additional manufacturing workers between 2024 and 2033, with as many as 1.9 million of those jobs going unfilled if workforce challenges are not addressed.
What is the difference between tribal knowledge and standard operating procedures?
A standard operating procedure captures the official steps of a process. Tribal knowledge is everything that surrounds and modifies those steps in practice: the small adjustments experienced operators make, the troubleshooting heuristics they’ve built through repetition, and the sensory and judgment-based cues that don’t fit neatly into a written document.
Is the manufacturing skills gap getting worse?
By most measures, yes. The 2021 estimate of the long-run economic cost of the skills gap was approximately $1 trillion; the 2024 follow-up by Deloitte and the Manufacturing Institute did not revise that figure downward, and unfilled manufacturing positions stood at roughly 409,000 in August 2025.
What is the most effective way to preserve operator knowledge before retirement?
The most effective approaches share three traits: they capture knowledge during normal work rather than in a documentation sprint at the end of someone’s tenure, they capture context (sensory, judgment, decisions) and not just steps, and they make the captured knowledge searchable so a frontline worker can actually retrieve it in the moment they need it. Modern egocentric-video and AI-indexing platforms aim at all three.