Testing and Inspection Attached to Every Weld

A weld can look finished while the job remains legally and commercially incomplete.

The metal is joined, and the component sits exactly where the drawing dictates, perhaps already painted. None of that matters if the customer cannot accept, commission, or restart the asset until someone qualified examines the work, interprets the result, and signs the record.

"Inspection attached to every weld" is shorthand, not a claim that every joint gets identical treatment. ASME Section IX, the B31.x piping codes, and API standards set requirements that vary by material, service, pressure, hazard class, and owner specification. A low-pressure water line might need nothing more than a visual check. A high-pressure hydrocarbon line doesn't get to see fluid until someone runs ultrasonic or radiographic testing and the report clears. The customer doesn't choose whether evidence matters, nor do they negotiate the extent of the examination. The code dictates the percentage of welds to be tested based on piping class, on a schedule that has nothing to do with the fabrication calendar.

That decision tree is the business.

Fabricators build the visible asset. Inspection sits at the choke point behind it, where evidence becomes permission. A general contractor can weld, but it rarely keeps qualified non-destructive testing (NDT) personnel on staff. A new entrant can buy the equipment, but it cannot buy a spot on a plant's approved vendor list or the trust that got someone else onto that list. The large multinational inspection firms can absorb the massive contracts, but a short turnaround at a single plant, staffed on a few days' notice, is often too small and too irregular to be worth their mobilization cost. The plant waiting on its restart clock is stuck choosing from a short list. Nobody with real scale is fighting to be on it.

The shadow ledger behind the weld

Industrial verification runs a second workstream behind fabrication, and it doesn't close when the weld does.

The sequence is rigid: qualify the procedure and the welder, make the joint, run the required examination, and interpret the result against acceptance criteria. If something fails, the repair triggers a re-examination and another entry in the ledger. The methods — visual, ultrasonic, radiographic, magnetic particle, liquid penetrant, eddy current — are mechanical enough to describe in a sentence each. What matters is which one the code demands and who is credentialed to run it. Owning an ultrasonic rig doesn't make a firm interchangeable with a radiography crew.

The backbone is old. The ASME Boiler and Pressure Vessel Code dates to 1914, born out of an era when boiler explosions were a routine industrial hazard. Catastrophic failures since — pipeline ruptures, refinery fires, structural collapses — have ratcheted the requirements upward rather than down. That makes the demand counter-cyclical to safety tolerance rather than to the broader economy.

New fabrication supplies the first inspection event, but the real volume lies in the recurring work. Corrosion, fatigue, turnarounds, and fitness-for-service reviews bring the inspector back years later. A refinery turnaround can mean thousands of welds and dozens of vessels crammed into a shutdown window measured in days. What the operator is actually selling is permission to restart, not inspection. And the loop never fully closes: sign the report, put the asset back in service, and the same code has already scheduled the next visit.

The market-size slide tells almost nothing

The global NDT and inspection market runs around $15 billion in 2025, projected near $22.3 billion by 2030 — roughly an 8.3% compound rate. The broader testing, inspection, and certification (TIC) category is fuzzier. Estimates for 2024 range from about $247 billion to $398 billion depending on what each research house counts as in scope, with forecast growth generally clustering between 3.5% and 6.1%.

A spread that wide serves as a warning label rather than a useful metric. "TIC" bundles different services, industries, and geographies under one heading, and a local inspection crew doesn't own a percentage of a global category just because both words appear in the same slide.

The broad category is exactly the part institutional capital can touch. It offers big enough tickets, established multinational players, and real liquidity. The inefficiency worth studying sits further down: one method, one industrial cluster, one approved crew, one turnaround calendar. So the question that actually needs answering has nothing to do with a market forecast. How many qualified inspection hours can this specific operator put on this specific plant floor by Tuesday?

Effective capacity wears a certification

NDT capacity gets reported as headcount. That is too generous a unit.

The tier structure in ASNT's SNT-TC-1A recommended practice defines what a technician may do — though Levels I and II are certified by the employer under its own written practice, not by ASNT centrally, which is part of why a technician's standing does not travel freely between shops. A Level I works under supervision, while a Level II sets up and calibrates equipment, evaluates results against the acceptance criteria, and supervises Level I personnel. Above them, a Level III provides the technical oversight, procedure development, and interpretation authority that allows the other two to operate.

The workforce carrying those credentials is aging out. PQNDT's salary and benefits survey put the average NDT technician's age at 47, and ASNT's own education writing describes a technician shortage driven by retirements. Buying a fleet of phased-array rigs takes an afternoon. Acquiring the years of exposure required to read a difficult weld profile under a compressed schedule is entirely different, and you certainly cannot buy a plant manager's willingness to trust that judgment.

Real capacity looks more like an arithmetic problem than a revenue line: qualified technicians, times workable shifts, times method-specific certification, times customer approval, times utilization. That theoretical maximum is then reduced by travel, calibration, recertification, safety briefings, documentation, and the schedule conflicts that never make it onto a projection. The workforce shortage acts as a tailwind for pricing, but it is simultaneously a hard ceiling on how fast any single operator can grow.

The forecast compounds cleanly at 8.3% a year. Technicians don't.

Scarcity can fool the analysis too

Constrained markets are the whole subject here, which is exactly why the framework has to be checked against the version of this trade where the constraint is real but the pricing power belongs to someone else.

A mandatory inspection requirement doesn't automatically transfer to owner pricing power. Customers can defer non-critical work, trim scope where the code allows it, or squeeze vendors during a rebid. A roll-up that buys ten small inspection shops can end up owning ten separate technician bottlenecks rather than one consolidated barrier. If the acquired revenue rides on a few senior Level IIs or a single Level III, those people can walk, taking the customer relationship and the judgment with them. The corporate entity changes hands; the scarce human asset keeps the leverage to leave on its own schedule.

Scarcity pushes bill rates up, but it also means the labor understands its own leverage. Wage inflation, overtime, per diem, and subcontracting markups can eat the premium before it ever reaches an owner's income statement.

So revenue and utilization should be measured per certified technician, split by method and location. Look for tenure, turnover, training pipeline, and overtime burden. A disciplined analyst needs to know how much of the book is recurring in-service inspection and turnaround work versus one-off new construction. Check approved-vendor status, safety records, rejected reports, and the share of work actually locked under a master service agreement. The termination clauses in those agreements require close reading, because plenty of them let the asset owner walk on short notice. The question underneath all of it is whether this is a protected local choke point or a rented labor shortage that the employees can monetize faster than the owner can.

Where small capital still fits

The patch worth occupying is mandatory, local, and just inconvenient enough to filter out the money that needs clean scalability. The work is too specialized for a general contractor to staff efficiently, yet too fragmented for a multinational inspection firm to prioritize. At the same time, the consequences of failure prevent the asset owner from handing the job to an unproven low bidder.

The operator worth studying sits inside a constrained industrial cluster, holds several experienced Level IIs and some Level III oversight, and shows a mix weighted toward repeat turnaround and in-service work rather than one-time new builds. Its growth plan should be sized to the crews it can actually recruit and qualify, not to a category forecast lifted from someone else's deck.

Adding two crews and a calibration lab can move a small independent operator's earnings materially. That same growth registers as rounding error to an institution whose smallest workable check is larger than the whole business. That mismatch is the entire trade. The edge disappears when the work standardizes enough for national bidding, or when ticket sizes swell to a point that draws multinational attention. It also vanishes if growth demands more qualified labor than the local market can supply.

The honest weakness in all of this is that "too small for institutions" and "too small to matter" are the same sentence read in two moods, and the arithmetic above cannot tell you which one you are holding.

Count deployable hours before addressable market.

The weld stays where the fabricator left it. The owner wants to move on. Someone with the right credentials still has to show up, run the method, interpret the signal, and sign a report that a regulator, an insurer, and a plant manager are all willing to trust. Then the valves open, the asset goes back into service, and the same code has already scheduled the next time someone has to come looking at it.