Direct Derek - Picks and Shovels

After the Ribbon: The Maintenance Annuity Behind Energy Infrastructure

2026-06-06T00:00:00+00:00

The ribbon gets cut. Construction photographs circulate. The crews collect their equipment, the temporary offices disappear, and the financing moves on to the next announcement.

The asset stays where they left it.

A turbine, solar array, battery system, substation, or transmission asset now has to operate through heat, vibration, wear, alarms, inspections, and component failures. Machinery remains politely indifferent to the narrative that financed it.

This is where the analysis becomes interesting. Large capital programs attract large crowds. The question worth asking is how many commissioned assets a qualified team can actually reach, inspect, and service.

Long after the cameras leave, a service vehicle approaches the gate.

The Number Underneath the Ribbon</h2>
The Inflation Reduction Act was scored at roughly $370 billion for clean energy in 2022, though subsequent legislation has curtailed much of that credit structure — one more reason to underwrite commissioned assets rather than announced policy. That expands the potential installed base, although it does not prove that every announced project gets built, earns an acceptable return, or creates work for an independent provider.
Equipment already commissioned or visibly under construction is the safer starting point. Steel in the ground is more persuasive than enthusiasm in a press release.
Operations and maintenance accounted for roughly 20–25% of lifecycle cost for European wind and solar plants as of 2017, a share that rises as capex per kW falls. For a hypothetical project with $100 million in lifecycle cost, that implies $20–25 million of O&M over its operating life. A quarter of the whole project economics sits on the far side of the ribbon-cutting.
That is not one cheque waiting for one contractor. It arrives across years, sites, equipment categories, monitoring, inspections, preventive work, corrective repairs, and component servicing. Each piece must be won and performed separately.
Construction is a project. Maintenance is a calendar.

"Recurring" Needs an Autopsy</h2>
Recurring revenue is one of those phrases that becomes less informative each time it appears in a presentation.
Energy O&M includes monitoring, scheduled servicing, corrective work, and maintenance of turbines, inverters, battery systems, substations, and lines. Those streams do not deserve the same valuation merely because they happen more than once.
Scheduled work may be predictable but competitively priced. Emergency work can produce attractive invoices while damaging crew utilization. Some work may be performed internally. Equipment requirements can also limit which providers are qualified to touch a particular asset.
The physical demand floor remains. Deferral does not abolish wear; it tends to convert manageable work into emergency labour, expedited parts, downtime, and a customer who has suddenly discovered the value of planning.
Industrial maintenance benchmarks show the broader mechanic. Maintenance cost as a percentage of replacement asset value is the standard yardstick. SMRP's top-quartile range runs from about 0.7% to 3.6% depending on industry, with roughly 2–3% a common world-class marker; benchmarks for reactive plants vary across sources from 6% up to 10%. SMRP also warns that a low ratio can mean under-maintenance rather than excellence. A plant at 6% is usually not maintaining twice the equipment of a plant at 3%. It is often performing substantially the same work at two or three times the unit cost because everything has become urgent.
A plant benchmark should not be transferred mechanically onto every wind or solar site. The cost logic still travels well. A capable servicer creates value by moving work onto a schedule rather than waiting for something expensive to fail.
Separate revenue into scheduled, corrective, emergency-driven, internally captured, and otherwise restricted buckets. Until the mix is known, "recurring" is decoration.

The National Market Disappears at Road Level</h2>
Aggregate spending numbers are useful for conference slides and nearly useless for describing the work a crew can perform on Tuesday.
The executable market consists of assets within a practical response radius, filtered by technician qualifications, equipment specialization, travel time, and available service windows. A national installed base can be enormous while the economically reachable market remains stubbornly local.
Density changes the economics. More serviceable assets inside a workable radius mean higher technician utilization, faster response, and less time behind a windshield. A scattered backlog can report impressive revenue while producing weak economics after travel and standby requirements consume the schedule.
A disciplined analyst would rather see a modest territory with dense routes than a heroic map covered in dots.
Capital has limited power over this constraint. It can buy vehicles, tools, inventory, and acquisitions. It cannot instantly create qualified technicians, local trust, or familiarity with an installed equipment base. A platform can acquire several crews, but it has acquired several local operating networks that still have to function locally. The logo is the easy part.
That creates the opportunity and sets the ceiling. A collection of contracts may support a good operator while remaining immaterial to a fund that requires scale to justify the work. The market stays protected because it is geographically bounded, relationship-dependent, and too small to absorb much capital without changing its character.
The better capacity denominator is backlog divided by available technician-days after travel and standby. If that figure deteriorates as revenue grows, the company is stretching the map instead of building density.

The Installed Base Is Not the Addressable Market</h2>
The maintenance tail is attractive because it outlives the construction cycle. That preference needs supervision.
Installed equipment creates an obligation to perform work, but it does not grant an independent servicer access at an attractive margin. Some demand remains internal. Other work requires specific qualifications or belongs to providers already embedded in the equipment relationship.
Underwrite commissioned assets and visible construction, therefore, rather than assuming every subsidy or forecast arrives intact. Then separate physical maintenance demand from the portion an outside provider can realistically capture.
The relevant variables are qualified headcount, route density, customer concentration, equipment mix, and the amount of work each crew can complete without wasting its week in transit. Hundreds of billions in policy spending may explain why more assets appear. It says very little about how much revenue fits inside one local service operation.
The recurrence can be genuine while the investable capacity remains quite small.

How the Service Thesis Fails</h2>
The first failure is confusing proximity with qualification. A nearby operator has no advantage if it lacks the credentials, equipment knowledge, or documented processes required for the work.
The second is mistaking site count for diversification. Ten facilities may still depend on one customer or one equipment category. The map looks broad until the ownership table is unfolded.
Growth can also worsen the business. Adding distant contracts may increase revenue while reducing technician utilization. Emergency work can interrupt scheduled jobs. A shortage of qualified staff can turn backlog into disappointed customers rather than future profit.
The questions worth asking are practical:

How many customers ultimately control the sites?</li>
Does growth improve route density or extend travel?</li>
Can the servicer charge for competence and response time?</li>
How much work can each qualified technician complete?</li>
Which revenue is scheduled, and which arrives only after failure?</li>
Does the customer relationship belong to the business or to one person?</li> </ul>
A fine local service company becomes a poor acquisition when priced as though geography, qualification, and trust have stopped mattering.
The Obligation Left on Site</h2>
After construction ends, the asset remains fixed behind a gate. It stays exposed to heat, wear, operating demands, and the consequences of delayed work. The service schedule continues without needing publicity.
Nothing here requires predicting which turbine, panel, developer, or policy narrative wins the decade. What matters is who can repeatedly reach the installed equipment, perform qualified work, and preserve the customer relationship without letting travel and standby consume the margin.
That is also where this framework is thinnest. The lifecycle share and the maintenance benchmarks are sector averages borrowed from a different industry and pointed at a site nobody in this argument has walked. Route density and technician-days are the numbers that decide the outcome, and they are exactly the numbers no filing publishes, which means the disciplined-sounding denominator above is a judgment wearing arithmetic as a costume.
The best territory is large enough to support the operator and small enough to remain inconvenient for capital that requires scale. Technician supply and geography limit growth, but they also protect the economics from competitors that need every opportunity to become a platform.
Construction crews leave an energy asset behind. For someone close enough and qualified enough, they also leave a long calendar of appointments.

Testing and Inspection Attached to Every Weld

2026-04-30T00:00:00+00:00

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</h2>
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</h2>
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</h2>
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</h2>
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</h2>
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.