The Inspection Illusion, Automated
A $300 million robotics contract will generate more hull and weld data than the Navy has ever had. Nothing in the record says anyone is obligated to act on it.
The Inspection Illusion, Automated
More data does not mean more accountability. It just means more data.
MD Marine Electric | The Current Conversation
The US Navy has expanded a maintenance contract with Gecko Robotics to $300 million.¹ The company's drones, wall-climbing robots, and fixed sensors crawl decks, welds, and hulls, collecting structural data at a density no human inspection team could match.¹ That is the headline. It is also, on its own, not news anyone in the fleet maintenance world should celebrate without a second question: what happens to the data once it exists.
We have been here before. In 2020, Navy leadership changed inspection procedures specifically to reduce the number of inspections performed, cutting them by nearly half, in order to preserve working relationships with contractors.² That is not a staffing shortage story. That is a documented policy choice to see less, on purpose, because seeing more was inconvenient for the business relationship. Call it what it is: the Inspection Illusion, the habit of treating inspection activity as a proxy for accountability when the institutional structures that would act on inspection findings remain untouched. The Gecko contract does not repeal that policy. It automates the collection half of a pipeline whose output half was deliberately narrowed four years before the robots showed up.
What the robots actually do
Trade-level specificity matters here, because the contract is not abstract. Gecko's platforms perform non-destructive structural assessment: wall-climbing units for vertical steel surfaces like bulkheads and tank interiors, drones for overhead and hard-to-access spaces, fixed sensors for continuous monitoring of welds, decks, and hull plating.¹ This is the kind of condition-based data a port engineer would once have had to assemble from a rotating crew of inspectors with ultrasonic thickness gauges and a clipboard, availability by availability, compartment by compartment. Robotic platforms do not get tired, do not skip a frame because the tide schedule is tight, and do not need the same per-diem budget line a human survey team requires.
That is a genuine capability gain. It is also exactly the kind of capability gain that, absent a corresponding commitment on the acting side of the ledger, produces more evidence of problems without producing more repairs. A hull survey that flags corrosion in a void space is only useful if someone with authority to fund the fix reads it, believes it, and acts before the next availability closes out. The 2020 inspection-reduction decision shows that the Navy has, within recent memory, chosen the opposite: fewer findings, preserved contractor goodwill, closed-out schedules.²
The data pipeline has no owner
This is where the Inspection Illusion compounds into something closer to the QAR Vacuum, the gap between inspection authority and inspection staffing that this series has documented before. Quality assurance representatives are the people who are supposed to translate inspection findings into enforceable contract action. If the volume of structural data entering the pipeline increases by an order of magnitude, as a $300 million robotics surge implies it will, and the number of QARs reviewing that data does not increase to match, the practical effect is a larger backlog of flagged conditions that nobody with signature authority has time to process.
The GAO's broader maintenance record supports the concern that volume alone does not fix throughput. Even with dock space available, fewer than 40 percent of ships finish their scheduled maintenance availabilities on time.³ That figure describes a system that already struggles to convert known, documented work into completed work within the window allotted. Adding a denser stream of newly discovered structural findings to that system, without adding the capacity to adjudicate and fund the resulting repairs, does not shorten availabilities. It lengthens the queue of things everyone agrees need fixing and nobody has scheduled.
Fixed-price bidding meets a flood of new findings
The timing compounds the exposure. The Navy has been shifting ship repair work from cost-plus contracting toward fixed-price, per-ship bidding, explicitly to widen the competitive pool of yards able to bid.⁴ That shift is the structural driver behind what this series has called the Bid-to-Win Trap: contractors win fixed-price work by underestimating scope, then recover margin through change orders once the ship is already in the yard and the Navy has no practical alternative but to pay.
Now introduce a robotics platform that discovers corrosion, weld fatigue, and coating failure at a resolution no prior survey method achieved. Every one of those findings is a potential change order under a fixed-price contract, because fixed-price bids are built against the scope known at bid time, not the scope a wall-climbing sensor platform uncovers six weeks into the availability. The Change Order Economy, the pattern by which contractors recover fixed-price losses through post-award scope growth, does not need the Navy to act in bad faith. It only needs better detection tools arriving after the contract price is locked. A $300 million robotics investment, layered onto a fixed-price bidding structure, is a near-guaranteed generator of change order volume, unless the Navy restructures how discovery findings are priced into the contract before the robots start climbing.
The Ticonderoga warning
The cost of discovering problems without a funded path to fix them is not hypothetical. The Navy spent $1.84 billion modernizing four Ticonderoga-class cruisers that were decommissioned before any of them deployed.⁵ That failure was not a data problem. The Navy knew the condition of those hulls. What it lacked was an institutional mechanism connecting known condition to a funding and scheduling decision that made the investment pay off before the ships were retired anyway. Robotic inspection solves the "knew the condition" half of that equation better than anything the fleet has had before. It does nothing, by itself, to solve the half that actually sank the Ticonderoga investment: the gap between documentation and decision.
The CBO's December 2025 analysis of ship repair capacity makes the same point from a different angle, examining the industrial base's ability to execute the maintenance work the fleet already knows it needs, independent of how that work gets discovered.⁶ The constraint the CBO describes, limited yard capacity, limited skilled labor, is not a sensor problem. Gecko's platforms will tell program offices more than they have ever known about hull and weld condition across the surface fleet. They will not build a single additional dry dock, train a single additional structural welder, or add a single QAR billet to review the findings.
What the contract gets right, and what it does not fix
None of this is an argument against the Gecko contract. Better structural data is better than worse structural data, and a Navy that can see corrosion before it becomes a hull breach is a Navy that can plan repairs instead of reacting to casualties. The argument is narrower and sharper: a $300 million investment in detection is only as good as the institutional commitment to act on what it detects, and the Navy's own recent record, cutting inspections nearly in half in 2020 specifically to protect contractor relationships, is the strongest evidence available that detection and action are not the same muscle.² The robots will do their job. Whether the findings they generate translate into funded repair work, priced correctly, inside the availability window, depends entirely on decisions the contract itself does not make.
A challenge to NAVSEA program offices
NAVSEA program offices will receive the Gecko data. The question is what happens in the ninety days after a wall-climbing sensor flags a weld fatigue pattern on a hull nobody budgeted to repair this cycle. If the answer is a change order negotiated under fixed-price pressure, the Navy has bought a better diagnostic tool and kept the same broken treatment plan. If the answer is a pre-negotiated discovery-pricing mechanism, built into the contract before the robots start climbing, the $300 million buys what it was supposed to buy: ships that get fixed, not ships that get photographed. Program offices have the data now. The next decision is theirs, and it will show up in the on-time completion numbers within two maintenance cycles.
Better eyes do not repair a hull. Only a funded decision does.
References
- Defense News, "US Navy, Gecko Robotics surge AI maintenance contract to $300M," October 20, 2026.
- GAO-25-106749.
- GAO maintenance availability analysis (on-time completion rate below 40 percent despite available dock space).
- USNI News coverage of Navy shift from cost-plus to fixed-price, per-ship bidding.
- GAO analysis of Ticonderoga-class cruiser modernization ($1.84 billion spent on four ships decommissioned before deployment).
- Congressional Budget Office, ship repair industrial base analysis, December 2025.
