Who keeps it going when no one is onboard?
Support can be thought of in two layers: the maintainers onboard, and the depth support behind them. A software-defined approach, and the Hybrid Navy, inevitably change the proposition for both, but nobody has yet said how.
The US Navy's production agreements for 30 medium uncrewed surface vessels (the award that prompted this series) are precise about range, speed, payload, price and the absence of a crew. However, they are silent on support. To my knowledge, none of the coverage says how these vessels will be sustained.[1] The reasonable inference is that the US Navy will continue its established model, with naval personnel maintaining and depth support through the shipyards. What everything is silent about, though, is who maintains the vessel at sea? Who delivers the preventive maintenance, and who fixes things in the middle of an engagement to keep the platform in the fight?
The same is true in the UK, with the Strategic Defence Review naming a 'New Hybrid Navy' and committing to 'making sure through-life support is considered from the outset'. The First Sea Lord's principle is 'uncrewed wherever possible; crewed only where necessary', and he has called the shift 'the biggest change of the culture, workforce training and skills for our sailors and marines'. However, nothing currently describes who keeps the uncrewed part of that fleet going.[2]
My last article ended with the thought that the real value of a software-defined shipyard is not the speed of the build but what the people who keep the ship going know about it afterwards. This article begins to look at what that might mean, and what happens when there is nobody on board to do the maintenance.
Two layers
Support is not a singular action or activity, and it helps to consider the different dimensions of support before deciding what autonomy might do to it. The MODs own vocabulary divides maintenance into Forward support (the tasks best done in the operating environment), and Depth support (the tasks that are not), and grades the work itself from Level 1 (servicing and repair by replacement), through to Level 4 (full reconditioning and major repair).[3] Essentially there are two layers.
The first is the ship's company: the women and men who sail with the ship and keep it running, and who also provide damage control, crew its boarding parties, provide its security alongside and support defence diplomacy in a foreign port.
The second is everything behind them: industry augmenting ship's staff alongside during fleet time, the dockyard and the equipment manufacturers taking the deeper work, and the planning system that decides what gets done when.
My argument is that software-defined capability, and the Hybrid Navy that comes with it, change the proposition for both layers at once, and that the benefits and the consequences arrive in roughly equal measure.
Layer one: the maintainer at the heart
A Type 23 sails with a ship's company of about 185; a Type 31 is designed for about 110, and that figure was arrived at by analysing engineering workload section by section and choosing equipment for its maintenance burden, with automation and a network of cameras and sensors allowing the machinery to be monitored remotely.[4] Babcock's own account of that design work, in a paper to the Royal Institution of Naval Architects in 2024, says the crew was sized not by the maintenance burden but by the resources needed to deal with a major fire or flood, which 'typically requires a large number of people to contain and manage', and 'there has not yet been a robot demonstrated that can replicate the full flexible and adaptable capability' of a trained sailor. The same paper notes that corrective maintenance 'will inevitably grow as a burden over time as systems become worn', and that keeping a complex uncrewed platform functioning at sea on an enduring operation is a problem nobody has yet solved.[5]
The current philosophy is that the maintainer sits at the heart of the support solution. The engineering branch was rebuilt under Project Faraday from the mid-2010s, and the Navy's own description of that work was that it 'brought back' hands-on and craft training so that technicians are 'taught to operate, maintain, diagnose and repair' the ship's systems.[6] The trend before Faraday was an increasing reliance on the equipment manufacturer to do the diagnosis and the repair, and a maintainer left enabled to do only the first-level support. The National Audit Office found the Navy's weapons engineers to be the single largest engineering shortfall across Defence, and judged the Navy's position 'unsustainable' by the Navy's own assessment.[7] Skills erosion becomes a retention problem before it is felt as an engineering one. The Ministry no longer publishes strength or outflow by branch,[8] but the engineering trades remain, and by every independent account, remains as the most acute shortage within the Royal Navy. If a sustainment model continues to follow a pattern that increasingly replaces a whole unit and returns it to the manufacturer, rather than repair it on board, then the short-term efficiency is being traded for a detrimental long-term impact on the whole employment model for the Navy.[9]
I saw this at first hand during my time working on the development of the Common Support Model. The drive to place the maintainer at the heart of the support solution was littered with stories about maintainers working exhaustively to understand or diagnose a fault, only for the OEM to turn up with a laptop and fix the defect with a couple of keystrokes.[10] The faults were not difficult to fix, but the support solutions were disintermediating the maintainer at the expense of the overall employment proposition for the Navy personnel. The response through the Common Support Model was to give maintainers the training and the authority to diagnose and repair again. It is worth remembering those lessons before we design a fleet in which the maintainer is no longer there.
Layer one without people
Take the person off the ship and three things go with them.
The secondary and tertiary roles go, and so does the human judgement in the loop that the Type 31 paper has in mind when it allows that a minimal crew may be retained 'for other purposes such as weapon control' to satisfy the rules of engagement.[5]
The career goes as the Hybrid Navy and autonomy will likely demand a different branch structure, certainly one that needs different skills and ingenuity, because the branch as it stands is built around a maintainer at sea.
The intervention goes as it is the on-board maintainer who notices that a pump is running hot, or re-seats a loose connector, or finds a way of getting a system back on line when the pressure is on.
The US Navy's own experience is an example of the third. Sea Hunter's transit from San Diego to Hawaii in 2019 required maintainers to board her to repair 'several broken systems', and the programme manager's conclusion was that the support infrastructure 'is optimized around manned warships'.[11] IA commonly held estimate is that around a third of a destroyer's crew exists to keep the ship running, which gives a sense of the scale of what is no longer there, and of the consequence of its absence.
One of the clearest statements of what the machinery itself must now do came from the US Congress, which mandated a demonstration of 720 continuous hours of uncrewed power and propulsion, with no human intervention and no corrective or preventive maintenance, before the acquisition of the larger vessels could proceed. The sixth and final engine passed in December 2024, in a land test cell, after three years of trying.[12] That is a 30-day bench test of an engine though, so worth keeping that in perspective. It has since become a certification, as in September 2026 the US Navy cleared Rolls-Royce's mtu Series 4000 for autonomous operation after two 720-hour tests, and the smaller Series 2000 on component testing and field hours, so that '30-days without maintenance' is now something an engine can be approved for rather than something a programme has to prove.[13]
The best at-sea record currently appears to be the Pacific deployment of four prototype vessels over five months and 46,651 combined nautical miles. During this trial, two of them had no humans on board at all, other than when entering and leaving port. Although impressive, nothing was published about maintenance, defects, reliability or availability.[14] The US Government Accountability Office found that the Navy's plans understated 'operations and sustainment and the supporting digital infrastructure', which a senior official said could run to billions. The US is now rebuilding how through-life costs are estimated for autonomous vessels, because the current reference class, built on crewed ships, is insufficient.[15] One serving officer has argued that the labour does not disappear but moves, from the people who operate the platform to the people who maintain it, and from the ship to the shore, offering the rough comparison that Waymo employs about 1,700 engineers and technicians for about 1,500 cars.[16] It is a useful corrective to the assumption that an uncrewed fleet is cheaper to maintain.
Layer two: depth support, done better
The second layer has been moving ashore for some time. Industry currently holds responsibility for complex warship upkeep and regeneration before handing the ship back to her crew, relieving pressure on crew availability and prioritising naval engineers for other operational roles.[17] The £5.2 billion Future Maritime Support Programme was let in 2021 with the aim of 'improved availability, resilience and certainty of delivery'.[18] It would be fair to say that the system it was applied to was already under strain. The Defence Committee was unconvinced in 2021 that escort availability could be lifted from 60 to 80 per cent and recorded the First Sea Lord's evidence that the crewing models which had improved availability elsewhere were not feasible for the Type 45.[19] Through 2024 two or three of the six destroyers were available at any time, and Type 23 refits that took 1-3 years a decade ago were taking 3-4 at the peak of the life extension.[20]
This is where software-defined support has the most obvious benefit. Connecting the installed performance of a platform to prognostic models, so that the near-term availability and the long-term resilience of the ship are planned from what the machinery is actually doing rather than from what a maintenance schedule assumes. This is one of the natural steps in a long journey towards doing the right maintenance at the right time. Lived experience says there is still a great deal of maintenance in the system that is done because the plan says so, rather than because the equipment needs it, and condition-based planning could help to take it out. One caution comes from the United States, who have also been on that journey, who report that condition-based maintenance 'went too far' by 2017. The approach saw deferral of structural work that then showed-up in the shipyard as unplanned growth. It did note that the useful application was seen in the wider systems and equipment: gearboxes, shafting, turbines and generators.[21] Alongside the challenge of secure transmission and constrained bandwidth, the UK MOD's Defence Support Strategy observes that 'our data is often in the wrong format or incomplete', and that Defence must be more assertive over data architectures 'especially data held by third parties'.[22]
All of that brings the two layers back together, crewed or uncrewed, software-defined support provides an opportunity to capture data systematically, but the knowledge-based inference and interpretation is gone. Therefore the key test for the support data comes down to defining what is essential, rather than interesting. The installed performance data that changes a maintenance decision matters - the rest is just data storage cost.
Where it works, the manufacturer runs both layers
Where uncrewed vessels operate at scale today, the equipment manufacturer doesn't just maintain them, it also tends to operate them. Australia's Bluebottle vessels average 75 days on task and are run around the clock by their builder from a facility on the naval base, and the Royal Australian Navy has just ordered forty more.[23] The Ghost Shark contract bundles 'delivery, maintenance and continued development' into one AUD$1.7 billion award.[24] The US Fifth Fleet's Task Force 59 ran its vessels contractor-owned and contractor-operated under naval oversight (COCONO), and the then Secretary of the Navy said unarmed platforms could stay that way for good, because it allowed faster refresh of the technology.[25]
At the far end of the spectrum Ukraine's one-way naval drones, at roughly USD$250k to USD$275k each and up to 50 a month, have no support model at all beyond building the next one.[26] These are 20-tonne boats and smaller. And of course, the scale matters - obviously nobody is proposing a one-way drone's support model for a Type 26, but the direction is unambiguous. For a small uncrewed vessel the natural answer is that the OEM provides the capability through a Platform-as-a-Service (PaaS) model, effectively collapsing both layers of support into one supplier.
Benefit and consequence
Both maintenance layers change the employment model for the Royal Navy and for industry, needing a different set of skills for strategic workforce planning. For the first layer, software-defined support could put the maintainer back at the heart of the solution, better equipped and better informed, working on the vessel alongside with skills that look more like a systems engineer's than a fitter's. The 'cost' is the person at sea, the roles they carried, and a career structure that will have to be rebuilt around a different kind of engineer.
For the second layer, the benefit is the right maintenance at the right time and a planning system that runs on evidence, supported by systemised knowledge and better access to assured information and data. The consequence is less obvious, as support for a complex vessel is already shared between an owner-operator, a maintenance provider and the equipment manufacturers, and each needs something different from the data. The Navy needs availability and the evidence to plan operations, the maintainer needs the diagnosis, the OEM needs the performance of its own equipment and has commercial reasons to keep it close. A data architecture that serves all three could disaggregate support from the current primes, which is not necessarily a bad thing, but that remains a contested space, and the Ministry's own strategy says that today the data too often sits with a third party.
If the manufacturer keeps the small boat going, operating it from the shore and holding the data that says how it is performing, then the question for a complex vessel is who does the same, and on whose terms? The small-craft answer scaled up is a new kind of prime with support locked-in, landing on a support landscape that is already fragmented across dozens of contracts and hundreds of suppliers. Whether that is a problem or simply the next model depends on what the Navy decides it must own for itself: the data, the skills, or the contract.
