The AUKUS story that ran last week was a communications story. The Minister for Defence Industry described the Submarine Rotational Force-West arrangement in terms that sounded time-limited. Washington asked what he meant, and within days he was confirming that United States submarines would be an enduring element of American force posture in Australia (Bourke, 2026). An analysis the following day traced how contradictory statements about the timeline for the base at HMAS Stirling had turned a piece of infrastructure into a political sensitivity (Greene, 2026). The reporting was reasonable and the questions it put to those statements were fair.
Two days before all of that, a paper appeared in Defence and Security Analysis proposing a way to ask a harder question, and it went unremarked. It contends that defence procurement failures of the Ajax kind stem less from cost and schedule overruns than from a structural gap between the risk an organisation retains and the risk it can actually govern (Risk and retention, 2026). The paper draws its lesson from Heathrow Terminal 5 and proposes a five-capability diagnostic that the United Kingdom’s Defence Equipment and Support could use to test retained risk before signing and through delivery.
The distinction between retained risk and governable risk repays a moment of attention, because it is easy to nod at and hard to apply. When a client body signs a major contract it decides which risks the supplier carries and which it keeps for itself. Keeping risk is the easy half of that transaction, and it happens constantly, often because no supplier would price the risk at all. Governing what you keep is the other half, and it demands capability: the ability to see the risk while it is still moving and to act on it before the options close. Organisations routinely retain risk they have no machinery to govern. The gap never appears on a risk register, because a register records what you hold and stays silent on whether you can do anything about it.
Apply that test to AUKUS Pillar One and the position becomes much clearer. ASPI published a piece on 30 July arguing that Pillar One will be won or lost on workforce. Australia’s ability to grow and sustain a specialised defence workforce should be treated as a first-order strategic priority, it argues, instead of a secondary implementation concern (ASPI, 2026a). That is a claim about retained risk. No contractor can make Australia produce nuclear technicians. The workforce risk sits permanently with the Commonwealth, and it cannot be transferred at any price.
The scale of what has been retained is worth stating plainly. Peter Briggs, writing on 1 August, sets out Australian Submarine Agency projections of roughly 20,000 direct jobs over the next thirty years, with a peak industrial workforce of up to 8,500 highly specialised trades, engineers and nuclear technicians (Briggs, 2026). A workforce of that size has to be drawn from a national labour market that Jobs and Skills Australia already describes as carrying severe trade shortages. The submarine enterprise is not competing for this labour against other defence programs alone. It is competing against every housing target and every hospital build in the country.
Briggs sets out a structural condition that splits the recruitment task in two. Osborne is the sole domestic nuclear construction yard. Henderson handles depot maintenance and operational support for visiting submarines. The two sites sit 2,700 kilometres apart, which rules out flexible movement of labour between them (Briggs, 2026). A shortfall at one yard cannot be relieved by a surplus at the other, so each site needs its own training pipeline and its own regional labour market to draw on. Dual-site strategies are usually sold on resilience. This one buys resilience by doubling the recruitment problem.
Now run the governability test. Australia has retained the workforce risk because it had no choice. Can it govern what it has kept? Governing this particular risk means moving levers inside the education and training systems and the migration settings, plus the housing conditions that decide whether a welder moves to Osborne and stays. Those levers are held by different ministers in different departments, and none of them answers to the submarine program. The Commonwealth holds the risk. The capability to govern it is distributed across a dozen bodies with their own priorities.
The diffusion problem will be familiar to anyone who read ASPI’s other piece from the same day, on legacy technology as a national security threat. Its argument is that fragmented ownership stretching from operators through to ministers diffuses accountability until the necessary upgrade decisions become impossible to make, and that the fix is clear mandatory decision triggers (ASPI, 2026b). The failure mode is identical. Diffuse ownership of a risk nobody can individually resolve produces drift, and drift on a thirty-year workforce pipeline is not recoverable in the way a schedule slip is.
There is a related category error that makes the problem harder to see, and a Dead Reckoning analysis published on 30 July names it precisely. Australia’s planned Kongsberg missile manufacturing facility will expand supply chain options for domestic needs and for Kongsberg’s global customers, and that expansion does not automatically translate into sovereign capability. Industrial presence and sovereign control are different things (Dead Reckoning, 2026). Counting jobs and calling the total capability is the same error wearing different clothes. A jobs projection is an input forecast, and it becomes a capability statement only once somebody demonstrates that the people can be found and then kept through a thirty-year program.
This is the pattern I traced through the Pacific in Seven Thousand Kilometres Past the Press Conference: an announcement-led architecture where the signing is treated as the achievement and the implementation depth goes untested. It also connects to the argument in Months to Destroy, Years to Rebuild, because specialised capability takes years to build and can be lost far faster than that. A nuclear technician pipeline interrupted in year eight does not restart in year nine.
Four things would demonstrate that the retained risk is being governed rather than merely held. Name the person accountable for the AUKUS workforce outcome who has authority reaching across training and migration levers together, and say what happens when those levers conflict. Publish the retained-risk position: which risks the Commonwealth has kept, and what capability it holds to govern each one. Test the Osborne and Henderson workforce plans as two independent problems, because the labour market treats them that way. And stop presenting jobs projections as evidence of capability, because they are evidence of demand.
The political question about AUKUS is whether the United States will honour the arrangement, and it consumed most of last week. That question is real, and it sits substantially outside Australia’s control. The deliverability question is inside our control, and it asks whether we can staff two yards on opposite sides of the continent from a labour market that is already short. Only one of them can be answered by anything Australia decides to do.
References
ASPI. (2026a, July 30). AUKUS Pillar One will be won or lost on workforce. The Strategist. https://aspistrategist.org.au/aukus-pillar-one-will-be-won-or-lost-on-workforce
ASPI. (2026b, July 30). Legacy technology is a national security threat hiding in plain sight. The Strategist. https://aspistrategist.org.au/legacy-technology-is-a-national-security-threat-hiding-in-plain-sight
Bourke, L. (2026, July 30). Pat Conroy scrambles to clarify AUKUS submarine plan blunder after US Navy ‘eviction notice’. The Nightly. https://thenightly.com.au/politics/latika-m-bourke-pat-conroy-scrambles-to-clarify-aukus-submarine-plan-blunder-after-us-navy-eviction-notice-c-22654897
Briggs, P. (2026, August 1). AUKUS infrastructure, part 4: Australia’s dual-site labour trap. Pearls and Irritations. https://johnmenadue.com/post/2026/08/aukus-infrastructure-part-4-australias-dual-site-labour-trap/
Dead Reckoning. (2026, July 30). Supply Chain Anatomy: Kongsberg NSM/JSM.https://geomastery.substack.com/p/supply-chain-anatomy-kongsberg-nsmjsm
Greene, A. (2026, July 31). Deadline conjecture over HMAS Stirling highlights political sensitivity of AUKUS. The Nightly. https://thenightly.com.au/politics/andrew-greeene-analysis-deadline-conjecture-over-hmas-stirling-highlights-politically-sensitivity-of-aukus-c-22656330
Risk and retention: learning from Heathrow Terminal 5 for defence procurement. (2026). Defence and Security Analysis. https://doi.org/10.1080/14751798.2026.2705015
Russell Buzby 
Leave a comment