Defence: budget versus strategy

The night shift at Osborne ends with a walk past a building site. On the northern edge of Adelaide, the yard that will one day build Australia's nuclear-powered submarines is still largely earthworks, sheds, hardstand and cranes — a shipyard being built in order to build a shipyard. A boilermaker who came across from the car industry when the last Australian-made cars rolled off the line in 2017 has been told, accurately, that the work here will outlast her career. She has also read the schedule. The first submarine assembled in that yard is not expected to be handed to the Royal Australian Navy until the early 2040s. She will be in her sixties by then. Her apprentice is currently in primary school.

Three thousand kilometres north, an armament technician at a bare base in the Northern Territory is doing a different sum. Hers is about how many of a particular guided weapon sit in the magazine, how many the squadron would use in a week of high-tempo operations, and how long the international queue is to buy more. Both of these are defence questions. Both of them are budget questions. They are not, however, the same budget question, and the distance between them is the subject of this audit. Australia's official strategic assessment says the country faces its most demanding circumstances since the Second World War, and describes that pressure as present rather than distant. The defence budget is rising to meet it — but most of the increase arrives in the 2030s, and the single largest commitment inside it delivers its most visible output in the decade after that.

What the government did: a new strategy, a rebuilt investment program, and a submarine

In April 2023 the government released the public version of the Defence Strategic Review, an independent review it had commissioned shortly after taking office in 2022. The review's language was unusually direct for a Commonwealth strategic document. It described Australia's strategic circumstances as the most challenging since the Second World War. It found that the long-standing planning assumption of a ten-year warning time — the idea that Australia would see a major conflict coming a decade out and could expand the force in that window — no longer held. And it judged that the Australian Defence Force, as then constituted and equipped, was not fully fit for the purpose the new environment demanded (Defence Strategic Review, April 2023).

The review's prescription was a change of shape as much as of size. It recommended abandoning the "balanced force" model — a defence force built to do a bit of everything, moderately well — in favour of a "focused force" concentrated on a smaller number of things Australia most needs to be able to do: deny an adversary the ability to operate against Australia through the northern maritime approaches, strike at long range, sustain a force away from home, and hold enough munitions to do any of it for more than a few days. In the review's terms this was a shift toward "impactful projection", meaning the ability to affect events at a distance rather than only to defend the continent's coastline (Defence Strategic Review, April 2023).

Twelve months later, in April 2024, the government published its answer: the first National Defence Strategy, intended to be reissued every two years, and alongside it a rebuilt Integrated Investment Program — the ten-year list of what Defence intends to buy, with indicative cost ranges by capability domain. The Integrated Investment Program published in April 2024 carried a headline value of roughly $887 billion over the decade to 2033-34 (Integrated Investment Program, April 2024). That number is the acquisition and sustainment investment pipeline, not the annual defence budget; it is the sum of everything Defence plans to spend on capability over ten years, and it is best read as a plan rather than as appropriated money.

The annual figures sit underneath it. The defence budget for 2026-27 is A$62.6 billion, equal to 2.02 per cent of gross domestic product (Budget 2026-27). Total defence funding for the same year, which adds national intelligence and cyber elements sitting outside the Defence portfolio line, is $63.4 billion (Budget 2026-27). Across roughly 10.9 million Australian households (approx.; ABS household estimates), $62.6 billion works out to something in the order of $5,700 per household per year — a useful reminder that defence is not a rounding error in the federal budget but one of its four or five largest single commitments.

The forward path is where the argument lives. The government has committed to lifting defence spending to 3 per cent of GDP by 2033 (Government commitment, as stated in Budget 2026-27 material). In dollars, total defence funding is projected to rise from $63.4 billion in 2026-27 to $106.7 billion in 2033-34 and $112.1 billion in 2035-36 (Budget 2026-27; Integrated Investment Program). That is a rise of roughly $43 billion in annual funding across seven years, and roughly $49 billion across ten — an increase of about 77 per cent in nominal terms over the decade. By the standards of Australian defence budgeting since the 1950s, it is a very large planned expansion.

Inside it sits AUKUS Pillar One: the acquisition of a conventionally-armed, nuclear-powered submarine capability. The "optimal pathway" announced in March 2023 has three stages. From around 2027, United States and United Kingdom submarines are to operate on rotation from HMAS Stirling in Western Australia under an arrangement known as Submarine Rotational Force – West. From the early 2030s, Australia is to purchase three Virginia-class submarines from the United States, with an option for up to two more, subject to decisions made in Washington at the time. And from the early 2040s, Australia is to take delivery of the first SSN-AUKUS boats built at Osborne. The government's own published estimate of the whole-of-life cost of the program is $268 billion to $368 billion over thirty years, with roughly $71 billion to $96 billion of that falling before 2036 (Government estimate, as published from March 2023 and carried forward in subsequent budget material).

Two institutional changes accompanied it. The Australian Submarine Agency was established as a standalone entity to run the program, with a departmental budget of about $512.5 million in 2026-27, up roughly 33 per cent year on year (Budget 2026-27) — that figure is the agency's own running cost, not the program spend it administers. And Parliament passed the Australian Naval Nuclear Power Safety Act in 2024, creating a dedicated regulator for naval nuclear propulsion, because no existing Australian regulatory body had the remit or the expertise to license a reactor at sea.

The evidence: what the primary record shows about timing and composition

The steel-man is strong enough that the critique has to be made carefully, from the published numbers rather than from rhetoric. Two things have to be established. First, that the funding curve really is weighted toward the 2030s rather than the near term. Second, that the composition of near-term spending has a genuine opportunity cost inside the window the government's own assessment identifies as most demanding. Neither claim requires believing that defence spending is falling, or flat, or inadequate in aggregate. It is rising, materially, and this audit takes that as established fact.

The gap between what is appropriated and what is projected

Australian budgets commit money over four years. The forward estimates — the budget year plus the three that follow — are the horizon over which the Commonwealth publishes numbers that carry real institutional weight, because they feed appropriations, contracts and departmental planning. Anything beyond the forward estimates is a medium-term projection. Projections are useful, honest and necessary. They are not money. They can be revised at any subsequent budget by any subsequent government, and historically they frequently are.

Apply that distinction to the 3-per-cent-by-2033 commitment. Defence is at 2.02 per cent of GDP in 2026-27 (Budget 2026-27). The target year, 2033, sits three years beyond the end of the current forward estimates. This means the great majority of the climb from 2 per cent to 3 per cent occurs in years for which no Parliament has yet appropriated anything. The dollar figures make the same point: total defence funding rises from $63.4 billion in 2026-27 to $106.7 billion in 2033-34 (Budget 2026-27; Integrated Investment Program), but the years in which the annual increment is largest are the later ones, when GDP is larger and the share is climbing at the same time.

This is not a hidden fact or an accounting trick. The government has published the trajectory openly, which is to its credit. But it does mean that the commitment currently rests, for most of its value, on decisions that future budgets will have to confirm — including budgets handed down after at least two more federal elections. That is a statement about institutional structure, not about anyone's sincerity.

What AUKUS actually consumes, and what it does not

Here the honest number is smaller than the rhetoric on either side suggests, and it is worth stating plainly because it disciplines the argument. The government's published estimate puts AUKUS Pillar One spending to 2036 at roughly $71 billion to $96 billion (Government estimate, from March 2023). Set that against an Integrated Investment Program of roughly $887 billion over the decade to 2033-34 (Integrated Investment Program, April 2024). On those two published figures, the submarine program accounts for something in the order of 8 to 11 per cent of the ten-year investment pipeline (approx., derived from the two published figures above). It is the largest single line in the program. It is not most of the program, and anyone who says it is has not read the documents.

But the dollar share understates the constraint, for a reason that matters more than the percentage. Most defence dollars compete for inputs that are broadly substitutable — general engineering capacity, project management, steel, construction trades. AUKUS competes for inputs that are not. There is a small and slowly-growing national pool of people qualified to work to nuclear standards: reactor engineers, nuclear-qualified welders and inspectors, radiation safety specialists, submarine designers, and the senior Defence and central-agency officials capable of running a program of this complexity. There is one Osborne. There is one deep-maintenance facility programme at Henderson in Western Australia. There is one new naval nuclear safety regulator, standing itself up from scratch under the 2024 Act.

Those are non-fungible inputs. When a program of this scale claims them, the cost to other programs is not measured in the dollars it takes but in the senior attention, the specialised workforce and the infrastructure sequencing it takes with them. That is the accurate version of the crowding-out argument, and it survives the observation that AUKUS is only about a tenth of the investment pipeline by value.

AUKUS takes about a tenth of the investment program by dollars. It takes considerably more than a tenth of the things that cannot be bought with dollars.

The capabilities the review said were urgent

The Defence Strategic Review was specific about what needed to happen quickly, and the Integrated Investment Program funded those areas. Guided weapons and explosive ordnance — the munitions the ADF would actually expend, and the domestic capacity to manufacture some of them — were allocated a sum in the order of $16 billion to $21 billion across the decade (approx.; Integrated Investment Program, April 2024). The northern base network — the airfields, fuel storage, accommodation, hardened facilities and port infrastructure across northern Australia from which any operation in the northern approaches would be mounted — was allocated a sum in the order of $14 billion to $18 billion across the same decade (approx.; Integrated Investment Program, April 2024).

Those are real commitments and they are new. The question the primary record raises is one of sequencing rather than of existence. A ten-year allocation is not a near-term allocation. Munitions stockpiles are a case in point: a stockpile is only strategically useful when it is already in the magazine, because the international queue for precision munitions lengthens precisely when demand rises. Domestic manufacturing capacity, which is the durable answer, takes years to certify and qualify — the Guided Weapons and Explosive Ordnance Enterprise was established with exactly that logic, and the first domestically-produced rounds of any complex weapon are the slowest to arrive. Base hardening in the north has a similar profile: the concrete and the fuel farms take years, and until they exist, aircraft and fuel are concentrated where they are today.

Sustainment, availability, and the fleet that already exists

There is a further category the aggregate numbers obscure. Capability delivered on paper is not capability available at sea or in the air. Sustainment — maintenance, spares, deep servicing, crewing — determines what proportion of the existing force can actually be used at any moment. The Collins-class submarines, Australia's current undersea capability, must be sustained through a life-of-type extension in order to bridge the gap to the nuclear-powered boats; that extension is itself a substantial program, and its schedule and cost have been the subject of repeated public scrutiny (Australian National Audit Office and Senate Estimates evidence, recent years). The surface fleet was itself reshaped after a separate review of the surface combatant fleet reported in early 2024, which reduced the planned Hunter-class frigate build and added a class of general-purpose frigates to be acquired more quickly (Government response to the independent analysis of Navy's surface combatant fleet, 2024).

The pattern to notice is that the near-term force is a bridging force. Between now and the early 2030s, Australia's deployable capability is largely the force that already exists, sustained harder and for longer than originally planned, while the new capability is under construction. Bridging forces are normal and often sensible. But they carry a specific risk: the older a platform gets, the more of its life is spent in maintenance rather than at sea, and availability tends to fall exactly when the schedule needs it to hold.

The delivery risk that sits outside Australia's control

One element of the pathway depends on decisions and industrial performance in another country, and the primary record is explicit about it. The planned purchase of Virginia-class submarines from the early 2030s is subject to a certification by the United States President at the time that the transfer will not degrade United States undersea capability — a condition written into United States legislation enacted in 2023. United States submarine construction rates have run below the build rate that programme requires, a fact documented in United States Navy budget material and analysed in Congressional Research Service reporting (SECONDARY: Congressional Research Service, Virginia-class submarine procurement, recent editions; United States Navy budget documents). Australia has also agreed to contribute to the United States submarine industrial base, with a contribution in the order of US$3 billion (approx.) and payments commencing from 2025 (approx.; Australian Government and Australian Submarine Agency public statements).

This is not a claim that the boats will not arrive. It is a statement of conditionality that the government itself has never concealed. A capability plan whose middle stage depends on a foreign legislative certification and a foreign production rate carries a different risk profile from one that does not, and it is reasonable for the public to see that risk described accurately rather than assumed away.

The workforce constraint, measured honestly

Money is not the only input in short supply, and on some capabilities it is not the scarcest. The permanent ADF has been running some thousands of positions below its funded strength in recent years, with annual separation rates in the order of 9 to 10 per cent (approx.; Department of Defence Annual Report, recent editions). The stated ambition is a workforce of around 80,000 by 2040 (Defence Strategic Review, 2023; National Defence Strategy, 2024). Meanwhile, the shipbuilding and sustainment workforce needed for the submarine enterprise alone runs to many thousands of new skilled roles over the coming two decades, competing for the same trades that the housing, energy transition and resources sectors are also trying to hire.

This cuts both ways in the analysis, and the fair reading has to acknowledge both directions. It supports the government's case that spending cannot simply be accelerated. It also sharpens the critique, because a workforce constraint is a reason to start the slow things earlier, not a reason to accept a later delivery date without comment.

The critique: an assessment written in years, funded on a curve written in decades

The central charge is not that the government is neglecting defence. On the published record, that claim is false, and it should be abandoned by anyone making it. The charge is narrower and harder to answer: the government's own strategic assessment describes urgency in the present tense, and the funding response is shaped to arrive later.

Consider the two documents side by side. The Defence Strategic Review, in April 2023, said the ten-year warning time was gone — meaning Australia can no longer assume a decade of notice in which to expand. That is a statement about the near term by construction. The funding response commits to 3 per cent of GDP by 2033, ten years after the review, with the largest annual increments falling in the years after the current forward estimates. The single biggest capability outcome, the Australian-built submarine, arrives in the 2040s, roughly two decades after the assessment that justified it.

Put plainly: the review removed the assumption of ten years' warning, and the budget was then built on a ten-year ramp. Those two things can both be defensible individually. Together they leave a period — broadly the late 2020s to the early 2030s — in which the assessment says the risk is highest and the capability response is least complete. That window is the whole of the critique.

The review said Australia could no longer count on ten years' warning. The funding curve then took ten years to arrive.

The composition question compounds it. Within that window, the resources most constrained — nuclear-qualified people, senior program leadership, specialised infrastructure at Osborne and Henderson, and the regulatory apparatus standing itself up — are heavily committed to a capability whose first Australian-built output falls outside the window entirely. The items that could most plausibly change what Australia can do inside the window are the less glamorous ones: filled munitions magazines, long-range strike weapons in service rather than on order, hardened and dispersed northern basing, fuel and logistics resilience, and enough sustainment funding to keep the existing fleet available. Each of those is funded in the Integrated Investment Program. None of them is funded on a profile that front-loads the window.

There is a second-order problem with a back-loaded curve, familiar from every other area of Commonwealth spending: projections beyond the forward estimates are the most fragile part of any budget. Between now and 2033 the Commonwealth faces a structural deficit, a rapidly growing public debt interest bill, and rising NDIS, aged care and health costs — all of them examined in separate audits on this site. Each of those pressures competes for the same fiscal space that a rise from 2 per cent to 3 per cent of GDP will require. A defence commitment whose value sits mostly in the out-years is a commitment that must survive contact with each of those pressures at every budget between now and then. It may well survive. But the risk that it is trimmed, re-phased or redefined is not zero, and it grows the further out the money sits.

Third, there is a transparency point that follows from the structure rather than from any concealment. Because the Integrated Investment Program publishes cost ranges by domain over a decade, and because the submarine estimate itself spans $268 billion to $368 billion — a range of $100 billion, wider than a year's entire defence budget — it is genuinely difficult for Parliament or the public to test whether a given near-term capability is being delayed to fund a long-term one. The information published is more than previous governments published. It is still not enough to answer the trade-off question directly, and the answer to that question is what the public most needs.

Finally, a point about what "3 per cent of GDP" actually measures. A share-of-GDP target is a measure of input, not of output. It says how much the country spends; it says nothing about whether the spending purchased usable capability, arrived on schedule, or delivered the specific things the strategic assessment said mattered. Australia can hit 3 per cent in 2033 and still have fewer available submarines, thinner munitions stocks and less resilient northern basing than the review contemplated, if the money went to programs that ran late. The ANAO's major projects work exists precisely because that outcome has occurred before. Judging defence policy by the share-of-GDP number alone would repeat the error this site makes about the budget generally: mistaking a headline figure for the structural reality underneath it.

What the government got right

Three things deserve explicit credit, and they are not small.

First, the 3-per-cent commitment is real and substantial. It is easy to be cynical about targets set for a year beyond the forward estimates, and this article has been. But an increase from roughly 2 per cent to 3 per cent of GDP is a structural expansion of about half again on the current base, undertaken by a government simultaneously managing a structural deficit and the fastest-growing interest bill in Commonwealth history. Very few comparable economies have made a commitment of that magnitude and then published the dollar trajectory to go with it. It would have been politically easier to announce a smaller number, or a vaguer one, or none at all.

Second, the Integrated Investment Program is a better planning document than what came before it. Its predecessor documents tended to list aspirations without a credible total, which allowed programs to be announced without anything being displaced. The 2024 program publishes indicative cost ranges by domain against a stated envelope. That makes it possible — for the first time in a long time — for outside analysts, the Parliament and this website to check whether the sum of the announcements exceeds the sum of the money. That is a real improvement in public accountability, and it is the reason much of the critique above can be written at all.

Third, and most creditably, the government made hard prioritisation decisions rather than pretending everything was affordable. Reshaping the investment program meant cancelling, reducing or deferring programs that had constituencies: the planned Hunter-class frigate build was reduced, some land programs were cut back, and other projects were deferred outside the decade. Each of those decisions cost the government something politically and cost specific communities and firms something real. Australian defence policy has a long history of adding capabilities to a plan without removing any, and then discovering a decade later that the plan was never funded. This government broke that habit. It should be said plainly, because it is the opposite of what governments usually do.

A caveat in the other direction is also owed. The submarine schedule Australia is now working to was shaped by a decision taken before this government took office. The Attack-class program with Naval Group was signed and then cancelled under a previous government, consuming billions of dollars and around half a decade of design work that produced no boat (Australian National Audit Office and Senate Estimates evidence, 2021–2023, approx.). The years lost there are unrecoverable, and the current delivery date carries them. Separating what was caused from what was inherited matters here more than almost anywhere else in this audit: the 2040s delivery date for an Australian-built boat is not principally a failure of the current government's budgeting. It is the compounding of a decision made in 2016, reversed in 2021, and replaced in 2023. What the current government owns is not the start date. It is the choice of what to do about the years between now and then.

It is also fair to record that recruitment and retention, which is genuinely binding on several capabilities, is not a problem any government solves quickly, and that the measures taken so far — retention bonuses, changes to eligibility and pay arrangements — are the kind of measures whose effects show up over years, not budget cycles. Criticising a government for not having fixed a workforce problem inside one term would be unreasonable. Asking whether the near-term force posture reflects the reality of that constraint is not.