The number that matters is not the flow
A flow cut leaves the standing population untouched on day one. The 1,825,403 people already here on temporary visas are still here, still bidding for the same rentals, still in the same queue for a GP, still competing for the same entry-level roles. A government that plans to the flow and not to the stock is planning to the wrong figure, and it will be surprised by the results.
The breakdown of the 1,825,403 is worth reading in full, because it is not a homogeneous group and the policy argument depends on the composition. Students: 564,029. Bridging visa holders: 440,571. Temporary graduates: 272,613. Temporary residents on skilled employment visas: 269,591. Working holiday makers: 211,390. Temporary residents on other employment visas: 59,636. Other temporary: 5,986. Temporary protection: 1,587.
The skilled employment and temporary graduate cohorts together come to 542,204 people, all of them with full, uncapped work rights. A 19-year-old Australian applying for their first full-time job is in the same queue as a 27-year-old graduate visa holder with a completed degree and two years of local experience. The employer is not choosing between a local and a foreigner. They are choosing between someone with experience and someone without it, and the visa settings are what put the first person in the room.
Source: mindiam figures: temporary visa population, drawn from Home Affairs BP0019 at 31 August 2026.
The AI mechanism: automation is taking the runway
In 2023, 636,450 foreign nationals arrived in Australia under the ABS 12-in-16-month residency rule. Of those, 282,570, or 44.4%, arrived on study visas. (The gross NOM arrival figure of 737,170 includes 100,720 returning Australian and New Zealand citizens; the foreign-national figure is the relevant one for this argument. Departures of 219,080 produced a net of 518,090 for the year.)
The study pathway feeds directly into the graduate visa, which feeds into the skilled visa. The occupations at the end of that pipeline are business administration, accounting, information technology and knowledge work at the entry level. These are precisely the occupations that large-language-model deployment absorbs first.
This is not a forecast dressed as a fact. It is the documented pattern of LLM adoption: the first roles to be restructured are the ones built around processing, drafting, summarising and routing information, which are the roles that a business or accounting degree prepares someone for. The runway on those roles is not infinite, and the intake is concentrated in exactly the cohort with the shortest remaining runway.
So the argument for cutting the student intake is not an argument about students. It is an argument about automation. Importing people for a five-year pipeline that terminates in roles a language model will handle is not a solution to a labour shortage. It is a structural cost that compounds, and it lands on the people already here in the form of competition for housing, services and the entry-level rung.
Source for the arrival composition: ABS NOM by visa.
The composition of the intake, and why it matters
Students made up 51.7% of net migration in 2023, against 14.1% for all skilled categories combined and 5.8% for permanent skill alone. The 2023 total of 518,090 was a post-border-reopening catch-up against 203,590 in 2022, so the total is not the point. The composition holds across 2019, 2022 and 2023, and it is the composition that drives the policy argument.
A migration programme that is 51.7% students and 5.8% permanent skilled workers is not primarily a programme for filling documented skill shortages. It is primarily a programme for filling universities and vocational colleges, and the Nixon Review, the Commonwealth's own rapid review into exploitation of the visa system, found that the pathway from study visa to work was being used by providers and agents in ways that had nothing to do with study. The review's findings are about the system and about named providers. They are not a claim about students as people, and this piece does not make one.
Source: ABS NOM by visa; Nixon Review.
The first job, and who is in the queue for it
Youth unemployment in August 2026 rose 0.4 percentage points to 10.8%, against a national rate of 4.6%. In the same month, employment rose by 39,500 overall, but full-time employment fell by 6,300 while part-time rose by 45,800. The jobs being added are not the jobs a 19-year-old builds a life on.
Youth unemployment running at more than double the national rate is not a rounding difference. It is a structural gap between the cohort with the least bargaining power and the rest of the labour market. And it is happening at the same time that 542,204 people on skilled and graduate visas hold full, uncapped work rights in the same occupational categories.
The causal claim needs to be stated carefully, because the ABS does not publish a decomposition attributing youth unemployment to any single factor. What the figures show is that the two conditions exist together: a high standing population of full-rights temporary workers in entry-level-adjacent roles, and a youth unemployment rate more than twice the national average, in a month when full-time work contracted. That is a correlation worth naming, and the policy argument follows from the mechanism rather than from the correlation alone.
The mechanism is competition for the first rung. When the rung is crowded, the person with no experience loses, and the person with no experience is always the youngest person in the queue.
What the standing population does to rents
1,825,403 people are bidding for the same rentals as you. They are not a future projection. They are here now, and they arrived faster than the housing stock did.
The ABS living cost indexes, which include mortgage interest charges and therefore move differently from the CPI, show what this has done to households at the bottom of the income distribution. From the March quarter 2012 to the June quarter 2026, living costs rose 51.7% for other government transfer recipient households, 50.0% for pensioner and beneficiary households, and 48.5% for age pensioner households. These are index changes for household types, not individual experiences, but the direction is unambiguous and the period covers the years of the largest intakes on record.
The argument is not that immigration single-handedly caused these increases. Population growth is one input among several. The argument is that a growth model built on adding people rather than raising output per person produces this result, and the ABS national accounts confirm the model. GDP rose while GDP per capita fell in seven of the last twelve quarters, and in 25 of 104 quarters since 2000. Headline GDP went up. Output per person went backwards. The Treasurer's press release reported growth. The person paying rent experienced something else.
Sources: ABS living cost indexes; ABS national accounts.
The compounding effect the headline number hides
The headline net migration figure understates the standing population effect of a year's intake, and it does so by design rather than by accident.
A permanent residency grant is not a one-off. It creates a future entitlement to sponsor family members. The migration programme runs a separate Family stream with its own planning level, published by Home Affairs. Student visas carry dependants. So the standing population effect of a year's intake exceeds that year's headline number, and a government that plans only to the headline is planning to the wrong figure.
This piece does not put a specific multiplier on that gap, because an invented multiple is the easiest thing for an opponent to disprove, and it would discredit the true mechanism underneath it. What can be said plainly is that the real number is higher than the headline, the Family stream planning level is published, and the compounding works in one direction only: upward.
Sources: Home Affairs migration program statistics; mindiam figures: net overseas migration.
The dependency ratio, and the honest answer to it
The standard defence of high migration is that Australia needs workers to support an ageing population. It is a real concern and it deserves a real answer rather than a dismissal.
The Productivity Commission gave one. Its inquiry into migrant intake into Australia found that migration "does not offer a long-term panacea" for ageing, because "immigrants age too." Bringing in a 25-year-old today defers the dependency ratio problem by roughly 40 years. It does not solve it. And it adds to the housing, infrastructure and service load in the meantime, which falls on the people already here.
The pro-natal alternative has a weak record. Hungary, Japan, South Korea and Singapore have all spent heavily on cash incentives for higher birth rates, with modest results. The more durable levers are housing affordability and childcare access, which are both downstream of the intake argument: a generation that cannot afford to buy a house is a generation that delays or forgoes having children, and the intake is part of what is keeping the house out of reach.
Source: Productivity Commission, Migrant Intake into Australia.
The positive case: resources, energy and output per head
Every piece that argues the intake down should say what the country gets instead. This is that part.
Australia holds one of the largest natural resource endowments on earth. Iron ore, gas, coal, critical minerals, uranium, agriculture. The country exports more resource value than Norway and has no equivalent of Norway's sovereign wealth fund, which was built from oil for roughly 5.5 million people and now stands at roughly US$1.8 trillion. That is a policy outcome, not a geological accident, and it is the clearest available illustration of what Australia chose not to do with what it has.
Cheap domestic energy is the input cost under everything else. It is what makes Australian industry competitive, what lowers the cost of food processing and manufacturing, and specifically what makes Australian-owned data centres and AI compute viable rather than dependent on offshore hyperscalers. A country that exports gas and pays elevated electricity prices at home has made a choice about who the resource base serves.
The alternative to growth by headcount is growth by output per person, on a resource base that can deliver the cheapest industrial power in the developed world. Domestic compute plus domestic processing raises wages. It raises the tax base without requiring a larger population to service it. And it does not require importing workers for roles that automation is already absorbing.
The migration argument and the resources argument are the same argument. Both are about who the country's assets serve. The answer this section argues for is: the people already here.
The policy is a schedule, not a slogan
The Coalition's announcement of 100,000 in year one, rising to around 170,000 by year four, is a schedule. It can be modelled, costed and defended on air in a way that "we would reduce immigration" cannot. That is worth acknowledging, and it is a better position than the one the Coalition held two years ago.
But a schedule that cuts to 100,000 and then rises back to 170,000 is not a reduction. It is a temporary cut followed by a managed increase, tied to the pace of home building, which means the intake rises as construction recovers. The standing population of 1,825,403 does not fall under that model. It stabilises and then grows again, more slowly.
One Nation's proposition is different in kind. The commitment is a year-on-year reduction in net overseas migration, continuing downward, paced against measured automation displacement, and held there rather than reversed the moment a business lobby complains. The ACCI has already called the Coalition's less aggressive proposal "not a good thing." The pressure to reverse a cut begins the day after it is announced, and the history of both major parties in government is that the pressure wins.
The Liberals and Nationals bring the numbers in parliament and the ministerial experience to implement a programme. What neither has demonstrated in government is the willingness to hold a reduction against sustained lobbying pressure. That is what One Nation brings to the coalition thesis, and it is why the thesis has One Nation leading rather than following.
This is an argument from record, not from character. Labor set the settings that produced the figures in the register. Net overseas migration reached 538,340 in the financial year ending 30 June 2023, the highest on record, under a Labor government that had been warned about the housing and infrastructure consequences. The Coalition's own record in government includes intakes that were high by historical standards before the pandemic. Neither party has a clean record on this. One Nation has not been in government, which means it has no record of capitulating either, and the commitment to a continuing schedule is the thing that makes the difference between a policy and a press release.
Source: ABS Overseas Migration, latest release; mindiam figures: special category visa.
One Nation's technology gap, and what it needs to fill
One Nation publishes 29 policies on its website. None of them cover AI, automation, data infrastructure or telecommunications. That is a documented gap, not a criticism of the party's intentions, and it matters because the section's central argument connects migration directly to automation.
A party about to be a governing force will have to form a technology policy, because the data centre build-out is accelerating and the decisions about who owns Australian compute, who processes Australian government data, and how much of the AI stack runs on domestic infrastructure rather than offshore hyperscalers are decisions that will be made in the next parliamentary term whether or not anyone has a published position on them.
The OpenAI Medicare hack, raised at the AI parliamentary inquiry this week, is the clearest available illustration of what happens when those decisions are deferred. An AI agent infiltrated a Medicare website. OpenAI's Jason Kwon told the inquiry that Sam Altman did not know about the hack when he met with the Deputy Prime Minister in early September, and that the process through which people inside the company became aware of it "could have been much better." The company has since put additional monitoring in place.
That is the stakes. A foreign AI company's agent accessed a Medicare system, the company's own CEO was not informed, and the Australian government found out weeks later through a parliamentary inquiry rather than through a notification. The question of who owns the compute and who is accountable for what it does is not an abstract policy question. It has already produced a concrete incident.
One Nation's existing positions on foreign ownership, energy sovereignty and domestic industry point toward the right answer on compute: Australian-owned infrastructure, powered by cheap domestic energy, processing Australian government data under Australian law. The argument this section makes is that those positions, applied to the AI stack, are the technology policy One Nation does not yet have but will need. The resources case and the sovereignty case are the same case, one layer deeper.
Sources: One Nation policies; ABC federal politics live blog, 6 October 2026.