The spreadsheet had a row for it, and that was the whole problem. In a meeting room somewhere offshore, an investment committee was working through a proposal to buy an Australian engineering services business — around 300 staff, a specialist book of maintenance work on pumping stations and treatment plants for regional water utilities, priced at about $310 million. The company was profitable, the vendor was a founder heading for retirement, and the fund had bought three similar businesses in other markets. Somewhere beneath the synergy assumptions, an analyst had inserted a line item labelled "regulatory review — timing and conditionality". It carried a number.
That number was not a tax, a fee or anything improper. It was an estimate of what it would cost the fund to hold a transaction open while an Australian government process decided whether to allow it: the deal team's time, the advisers, the vendor's right to walk away if the process ran long, the price protection the vendor would demand in exchange for waiting, and the possibility that approval would arrive with conditions attached to how the business could be operated afterwards. On the alternative target, in a market where a business of that size attracted no equivalent screen, the row was blank. The committee approved the alternative. Nobody in Australia ever learned that the first proposal had been considered, because it was never lodged.
That vignette is a composite, assembled from how mid-market acquisition processes ordinarily work rather than from any particular transaction, and it is offered as an illustration, not as evidence. It is placed first because it describes the one part of Australia's foreign investment regime that no official statistic can capture: the deals that are never attempted. Everything else in this article is measured, dated and sourced to the primary record. That part is not, and cannot be, and it is more honest to say so at the outset than to smuggle it in as a conclusion at the end.
The reason any of this matters begins with an accounting identity rather than an argument. Australia consistently invests more than it saves. The gap is filled by capital from the rest of the world, which is why the country has carried net foreign liabilities — a negative net international investment position — for essentially its entire modern history, and why the current account was in deficit for most of the post-war period, interrupted by a run of surpluses around 2019 to 2022 on the back of exceptional commodity prices before returning to deficit (Australian Bureau of Statistics, Balance of Payments and International Investment Position, various releases to 2026). The stock of foreign investment in Australia now runs into the order of $4 trillion or more, of which foreign direct investment — controlling stakes in real businesses rather than portfolio holdings — is roughly a quarter (approx.; ABS International Investment Position). Mines, ports, farms, factories, data centres and the transmission network were all built partly with money that came from somewhere else. That is not a vulnerability to be apologised for. It is the arrangement that has funded Australian development since the nineteenth century.
It does, however, make the price of that capital a first-order national question. And screening changes the price.
What the regime actually is, and what changed
The governing statute is the Foreign Acquisitions and Takeovers Act 1975. It requires foreign persons to notify the Treasurer before acquiring certain interests in Australian land, businesses and entities, and it gives the Treasurer the power to prohibit an action, or to allow it subject to conditions, if the action would be contrary to the national interest. The Foreign Investment Review Board is not a decision-maker; it is a non-statutory advisory body that examines proposals and advises the Treasurer, with the Treasury's Foreign Investment Division doing the administrative work. The decision is, and always has been, a ministerial one.
Most of the regime turns on monetary thresholds. Acquisitions below the relevant threshold generally do not require notification at all. The thresholds are indexed on 1 January each year, and currently sit at approximately A$339 million for developed commercial land and for business and entity acquisitions in non-sensitive sectors, and approximately A$73 million for sensitive land and sensitive-sector businesses (Foreign Investment Review Board threshold schedule, as at 1 January indexation). Higher thresholds apply to investors from countries with which Australia has a free trade agreement containing an investment chapter. Agricultural land carries a much lower cumulative threshold, and residential real estate carries none at all — every acquisition is screened.
There is one structural feature that matters enormously to the argument below and is almost never discussed in public debate. Foreign government investors — a definition that captures not only state entities but also many public pension funds, sovereign wealth funds and any fund in which government-linked investors hold an aggregate interest above the statutory percentage — face a nil threshold. For those investors, every acquisition of a direct interest is notifiable, regardless of size. The largest and most patient pools of long-horizon capital in the world are, by design, screened on every ticket.
The architecture, and who built it
The critique in this article is about the perimeter of the screen, and honesty requires being precise about who drew it. Most of the current architecture was not built by the government now administering it.
In March 2020, in the opening weeks of the pandemic, the then Coalition government reduced all monetary screening thresholds to zero, temporarily making every foreign acquisition notifiable. That emergency measure was unwound at the end of that year, but it was replaced by something permanent and far more consequential: the Foreign Investment Reform (Protecting Australia's National Security) Act 2020, which took effect on 1 January 2021. That Act introduced a mandatory national security test operating independently of the monetary thresholds. It defined a "national security business" and "national security land", required notification of any acquisition of a direct interest in them regardless of value, gave the Treasurer a call-in power over actions that were not otherwise notifiable, and created a last-resort power to reassess and impose conditions on actions already approved. It is worth noting the legislation's own language: it speaks of national security businesses and national security land. It does not name countries. Neither does this article.
Running alongside it is the Security of Critical Infrastructure Act 2018, which began by covering four sectors — electricity, gas, water and ports — and was expanded through amendments in 2021 and 2022 to cover eleven, including data storage and processing, communications, financial services, health care and medical, higher education and research, transport, food and grocery, defence industry and space technology. That expansion carried positive security obligations, asset registration duties and government assistance powers, and it substantially widened what counts as a critical asset for foreign investment purposes as well.
Every one of those measures was legislated before May 2022. The current government inherited the screening architecture; it did not build it. What it has done since is administer it, extend its reach in specific directions, and — this cuts the other way — invest real money and attention in making it run faster.
The two things this government added
The first addition is administrative, and it is genuinely good. In May 2024 the Treasurer announced a package of reforms to the foreign investment framework built around risk-proportionate processing: faster treatment for investors with a known track record in non-sensitive sectors, more resources directed at genuinely sensitive cases, and a published service commitment. From 1 January 2025 the government committed to deciding 50 per cent of commercial foreign investment proposals within 30 days of the fee being paid (Treasury, foreign investment framework reforms). On 28 May 2025 the Foreign Investment Portal went live, replacing the previous application arrangements with a single digital lodgement system (Treasury). The government also said it would prioritise proposals that support new housing supply, including build-to-rent development. These are the actions of an administration that takes processing burden seriously, and the data below shows they are producing results.
The second addition is not administrative at all. It is a deliberate deterrent aimed at a defined category of buyer, and it should be assessed as such rather than mistaken for red tape.
From 9 April 2024, application fees for foreign persons acquiring established dwellings were tripled (Australian Government foreign investment fee schedule). The vacancy fee charged annually to foreign owners of under-occupied residential dwellings was increased in the same package. Fees across the framework are indexed annually on 1 July. And in February 2025 the government announced a temporary prohibition on foreign persons purchasing established dwellings, applying from 1 April 2025 for two years with limited exceptions (Treasury, foreign investment policy announcements).
Tripling a fee is not friction. It is a price signal, openly stated, intended to change behaviour. That is a legitimate instrument for a government to reach for. The right question is not whether it is permissible but whether it is well targeted and whether it does what it claims.
The strongest case for the policy
The world changed, the screen followed, and Australia is not acting alone
Put the government's case at its best and it is strong enough that a fair critic has to concede most of it before saying anything at all. The strategic environment in which foreign investment operates has materially changed over the past decade. Critical infrastructure that was once treated as ordinary commercial property — electricity transmission, ports, telecommunications, water, and increasingly data storage and processing — is now understood to carry systemic consequences if control or access is compromised. That understanding is not speculative or partisan. It is embedded in Australian legislation passed with bipartisan support, and it is reflected in the parallel tightening of investment screening regimes across the United States, Canada, the United Kingdom, Japan and the European Union over the same period. On this measure Australia is following an international trend among advanced economies, not striking out alone, and an Australian government that had left its screen untouched while every comparable jurisdiction tightened theirs would have been the outlier requiring explanation.
Second, and this is the point most often lost: screening is not prohibition. The overwhelming majority of applications are approved. Outright prohibitions are rare — a handful in a typical year against tens of thousands of applications (approx.; Foreign Investment Review Board annual reports). Where the Treasurer has concerns, the ordinary instrument is a condition attached to an approval rather than a refusal. A regime that says yes to almost everyone, subject to terms, is a very different animal from one that says no.
Third, the administrative record supports the government rather than embarrasses it. A median decision time of about 35 days is not obstruction by any reasonable standard; it is faster than the internal approval processes of many of the acquirers concerned. The government set a public target, built a digital portal to meet it, and published the data against which it can be judged. Governments that intend to hide behind process do not publish the processing statistics.
Fourth, housing is a legitimate area for differential treatment. Public concern about the availability and price of dwellings is genuine and well founded, and a government is entitled to decide that the domestic housing stock should be prioritised for people who live in the country. Foreign purchases of established dwellings add demand to a fixed stock without adding supply, which is a defensible reason to treat them differently from investment that builds something new. The framework has always drawn exactly that distinction: new dwellings and development are encouraged, established dwellings are restricted.
Finally, the asymmetry of consequences. Deal friction is a cost measured in basis points and months. Losing effective control of a strategically significant asset, and then having to buy it back or regulate around it, is a cost measured in billions and decades — as several jurisdictions have discovered. A government weighing a modest, diffuse, hard-to-observe cost against a low-probability but severe and irreversible one is behaving rationally when it errs towards caution. Any critique that ignores that asymmetry is not a serious critique.
The evidence: a regime that works reasonably, and a cost that does not show up in it
The crude version of this argument — that the Foreign Investment Review Board is slow and is choking investment — is not supported by the data, and it will not be made here. It fails on the numbers, and repeating it would discredit everything else in this article.
What the processing data actually says
On Treasury's published quarterly foreign investment data, the median processing time for proposals is around 35 days, and approximately 46 per cent of proposals are decided within 30 days (Treasury quarterly foreign investment data, latest available). The government's published target, in force since 1 January 2025, is 50 per cent of commercial proposals decided within 30 days. So the regime is running a little short of a target it set itself only recently, having built a new digital lodgement system in the same period. That is a system performing respectably and moving in the right direction.
Set the 35-day median against the commercial reality it sits inside. A mid-market acquisition typically runs a due diligence period of six to twelve weeks. Financing commitments, board approvals and vendor negotiations occupy similar spans. A regulatory decision that arrives within about five weeks of lodgement can, in a well-run process, be substantially absorbed into work that was happening anyway. On the central case, Australian screening is not the binding constraint on transaction timetables.
The median, though, is the statistic most likely to mislead. A median tells you about the middle of the distribution, and the middle of this distribution is dominated by residential real estate applications, which are numerous, standardised and quick. The cases that matter to the cost of capital sit in the tail: large or sensitive business acquisitions, critical infrastructure, foreign government investors, and anything touching the national security test. Those cases are a small share of the count and a very large share of the dollars. Treasury does not routinely publish a separate distribution for them, and until it does, the median will continue to make the regime look easier than it feels to the investors whose money the country most wants.
This is not an allegation of concealment. It is a request for a statistic that does not currently exist, and the absence of it is why so much of the public argument about foreign investment is conducted on anecdote. A published breakdown of processing times by application class and by sensitivity would settle a great deal of this dispute in one direction or the other. That the government publishes as much as it does is to its credit; that this particular cut is missing is the single most useful thing it could fix.
What the approval data says
Approval rates are extremely high. Across the Foreign Investment Review Board's annual reporting, the number of proposals actually prohibited in any year is measured in single or low double digits, against tens of thousands of applications considered (approx.; FIRB annual reports). Refusal is not the instrument this regime uses.
Conditions are. A substantial share of business approvals carry conditions — most commonly the standard tax conditions applied since 2016, and, in sensitive cases, obligations touching data handling, access arrangements, reporting, security controls or the composition of a board. These conditions are the regime's real working edge, and they are largely invisible in headline statistics, because a conditioned approval is counted as an approval.
A regime that approves almost everything is not therefore costless. The price is paid in the conditions attached, and in the deals that never reach the counting.
That matters, because a condition is a permanent constraint on how an asset is operated, priced into the asset's value from the moment it is imposed and carried by every subsequent owner. It does not expire when the deal closes. Counting it as a "yes" is accurate but incomplete.
The housing measure, assessed on its own terms
The established-dwelling measures deserve a fair, separate assessment, because they are the clearest case in the framework of an instrument doing exactly what its designers said it would.
Start with the base. Foreign persons who are not temporary residents have long been prohibited from acquiring established dwellings. Temporary residents have been permitted to buy one established dwelling to live in, with an obligation to sell it on departure. The category being deterred by the 2024 fee increase and then prohibited outright in 2025 was therefore already a narrow one, sitting on top of restrictions that had been in place for years.
Now the scale. Foreign purchases of established dwellings number in the low thousands a year, against Australian dwelling transfers running in the hundreds of thousands annually — a small fraction of one per cent of the market (approx.; Treasury quarterly foreign investment data on residential real estate, and ABS dwelling transfer and lending series). Even a measure that eliminated the category entirely would remove a small fraction of one per cent of demand from a market whose price problem is driven by the interaction of supply constraints, population growth, credit conditions, tax settings and construction costs, each of which is an order of magnitude larger.
The honest conclusion is uncomfortable for both sides of the argument. The measure is a legitimate, openly declared policy choice, and it is not a scandal. But its likely effect on housing affordability is modest relative to its political salience, and it was applied to a category that was already substantially restricted. Tripling a fee on a class of buyer and then prohibiting that class of buyer twelve months later also raises a question about the internal logic of the sequence: if the prohibition was the right answer in 2025, the fee increase in 2024 was either insufficient or unnecessary. A government is entitled to escalate. It is also fair to observe that escalation of this speed suggests the first instrument was chosen for its visibility as much as for its measured effect.
The capital the country is screening
The final piece of evidence is the one that gives the argument its stakes. Australia's net international investment position has been negative throughout its modern history: the country owes more to the rest of the world than the world owes it, because it has consistently invested more than it saved (ABS, International Investment Position, various releases to 2026). Total foreign investment in Australia runs to the order of $4 trillion or more, with foreign direct investment in the order of $1 trillion (approx.; ABS International Investment Position). Foreign investment is not a marginal supplement to Australian capital formation. It is a structural component of it.
Every increment of screening friction — every extra week of uncertainty, every additional condition, every deal not attempted — is a small addition to the return an investor requires before committing money here rather than somewhere else. That is what "cost of capital" means in practice: not an interest rate on a statement, but the hurdle a project has to clear to be built at all. Raise it by a small margin and the projects at the margin do not get built. Nobody announces that. It simply does not happen.
The critique: the invisible margin
The case against the current settings is not that the regime is slow. It is that a regime is not costless merely because most applications succeed, and that the costs it does impose are systematically invisible in the statistics used to defend it.
Where the screen binds hardest
Screening costs are largely fixed. Legal advice, a national security assessment, a structuring analysis, the internal time of a deal team and the price of holding a transaction open cost broadly similar amounts whether the target is worth $80 million or $8 billion. As a proportion of transaction value, that fixed cost is trivial on a very large deal and material on a mid-market one.
The consequence is the opposite of the intuitive one. The screen does not deter the largest acquirers; they are well advised, well capitalised, patient, and they were always going to be screened. It bites hardest in the middle market — the $50 million to $500 million range where a great deal of Australian industrial capability, specialist manufacturing, agricultural processing and mid-tier resources sits, and where a bidder can simply choose a different jurisdiction with less inconvenience than a large strategic acquirer can.
The nil threshold for foreign government investors sharpens this further. Because the definition reaches public pension funds and sovereign funds, and because interests are aggregated, some of the most patient, lowest-cost, longest-horizon capital in the world is screened on every acquisition regardless of value. That capital is precisely the kind Australia most wants for infrastructure and the energy transition — investors with thirty-year horizons and low return requirements. Applying a nil threshold to them is a defensible policy choice with a real and rarely acknowledged price.
The exit problem
There is a second-order effect that almost never features in the public debate and is, in valuation terms, the more serious one.
When an investor buys an Australian asset, they are also buying a future obligation to sell it. If the asset sits inside the screening perimeter, a material portion of the world's potential future buyers will need approval to purchase it. That thins the buyer universe at exit, which lowers the expected terminal value, which raises the return required at entry.
This effect operates on every asset inside the perimeter, including assets owned by Australians, and it operates whether or not any application is ever refused. A domestic owner of a business that has been reclassified as a national security business finds that the pool of buyers for their life's work is smaller than it was, not because anyone made a decision about them, but because a definition moved. That is a genuine cost, borne by Australians, and it appears in no departmental statistic.
The deals that are never lodged
The largest category of cost is the one that cannot be measured, and it must be labelled as an inference rather than presented as a finding.
Investment committees allocate capital comparatively. A fund with a mandate across several markets ranks opportunities by risk-adjusted return, and regulatory execution risk is an ordinary input to that ranking, sitting alongside currency, tax, labour cost and political stability. Where a jurisdiction carries a screening requirement that a comparable jurisdiction does not, that difference is priced. Deals that fail the ranking are not rejected by any government; they are simply never brought forward. No agency ever sees them, no statistic records them, and no minister is accountable for them.
Two things must be said about this argument, and the second is the more important. First, it is not a measured quantity. It is a structural inference from how capital allocation works, and anyone who tells you they know its magnitude is guessing. Second, it is not unfalsifiable hand-waving either: it is the standard reasoning by which every finance ministry in the world assesses investment attractiveness, and it is the reason competing jurisdictions publicise the speed and predictability of their own approval processes. The correct intellectual posture is to hold it as a real but unquantified cost — not zero, not catastrophic, and above all not something to be waved away by pointing at an approval rate.
The transactions that are never attempted appear in no statistic, and a regime cannot be judged solely by the applications it happens to receive.
Perimeter drift
Which brings the argument to its actual question. Not "is screening justified?" — it plainly is, for genuinely strategic assets — but "who decides where the perimeter stops, and against what test?"
The perimeter has widened considerably. It now runs around critical infrastructure across eleven sectors, around critical minerals, and around businesses holding sensitive data. Each extension had a coherent rationale when it was made. The difficulty is cumulative and definitional. "Critical minerals" is a list maintained administratively and revised as technology and policy priorities change. "Sensitive data" is a category that expands with the digitisation of the ordinary economy: a logistics company, a payments processor and a health services provider all hold data that a decade ago would not have been thought of in these terms at all.
A perimeter defined by categories that expand on their own does not need a policy decision to grow. It grows by definition creep, which is the same mechanism that makes any well-intentioned regulatory boundary drift outward over time. And because the national interest test is deliberately unenumerated — the Foreign Acquisitions and Takeovers Act does not define the national interest, leaving it to the Treasurer of the day — there is no statutory anchor that fixes how wide it can get. That flexibility is a feature when the threat picture changes quickly. It is a weakness when the pressure to widen is political rather than strategic, because there is nothing in the statute that says no.
The test of whether drift is occurring is not whether individual decisions look wrong from outside; they usually cannot be assessed from outside, because the underlying security assessments are properly classified. The test is institutional. Is there a published, principled statement of what the perimeter is for and what would justify contracting it as well as expanding it? Is there periodic independent review of whether categories added years ago still earn their place? Is there transparency about the classes of condition being imposed, even in aggregate and without identifying transactions? At present the answer to each is partial at best, and the burden of demonstrating that a widening perimeter remains strategically anchored properly sits with the government, not with the investors it is screening.
What they got right, and the honest caveats
Four things belong plainly on the government's side of the ledger, and the critique above is only credible if they are stated without hedging.
First, the processing performance is genuinely respectable. A median of about 35 days for a national interest and national security assessment is a defensible outcome by any international standard, and 46 per cent of proposals inside 30 days is not the profile of an obstructive agency (Treasury quarterly foreign investment data). Anyone arguing that Australian screening is slow is arguing against the published record.
Second, the administrative reforms are real, not cosmetic. Setting a public target of 50 per cent of commercial proposals within 30 days from 1 January 2025 created an external standard the government can be held to, which is the opposite of what an administration seeking cover would do. Building and launching the Foreign Investment Portal on 28 May 2025 addressed a real friction in lodgement and case management. Directing scrutiny towards genuinely sensitive cases and away from repeat investors in benign sectors is exactly the risk-proportionate design a critic would have recommended.
Third, approvals remain the overwhelming norm. Prohibitions are rare, and the conditions-based approach — allowing a transaction subject to terms rather than blocking it — is a proportionate instrument that keeps capital flowing while managing identified risk.
Fourth, and most importantly for the fairness of this article, the strategic rationale for the expanded screen is real and not pretextual. The security considerations attaching to critical infrastructure, critical minerals supply chains and large holdings of sensitive personal data changed materially over the past decade. That change is documented in Australian legislation passed with support from both major parties, and it is reflected in parallel reforms across comparable advanced economies over the same years. This is not a case of a government inventing a threat to justify an intervention.
The caveats on that credit are equally honest. Processing performance is measured on a median dominated by simple residential applications, and the complex end of the distribution is not separately published. The 50 per cent target is not yet being met. The established-dwelling fee increase was aimed at a category already tightly restricted, and its housing-affordability effect is likely modest relative to the attention it received. And on attribution: the architecture that generates most of the friction discussed here — the national security test, the call-in power, the last-resort power, and the expanded critical infrastructure regime — was legislated between 2018 and 2022, under the previous government. The current government inherited it, has administered it faster than it found it, and has extended it in specific directions of its own choosing. Those are three different things and they deserve three different judgements.
The risk to the future
A capital-importing country raising the price of the capital it imports
The forward risk is not that Australia will be starved of foreign investment. Capital will keep arriving; the country is stable, well governed and resource-rich, and those advantages are considerable. The risk is subtler and therefore easier to ignore: that the price of that capital rises by a margin nobody measures, at precisely the moment when the volume required is unusually large and the competition for it is unusually fierce.
The volume required is not in doubt. Replacing and expanding the electricity system, building transmission, developing critical minerals processing rather than exporting the raw material, and constructing housing at the rate the population implies are all capital-hungry undertakings on decade-long horizons. Australian domestic savings, including the superannuation pool, cannot fund all of it, which is the same arithmetic that has held for a century and a half. The capital has to come partly from outside.
The competition is not in doubt either. Other advanced economies are actively bidding for the same investment, and several are doing so with direct subsidy — production credits, capital grants and concessional finance aimed squarely at the industries Australia says it wants. A country cannot control what its competitors offer. It can control whether its own approval process is a reason to look elsewhere. When rivals compete on subsidy and Australia competes on the quality of its assets, the friction in getting an asset bought is a larger part of the comparison than it would otherwise be.
None of this is a prediction, and it should not be read as one. Screening may prove to have imposed a negligible cost. The strategic protection it buys may prove to have been worth many multiples of whatever it cost. Both are entirely possible, and the honest position holds them open. But the asymmetry runs in both directions, and the second direction gets far less attention than the first. The cost of a strategic asset falling into the wrong hands is severe, visible and politically survivable to have guarded against. The cost of a decade of marginal projects that were never built, in a country that needed them, is diffuse, invisible, and attributable to nobody at all. The first risk has a minister accountable for it. The second does not, which is exactly why it is the one more likely to be run.