The quiet revolution in Australian lending.

1. Summary

Two things happened to Australian non-bank lending at the same time, and they point in opposite directions.

Capital withdrew. Total fintech investment in Australia fell from approximately $2.9 billion in 2024 to approximately $940 million in 2025, a decline of around 68%.¹ The number of active independent fintech firms has fallen for four consecutive years, from 859 in 2022 to 801 in 2025, as weaker operators exited and stronger ones acquired.²

The compliance floor rose. Buy now pay later moved under full credit regulation in June 2025. Consumer Data Right obligations extended to non-bank lenders from 13 July 2026, with consumer data sharing phasing in from 9 November 2026 for initial providers and 10 May 2027 for larger ones.³

Demand did not move at all. The borrower cohorts that the major banks systematically decline are large, growing, and no better served now than they were three years ago.

This note argues that the combination produces a specific and familiar situation for a lower middle market investor. Not a growth market to enter, but a set of licensed operating businesses whose value is constrained by capital availability and compliance cost rather than by any deficiency in their commercial position.

2. The market did not fail. It rationalised.

The headline decline reads worse than the underlying position. The rate of contraction in firm numbers has slowed materially, which is consistent with an ecosystem that has finished shedding speculative entrants and reached a more sustainable size.²

Australian activity in 2025 was characterised by consolidation and acquisition, with stronger players strengthening position while weaker firms were forced to exit.² That is a description of a buyer’s market. It is also a description of the conditions under which good assets become available at prices that bear no relationship to their operating quality, because the constraint on the vendor is funding rather than performance.

3. What the first neobank wave established

Australia’s first cohort of challenger banks largely failed. Volt Bank, Xinja and 86 400 all exited or were absorbed between 2020 and 2023. Up is now held within Bendigo and Adelaide Bank. Judo, focused on business lending, built a defensible position and remains.

The pattern is worth stating precisely, because it is regularly misread as evidence that the sector does not work. It is not. It is evidence that capital-intensive full-service banking without a defined credit niche does not work. The operators that survived picked one borrower segment and built infrastructure around it.

Neobank account penetration in Australia sits at roughly 6%, well below comparable measures in the United Kingdom and Brazil. That gap is often presented as headroom. It is more accurately read as a closed window: the capital required to open a full-service challenger is no longer available at the terms that made the first wave possible.

4. The borrowers the assessment model cannot see

Traditional bank credit assessment in Australia was built on a narrow data model. It serves salaried homeowners reliably. It excludes several large cohorts whose income is real and verifiable but does not fit the template.

Table 1. Underserved borrower cohorts

Cohort	Indicative population	Why the model declines them
Self-employed and contract workers	1.1 million	Income is substantial and verifiable but arrives on irregular cycles. Automated scoring models built around fortnightly pay fail them on pattern rather than capacity.
Recent migrants	~500,000 per annum	Arrive with employment, savings and financial literacy, and no domestic credit history. Bureau scoring treats an absent file as an adverse one.
Post-event borrowers	~800,000	A default, hardship period or business failure followed by rebuilt and stable income. The bureau score carries the event for years after the underlying risk has resolved.
Small business owners	2.5 million+	Requirements of $5,000 to $50,000 attract application processes and approval timeframes calibrated for far larger exposures.

Population figures are indicative segment sizes drawn from Australian Bureau of Statistics and Treasury data. They are not addressable market estimates, and should not be read as such.

The gap here is not demand and it is not willingness to lend. It is infrastructure. Assessing these cohorts responsibly requires a data pipeline, a risk model and a product architecture that the incumbents have had no commercial reason to build.

5. The compliance floor is now the barrier to entry

Two regulatory changes have altered the economics of entry, and both cut the same way.

Buy now pay later moved under full credit regulation in June 2025, converting a category built specifically to sit outside the credit perimeter into one that carries responsible lending obligations and their associated cost.

Consumer Data Right obligations reached non-bank lenders on 13 July 2026, requiring product data sharing covering interest rates, fees, charges and eligibility criteria. Consumer data sharing follows from 9 November 2026 for initial providers and 10 May 2027 for large providers.³

For a new entrant this is straightforwardly bad news. Compliance infrastructure must now be built before the first loan is written rather than retrofitted at scale, and it must be built at a moment when funding for exactly that kind of pre-revenue build has fallen by two thirds.

For an existing licence holder the same change is protective. The obligations are a fixed cost that is already partly absorbed, and the data flowing through the regime narrows the information advantage that the major banks have historically held over the cohorts in Table 1.

6. What this means for a lower middle market investor

The situation resolves to a structural constraint rather than a commercial one, which is the condition this firm looks for.

There exist licensed Australian non-bank lenders with an Australian Credit Licence, AUSTRAC registration, AFCA membership, an operating loan book and demonstrated borrower demand, whose growth is limited by access to capital and by the cost of compliance obligations arriving on a fixed timetable. Nothing about their commercial position is broken. The constraint sits in the capital structure and in the regulatory calendar.

Three propositions follow.

6.1 The licence and the loan performance history are the scarce assets

Neither can be acquired quickly. An Australian Credit Licence, a risk model trained on real repayment behaviour rather than backtested assumptions, and embedded compliance infrastructure are built over years. A well-funded competitor cannot assemble them in a quarter. That is the definition of a durable position, and it is held by operators who are currently short of capital.

6.2 Building is the wrong side of the same trade

A new entrant must fund licence acquisition, compliance build, technology build and the loan book itself before writing meaningful volume, in a funding environment that has just contracted sharply. The same capital directed at an existing licence holder buys all four already in place, together with the borrower relationships and the performance data.

This is the same reasoning applied elsewhere in this research to modular construction. Where a supply base already exists and is competitive, capital deployed into building a new one chases an asset the incumbents already own.

6.3 The narrow product set is the discipline, not the limitation

The operators that survive the next cycle will not be those with the broadest product range. Concentration in a defined borrower segment reduces early capital risk, generates the volume of comparable decisions required to refine a risk model, and avoids the scope expansion that consumed the first neobank cohort.

7. What would falsify this

Three developments would materially weaken the argument above, and an investor should watch for each.

A recovery in fintech funding on 2021 terms would reopen the entry window and erode the advantage held by incumbent licence holders. Investor appetite is described as gradually recovering, assisted by expectations of lower rates and improved liquidity.² A partial recovery does not change the picture. A full one does.

A major bank building genuine capability against the cohorts in Table 1, most plausibly by acquisition rather than internal build, would remove the structural gap the thesis depends on.

A materially adverse turn in Australian consumer credit performance would test whether alternative assessment models hold outside benign conditions. No risk model in this segment has yet been observed through a full credit cycle, and that is the largest single unknown in the sector.

8. Conclusion

Australian lending has already been digitised. The open question is who owns the infrastructure that does it within the regulatory perimeter, and the answer is being decided now, in a market where the operators holding the licences are the ones short of capital and the parties holding capital are largely absent.

That is not a growth story. It is an availability window, and windows of this kind close when funding returns rather than when demand changes.