Key implication: Australia’s credit appetite is reaccelerating at the exact moment balance sheets are stretched. Mortgage enquiries are at a three‑year December high in 2025, up 17.9% year on year (Equifax Consumer Market Pulse), while credit card debt jumped about 15% over the period. That mix — secured credit interest returning alongside heavier unsecured usage — is a late‑cycle tell. For lenders, the margin opportunity is real, but so is the tail risk. The edge now lies in AI‑enabled underwriting, precision pricing and proactive collections.

Market context: a two‑speed credit cycle

On the demand side, Equifax data shows a sharp rebound in mortgage intent, with Queensland, Western Australia and New South Wales leading state‑level gains. Simultaneously, consumers leaned harder on credit cards and personal loans, signalling pressure from cost‑of‑living and rental inflation. It’s the classic two‑speed cycle: higher‑ticket secured borrowing returns as buyers anticipate stabilising rates, while revolving credit becomes a bridge for household cashflow.

For banks and non‑banks, this splits into two P&L realities. Mortgages promise scale and customer lifetime value but carry thin front‑book margins and heightened funding sensitivity. Unsecured lending offers yield but amplifies loss volatility and fraud exposure. The commercial question is not whether to participate, but how to re‑price, re‑underwrite and re‑collect in a market that is both warming and riskier.

Business impact: the new lending economics

Map this through a unit economics lens:

• Acquisition and conversion: Broker channels should see volume tailwinds as enquiry funnels reopen. Speed to yes — not just rate — becomes the conversion wedge.
• Risk and provisioning: Heavier unsecured usage is a leading indicator of stress that can travel into mortgage arrears with a lag. Expect pressure to lift expected credit loss overlays even as headline demand rises.
• Funding and capital: Non‑banks reliant on warehouse lines must lock in flexibility; banks face capital optimisation decisions as the mortgage mix shifts. Precision risk‑based pricing can protect net interest margins without bluntly choking growth.
• Cross‑sell and retention: Surge periods create refi churn. Without data‑led retention triggers and offer personalisation, portfolios will leak the best credits to competitors.

Net result: growth is back, but it is ‘expensive growth’ unless underwriting and collections efficiency improve in step.

Competitive advantage: AI as the operating lever

Australia’s ecosystem is primed for a technology‑led response. The UK–Australia FinTech Bridge underscores supportive regulation and high adoption, while local momentum around AI is tangible — from Fortiro’s 2024 award for Best Use of AI in FinTech to Microsoft’s AI FinTech roadshow in Sydney and capability uplift programs from CA ANZ. The play is not shiny demos; it’s operational deployment across the lending value chain:

• Application triage: Machine learning models segment low‑risk, standard files for straight‑through processing and route edge cases to specialists, cutting decision times and abandonment.
• Affordability and income verification: AI aids extraction and reconciliation of income and expense data, reducing manual errors and speeding assessment. Combined with Consumer Data Right consents, lenders can shift from stated to observed affordability.
• Fraud and document integrity: AI‑driven document analysis spots tampering patterns, synthetic identities and inconsistencies — critical as fraud attempts typically rise with unsecured credit growth.
• Pricing and retention: Real‑time risk‑based pricing engines allow micro‑segmented offers that defend margin while keeping the best credits in‑house.
• Early collections: Behavioural models and nudges trigger earlier, more empathetic outreach, improving cure rates and reducing roll rates.

Industry practitioners increasingly frame AI ROI in terms of double‑digit reductions in time‑to‑yes, measurable drops in manual touchpoints and improved fraud catch rates — outcomes that directly support both growth and risk agendas.

Implementation reality: speed without breaking risk

Execution is where advantage compounds. Global supervisors, including the IAIS’s 2023–2024 work on AI and third‑party risk, highlight model governance, cyber resilience and vendor oversight. Translate that to the Australian shop floor:

• Model governance: Establish clear ownership, validation and monitoring for machine learning models used in credit decisions. Log and explain decisions to align with responsible lending expectations.
• Data pipelines: Prioritise secure ingestion of CDR and payroll data with lineage, quality checks and consent management. Bad data will undermine even the best models.
• Third‑party risk: Assess cloud, model and document‑analysis vendors for resilience and recovery. Concentration risk matters when decisioning is outsourced.
• Human‑in‑the‑loop: Keep adjudicator checkpoints for borderline files and periodic challenger reviews. AI can prioritise, humans must finalise.
• Change management: Upskill credit teams. Programs from professional bodies such as CA ANZ signal the skills shift under way across finance teams adopting AI.

The practical hurdle is integration debt: legacy LOS, fragmented broker portals and batch‑based data flows. Tactical progress comes from API‑first components that wrap legacy cores rather than big‑bang replacements.

Industry transformation: brokers, non‑banks and the data dividend

Brokers remain the front door for a majority of Australian mortgages, so fintechs that embed AI into broker workflows — income verification, document integrity, pre‑assessment — will win share of attention and referrals. Non‑bank lenders, often nimbler on technology, can turn AI speed into pricing flexibility, provided they match it with disciplined funding. Banks, meanwhile, can lean on balance‑sheet strength to scale straight‑through approvals for prime segments and use data to surgically defend against churn.

Expect more cross‑border collaboration under the FinTech Bridge, especially in fraud analytics and explainable AI. The strategic frontier is turning Australia’s CDR data into sustained advantage: more precise, real‑time affordability means fewer unpleasant surprises later in the cycle.

Scenario outlook: three paths to 2026

• Soft‑landing base case: If inflation moderates and rate cuts emerge, today’s enquiry surge converts into settlements and a refinancing mini‑wave. Priority: scale capacity, automate verification, and deploy retention pricing early.
• Sticky inflation case: Higher‑for‑longer rates keep unsecured stress elevated, lifting arrears risk. Priority: tighten scorecards at the margin, strengthen early‑collections, and raise ECL overlays while protecting prime growth.
• Downside case: Labour market softens and delinquencies rise. Priority: pivot to resilience — harden underwriting, slow higher‑risk segments, and redeploy AI to collections and hardship support to minimise loss‑given‑default.

Action plan for decision‑makers

• Re‑segment risk: Use unsecured utilisation and expense inflation as forward indicators in mortgage scorecards; adjust cut‑offs by cohort, not blunt portfolio caps.
• Compress time‑to‑yes: Target near real‑time decisions for low‑risk files by deploying AI triage and CDR‑driven verification; make broker SLAs a competitive weapon.
• Harden fraud controls: Integrate AI‑based document and identity verification at lodgement; measure false‑positive rates and continuously calibrate.
• Dynamic pricing and retention: Stand up risk‑based pricing that responds to enquiry spikes; use proactive retention offers for high‑value customers at pre‑refi triggers.
• Strengthen governance: Align AI model governance, vendor risk and cyber controls with global best practice; ensure ASIC and APRA expectations on responsible lending and operational resilience are front‑of‑mind.
• Invest in talent: Upskill credit and data teams; partner with fintech specialists where it accelerates time‑to‑value.

Bottom line: The credit taps are opening, but the water pressure isn’t uniform. Those who combine disciplined risk with industrial‑grade AI will convert a volatile upswing into durable market share and superior returns.