Skip to content
English
Try our demo now

Business failures in 2025: how to anticipate with AI and Open Banking

 

 

Introduction: the comeback of business failures post-Covid

During the Covid period, insolvencies across Europe plummeted. Governments stepped in with moratoria, guaranteed loans (PGE in France), and massive subsidies. But in 2025, the picture has reversed: business failures are rising sharply.

👉 For banks, insurers, fintechs, brokers and large enterprises, this is a turning point. Relying only on historical data and reactive processes is no longer sustainable. What’s needed is a shift towards predictive, explainable and real-time risk monitoring powered by AI and Open Banking.

This article covers:

  • The latest figures on business failures in 2025.

  • Why traditional models are failing.

  • How AI and Open Banking change the game.

  • Real-world use cases by sector.

  • A practical checklist to prepare.

 

 

1. Business failures in 2025: key figures and trends

1.1 France and Europe

  • France: +15% increase in insolvencies year-on-year (Banque de France).

  • Eurozone: insolvencies back to pre-Covid levels, with steeper increases in construction, hospitality, and retail.

  • SMEs/TPEs: most exposed, representing nearly 75% of failures.

 

1.2 Driving factors

  • Rising interest rates → heavier debt burden.

  • Inflation → margin squeeze, especially in energy-intensive sectors.

  • Supply chain fragility → delayed supplier payments cascade.

  • End of Covid aid → disappearance of the safety net.

 

👉 Bottom line: systemic risk is back, with failures potentially triggering chain reactions.

 

 

2. Why traditional approaches fail

2.1 Annual balance sheets

  • Published with a 12–18 month delay.

  • Fail to capture weak signals (cash erosion, payment delays).

  • A company can look healthy on paper while collapsing in reality.

 

2.2 Classical statistical models

  • Built on a handful of ratios (liquidity, leverage).

  • Poor coverage for thin-file SMEs and startups.

  • Not designed for dynamic, high-frequency data.

 

2.3 Lagging alerts

  • Insolvencies often detected too late, once liquidity is already gone.

  • Leads to heavy provisioning and write-offs for lenders and insurers.

 

👉 Traditional methods = reactive and costly.

 

 

3. AI and Open Banking: a new paradigm

3.1 Real-time banking data analysis

  • Open Banking APIs give access to company accounts (with consent).

  • Detect weak signals: declining revenues, growing supplier arrears, abnormal seasonality.

 

3.2 Predictive AI/ML models

  • Combine banking, accounting, legal and public datasets.

  • Learn from historical patterns of deterioration.

  • Generate dynamic risk scores, updated continuously.

 

3.3 Explainability (XAI)

  • Each alert comes with factor-level explanations.

  • Example: “Net cash down 35% over three consecutive months.”

  • Enables targeted, corrective action.

 

3.4 Inclusivity

  • Reliable scoring for SMEs, startups and atypical sectors.

  • Use of alternative data when limited history is available.

 

 

4. Real-world use cases

4.1 Bank: reducing cost of risk

  • Before: defaults only flagged after 90 days past due.

  • With AI + Open Banking: alerts triggered on early stress signals → proactive restructuring.

  • Impact: 25% reduction in losses on SME portfolios.

 

4.2 Commercial insurer: dynamic pricing

  • Before: premiums calculated annually on stale financials.

  • With AI: premiums adjusted dynamically based on real-time health.

  • Impact: improved loss ratio and fairer pricing.

 

4.3 Broker: dossier qualification

  • Before: heavy manual review for each file.

  • With RocketFin API: instant pre-scoring and explainability.

  • Impact: better hit rate, faster time-to-market.

 

4.4 Large enterprise (ETI): supplier risk monitoring

  • Before: critical supplier insolvencies hit without warning.

  • With continuous AI scoring: early alerts on deteriorating vendors.

  • Impact: protected supply chain continuity.

 

 

5. Practical checklist for 2025

  • Integrate Open Banking APIs with secure consent flows.

  • Deploy AI models combining banking, accounting, legal and public data.

  • Move to real-time, dynamic scoring.

  • Ensure explainability (factor-level breakdowns, reason codes).

  • Score inclusively: SMEs and thin-file companies included.

  • Implement webhooks for proactive alerts.

  • Stay compliant with GDPR, EU AI Act, PSD3.

 

 

Conclusion

The surge in business failures in 2025 is a wake-up call. Institutions must abandon post-mortem analysis and embrace predictive, explainable, real-time monitoring.

👉 By combining AI + Open Banking, financial institutions and corporates can:

  • Detect early warning signals.

  • Take proactive action.

  • Reduce losses and protect their value chains.

 

With RocketFin, you get real-time, explainable, inclusive scoring, delivered via API and enhanced with continuous monitoring.

 

Next step: explore the RocketFin demo and test proactive risk detection on your portfolio : https://www.rocketfin.ai/demo