June 6, 2026|⏱️~10minutes

By Nicholas Brennan

Small and medium manufacturers around the world face the same problem. They need money to grow, but they often lack the property or land that banks typically ask for as collateral. At the same time, banks find it hard to check how a business is really doing. Financial statements can be polished. Audit results can be out of date. But the bank cannot see if the machines are actually running.

In the past few years, a new trend has begun to attract attention. There are signs that some lenders have started looking at realtime machine data – hours of operation, energy use, maintenance records – as part of their loan decisions. This approach is still early in its development. But the shift in thinking is worth looking at: moving from static records and physical collateral to dynamic, realtime operating data.

1. What problem does machine data solve?

Traditional lending has a basic problem: information asymmetry.

Banks want to lend to businesses that are stable and can repay. But they cannot watch how a business is doing every day at low cost. Many smaller companies do not have complete financial statements. And valuing and selling collateral takes time and money.

Machine data offers a different path. When equipment is connected to a network, it records its own hours of operation, load levels, unplanned stops, and maintenance frequency. This data tells you how active the business really is.

A reasonable assumption is that steady, consistent machine use usually means steady orders and steady revenue. Machines that sit idle or break down often may signal trouble. Unlike financial statements, machine data is harder to fake. While not impossible, tampering with sensors costs more and requires more skill than altering paperwork.

2. Two models in practice

Different markets have tried slightly different ways to use machine data for lending.

Model one: Repayments follow machine usage.
In 2018, Germany's Commerzbank worked with machine tool maker EMAG to launch what was called a "payperuse" loan. Companies borrowed money to buy equipment, but their repayments were not fixed. Instead, they went up or down based on how many hours the machine actually ran. In slow months, they paid less. In busy months, they paid more. EMAG's COO at the time said this linked the cost of the investment directly to the company's revenue.

Model two: Machine data as a reference for credit limits.
In some supply chain finance platforms in Asia, companies allow lenders to access their equipment data. Lenders use that data to assess how active and stable the business is. Machine data is rarely the only factor, but it gives an extra layer of proof for companies that lack traditional collateral.

It is worth noting that both models are still limited in scale. More data is needed to see how well they hold up over time.

3. Why this might work – economic logic

There are a few economic reasons why machine data could work as a lending tool.

First, it lowers the cost of signaling.
Nobel Prize winner Michael Spence showed that highquality sellers need a credible way to prove their quality. For a manufacturer, steady machine use is costly to achieve and hard to fake. A struggling factory cannot maintain high operating hours and low failure rates for long. So machine data can serve as a signal that separates good borrowers from weak ones.

Second, it allows dynamic risk monitoring.
Traditional lending checks risk mostly before the loan is made. After that, the bank has limited visibility. Machine data changes that. If operating hours drop suddenly, or energy use falls, the system can send an early warning. This does not prevent default, but it may give the lender more time to respond.

Third, it reduces reliance on selling collateral.
The old problem: if a borrower defaults, the bank must seize and sell assets. That process takes months and recovery rates vary. A loan based on machine data focuses on what the equipment is producing right now – not what it could be sold for. A 2023 report from the China Academy of Information and Communications Technology called this a shift from "stockbased credit" to "flowbased credit". That idea makes sense in theory, but its realworld sustainability needs more testing.

4. Examples and numbers

Commerzbank's pay-per-use loan (Germany, 2018).
The repayment amount was calculated based on actual usage data from connected machine tools. According to the bank and EMAG, it was one of the few successful attempts to turn Industry 4.0 theory into practical lending.

SenseLink IoT platform.
A Xinhua News report from December 2024 said that this platform worked with multiple banks to monitor warehouse goods using IoT sensors. By the end of 2024, it had helped thousands of companies secure over 240 billion yuan (about $33 billion USD) in loans. The reported nonperforming loan ratio was zero. It should be noted that a zero default rate may reflect an early stage or very specific risk controls. It is not necessarily repeatable everywhere.

Zhongqi Yunlian's "de-anchored" financing model.
As of 2025, this platform reported serving over 300 core supply chain companies and thousands of small businesses across construction, manufacturing, agriculture, and nearly 20 other sectors. The platform's own data claimed financing costs dropped by about 50% and efficiency improved by roughly 70%. These figures have not been independently verified.

Dynamic IoT risk monitoring.
An industry analysis from late 2025 suggested that small business loans using IoTbased dynamic monitoring improved early warning accuracy by roughly 40%, and saw average nonperforming loan ratios fall by over 30% compared to traditional methods. This was based on a specific report's methodology. Results may vary by industry and lender.

5. Current limits and open questions

Despite the logic, this approach faces several realworld limits.

Limited scope of application.
Machine data works best for manufacturing, logistics, and agriculture – sectors that own production equipment. For services, retail, or pure trading businesses, machine data does not tell you much. This is not a universal solution. It is a niche tool with potential in specific industries.

Unclear data ownership and privacy.
When equipment makers, IoT platforms, lenders, and the business itself all see the same machine data – who owns it? If a company switches to a different bank, can the old bank keep that historical data? Laws vary widely across jurisdictions, and most have not caught up.

Security and tampering risks.
It is generally harder to fake machine data than paper records. But no system is perfect. IoT devices can be hacked, and data streams can be intercepted. Lenders will need to keep investing in security.

Potential algorithm bias.
Consider seasonal manufacturers – for example, a factory that makes Christmas decorations. It runs at full capacity for only four months a year. The rest of the time, its machines sit idle. That does not mean it is a bad business. But a scoring model that only looks at average utilization might wrongly flag it as risky. There is no public evidence yet that lenders' models have solved this problem well.

Platform concentration risk.
More and more machine data and lending decisions depend on a small number of industrial data platforms. If one of those platforms has a technical failure, a security breach, or goes out of business, many lenders and borrowers could be affected at once. Regulators have not yet set clear requirements for the resilience of these new systems.

6. An important distinction

It helps to be clear about what machine data is and is not. Machine data is a credit assessment tool. It is not credit itself. A loan is still based on the borrower's ability and willingness to repay. Machine data offers a new window into that ability – especially for manufacturers – but it cannot fully replace traditional financial analysis and due diligence.

7. What might happen in the next few years

Based on current trends, a few cautious predictions are possible.

Combining multiple data sources.
Machine data will likely be used alongside other information – electricity bills, tax records, bank account activity, supply chain transactions. A multisource picture is likely to be more reliable than any single data stream.

More flexible loan products.
We may see loans where the limit automatically adjusts based on recent machine utilization, or where the interest rate varies with maintenance records or energy efficiency. The technology already exists. Whether the business model and regulation will support it is another question.

Infrastructure growth.
China's Ministry of Industry and Information Technology reported in January 2026 that the core industrial internet sector was expected to exceed 1.6 trillion yuan in 2025, with over 100 million connected devices on major platforms. In May 2025, China's central bank published technical guidelines for IoT applications in finance (JR/T 03382025). These kinds of infrastructure and policy moves may create more favorable conditions for datadriven lending.

Still, this approach will not "solve" small business lending. At best, it can ease constraints for a specific group of solid but collateralpoor manufacturers.

Closing thoughts

Machine runtime data is starting to be used by some lenders as a supplementary input for loan decisions. The core idea is simple: shift attention from what a company used to own to what it is actually doing right now.

This shift is not going to overturn the entire lending system. But for a certain type of business – a manufacturer with real production but no property to mortgage – it offers a new possibility.

At the same time, this model is still early. It faces real uncertainties around data ownership, algorithm bias, platform security, and regulatory clarity. Whether it will grow significantly in the coming years depends on falling technology costs, better rules, and lenders finding a sustainable balance between risk and return.

One plausible conclusion: machines are becoming an honest witness in the business of credit assessment. How much that witness will change the rules of lending is a question that only more practice – and more time – can answer.

Disclaimer: This article is for information and analysis only. It does not constitute financial, investment, or lending advice. The examples and data cited come from public sources or industry reports and do not imply endorsement of any product or institution. The author has no affiliation with any organization mentioned. Readers should consult qualified financial professionals before making any loan or financing decisions.

References

[1] Commerzbank AG. "Commerzbank the first German bank to offer new databased loans for corporate clients." Press release, June 5, 2018.

[2] Shah, Vitarag. "How IoTEnabled Loan Origination Platforms Are Transforming Credit Decisioning in the UK." IoT For All, April 29, 2026.

[3] China Academy of Information and Communications Technology (CAICT). "IoTEnabled Financing for Small and Micro Enterprises." July 2023.

[4] SenseLink platform. Xinhua News Outlook Weekly report on "Digitalising movable assets for credit". December 2024.

[5] People's Bank of China. "Guidelines for IoT Technology in Financial Applications" (JR/T 03382025). May 2025.

[6] People's Bank of China, Ministry of Industry and Information Technology, et al. "Guidance on Financial Support for New Industrialisation." August 2025.

[7] EMAG GmbH & Co. KG. "Payperuse loan" technical report. 2018.

[8] Spence, Michael. "Job Market Signaling." The Quarterly Journal of Economics. 1973.

[9] Ministry of Industry and Information Technology (China). Industrial internet development data. January 2026.

About the Author

Nicholas Brennan is a long-term observer and writer in the field of fintech. Over the past decade, his work has focused on global payment systems, digital currencies, and the modernization of bank core systems. He is skilled at translating complex underlying technical logic into clear business narratives. He has served as a technical and strategic advisor at several international financial institutions and consulting firms. Currently, he mainly writes in-depth analyses for industry publications, tracking how financial infrastructure is evolving globally.