How AI Layoffs and $7 Billion in Square Loans Reshape AI Underwriting for MCA Lenders

Platform Lenders Just Raised the Bar for Everyone

When Square Loans closes out a year with $7 billion funded to merchants and describes its nonperforming loan balance as "immaterial," every MCA lender should pay attention. That level of volume combined with that level of portfolio quality does not happen by accident. It happens because AI underwriting for merchant cash advance has matured from an experiment into an operating system. And as deBanked recently reported, the same company then laid off 40% of its entire staff, citing AI as the reason. That is not a headcount trim. That is a structural transformation.

In the same reporting cycle, Stripe Capital quietly originated 81,000 merchant cash advances and business loans through its subsidiary. Stripe did not even disclose the total dollar volume, but at its reported $159 billion valuation, the lending arm is clearly a strategic priority. Both of these platform lenders share a common advantage: they sit directly on top of the merchant's transaction data, and they have built AI systems that turn that data into instant underwriting decisions.

For independent MCA lenders, brokers, and funders, the message is clear. The competitive floor is rising fast. The question is no longer whether to adopt AI-powered tools for bank verification and underwriting. The question is how quickly you can get there before the gap becomes permanent.

What Square and Stripe's Numbers Actually Teach MCA Lenders

Massive Volume with Minimal Losses Is an AI Story

Square's near-zero nonperforming loan rate across $7 billion in originations is remarkable, but it is not magic. Square has real-time access to every card swipe processed through its point-of-sale network. Its machine learning models analyze daily revenue patterns, seasonal trends, transaction velocity, and dozens of other signals to build a continuously updated risk profile for each merchant. Underwriting is not a one-time event. It is an ongoing, automated process.

Stripe operates similarly. With direct visibility into a merchant's payment processing data, Stripe Capital can underwrite based on actual cash flow rather than self-reported financials or static bank statements. The 81,000 originations in 2025 suggest a highly automated pipeline where human review is the exception, not the rule.

Independent MCA lenders do not have this embedded data advantage. They rely on submitted bank statements, applications, and supporting documents. But this does not mean they cannot achieve comparable underwriting speed and accuracy. The key is building an intake and analysis pipeline that extracts the same caliber of data from the documents merchants provide. That is exactly where AI document extraction speeds up the underwriting process by pulling business info, financials, and owner details automatically from uploaded PDFs.

When AI Replaces 40% of Your Workforce, the Model Has Changed

Block's decision to cut 40% of staff is the most aggressive AI-driven workforce reduction any major fintech has announced. This is not about trimming a back-office team. Block is signaling that AI now handles functions previously requiring thousands of human employees: customer support, risk analysis, compliance review, data entry, and document processing.

For MCA lenders operating with lean teams, the lesson is both a warning and an opportunity. The warning: your competitors, whether platform giants or well-funded ISO shops, are automating aggressively. If your underwriters are still manually keying in bank statement data, calculating average daily balances by hand, or chasing applicants over email for missing documents, you are spending human time on tasks that AI handles in seconds.

The opportunity: you do not need a $159 billion valuation to automate. Purpose-built tools designed for alternative lending workflows can deliver the same efficiency gains at a fraction of the cost. Let's Submit, for example, allows lenders to send a single upload link to applicants, collect all documents in one place, and use AI-powered extraction to parse business information, financials, and owner details automatically. That is the kind of workflow automation that turns a five-person underwriting team into a team that can handle three times the volume.

Embedded Lending Is Growing, but Independent Lenders Still Have an Edge

Platform lenders like Square and Stripe have a structural advantage in data access. But they also have a structural limitation: they can only lend to merchants already on their platform. A restaurant using Clover, a retailer on Shopify, or a service business that does not process cards through Square will never see a Square Loans offer.

Independent MCA lenders serve the entire market. They work with brokers, ISOs, and direct applicants across every vertical and payment processor. The challenge is that this broader market access comes with messier data. Documents arrive via email, fax, broker portals, and phone photos. Bank statements come in dozens of formats from hundreds of institutions. Applications are incomplete, inconsistent, or duplicated.

This is precisely the problem that building a scalable MCA application pipeline solves. When intake is standardized and document parsing is automated, independent lenders can process applications at speeds that approach what platform lenders achieve with embedded data. The playing field does not need to be perfectly level. It just needs to be close enough that your funding speed and approval rates keep brokers sending you deals.

Practical Steps to Implement AI Underwriting in Your MCA Operation

Adopting AI underwriting does not require ripping out your existing systems or hiring a machine learning team. The most impactful changes are incremental, and they start with the intake process.

First, eliminate manual document collection. Every hour your team spends emailing applicants for missing bank statements or re-requesting illegible PDFs is an hour that a platform lender spends funding deals. A secure upload portal, like the one Let's Submit provides, gives applicants a single link where they can drag and drop every required document. Your team gets a complete submission package without the back-and-forth.

Second, automate data extraction from bank statements. This is where the real time savings happen. AI-powered OCR and document classification can parse multi-month bank statements, identify deposits and withdrawals, flag NSF transactions, calculate average daily balances, and detect potential MCA stacking patterns through smarter bank verification. Manual data entry is not just slow; it introduces errors that compound through the underwriting decision.

Third, build a review layer, not a rebuild layer. The best AI underwriting systems in 2026 are not fully autonomous. They extract and organize data, flag anomalies, and present a clean summary for a human underwriter to review. The underwriter's job shifts from data entry to decision-making. This is the model Block is moving toward, and it is the model that keeps portfolio quality high while dramatically increasing throughput.

Fourth, track everything. Platform lenders have comprehensive audit trails built into their systems by default. Independent lenders need the same. Every document uploaded, every data point extracted, every review action taken should be logged. This is not just a compliance requirement. It is the foundation for training better AI models over time. The more structured data you capture now, the smarter your underwriting becomes in the future.

Finally, consider how your workflow integrates with CRM and syndication partners. Extracted data that lives in a PDF on someone's desktop is wasted data. Systems that sync to Salesforce or export structured records allow your sales and operations teams to act on underwriting insights immediately. Let's Submit is building exactly this kind of integration, with Salesforce sync and structured data export designed to keep information flowing downstream.

Frequently Asked Questions

How does AI underwriting work for merchant cash advance?

AI underwriting for merchant cash advance uses machine learning models and document analysis tools to evaluate a merchant's financial health automatically. Instead of an underwriter manually reading bank statements and keying numbers into a spreadsheet, AI-powered OCR extracts transaction data, categorizes deposits and withdrawals, calculates cash flow metrics, and flags risk indicators like NSF fees or declining revenue trends. The extracted data feeds into scoring models that estimate probability of default and expected loss. Human underwriters then review the AI's output and make the final funding decision, spending their time on judgment calls rather than data entry.

Can small MCA lenders compete with Square and Stripe on speed?

Yes, but only with the right infrastructure. Square and Stripe have embedded data advantages because they process merchant payments directly. Independent lenders compensate by automating their document intake and extraction pipeline. When an applicant can upload bank statements, applications, and IDs through a single secure link, and AI parses that data within minutes, the effective turnaround time approaches what platform lenders achieve. The gap is not in data quality; it is in how quickly that data moves from raw documents to actionable underwriting summaries.

What is the biggest risk of not adopting AI in MCA lending?

The biggest risk is losing deal flow to faster competitors. Brokers and ISOs send applications to the funders who respond quickest with competitive offers. If your intake process takes 24 to 48 hours because staff are manually processing documents, while a competitor returns a decision in two hours using automated extraction and scoring, you will see fewer submissions over time. As the Federal Reserve's small business lending research consistently shows, speed of access to capital is one of the top factors driving borrower satisfaction in alternative lending.

Is AI underwriting accurate enough to trust for MCA decisions?

Modern AI extraction tools achieve high accuracy rates on standard bank statement formats, often exceeding 95% on clean documents. The key is building a human-in-the-loop review step where underwriters verify extracted data before making funding decisions. This hybrid approach captures the speed benefits of automation while maintaining the judgment and accountability that regulators and investors expect. Purpose-built AI systems trained specifically on financial documents, rather than general-purpose language models, deliver the most reliable results for lending workflows.

Conclusion

Square's $7 billion year and Block's 40% workforce reduction are not isolated headlines. They are signals that AI-powered underwriting has moved from competitive advantage to baseline requirement. Stripe's 81,000 MCA originations reinforce the same point from a different angle. The platform lenders are scaling with technology, and independent MCA lenders who rely on manual processes will find it increasingly difficult to compete on speed, accuracy, or cost.

The good news is that you do not need to be a $159 billion company to automate. Tools built specifically for MCA workflows can close the gap. Let's Submit gives lenders AI-powered document extraction, secure applicant upload portals, real-time application tracking, and a streamlined review process that turns raw documents into underwriting-ready data in minutes. Visit letssubmit.ca to see how async verification and AI extraction fit into your operation before the next wave of platform competition makes the decision for you.