How Fintech Startups Get Funded

Fintech founders routinely raise a seed round, launch a lending product, and discover within two quarters that the round is gone — not spent on salaries, but lent out. They have funded a loan book with equity capital, which is roughly the most expensive money in existence.

The lesson underneath is the defining feature of the sector: many fintech companies need two entirely separate capital structures, raised from different people, on different terms, for different purposes.

This guide covers those two stacks, how licensing strategy is really a financing decision, the unit economics investors actually test, what diligence looks like in a regulated business, and how the milestones sequence.

1. Two Kinds of Capital

Corporate equity

Funds the company: engineering, compliance, sales, marketing, licences, regulatory capital. Raised from venture investors on conventional terms. This is the money most founders think about.

Balance sheet or facility capital

Funds the assets: the loans you originate, the advances you make, the float you carry. Raised from credit funds, banks and specialist lenders, priced on the quality of the asset rather than on your growth story.

The critical insight is that equity is by far the most expensive capital available, and using it to fund an asset that yields a predictable return is enormous value destruction. If your loan book earns a spread, that book should be financed with debt costing a fraction of what equity costs, with the equity reserved for building the company.

Companies that do not need a balance sheet — payments infrastructure, financial software, brokerage technology, compliance tooling — raise like ordinary software companies and can skip most of this. Companies that carry assets cannot.

2. How Facility Capital Actually Works

The progression is fairly standard, and each stage is a genuine milestone.

• Founder or friendly capital. The first loans are funded from anywhere available, purely to generate data.
• A small warehouse facility. A credit fund provides a line secured against the loans, with you contributing first-loss equity — typically a meaningful percentage of each loan. Expensive, restrictive, and usually the hardest facility you will ever raise, because you have no performance history.
• A larger warehouse, once you have twelve to eighteen months of cohort performance. Better pricing, higher advance rate, more capacity.
• Forward flow. A buyer commits in advance to purchase loans meeting agreed criteria. This is the transformational structure: you originate and sell, so you are no longer using your own capital to hold assets and your growth is not constrained by your balance sheet.
• Securitisation, at scale, for the cheapest cost of funds.

Terms to understand before signing anything:

• Advance rate — how much of each loan the facility funds. The remainder is your equity, and it directly determines how much equity your growth consumes.
• Cost of funds — the benchmark plus spread, which must sit comfortably below your yield.
• Eligibility criteria — which loans qualify. Loans that fall outside sit on your own balance sheet.
• Concentration limits by borrower, geography, channel and vintage.
• Performance triggers. If delinquency exceeds a threshold, the facility can stop funding new loans and begin amortising. This is the clause that kills lending companies, because it fires exactly when you can least withstand it.
• Minimum utilisation and unused-line fees, which make an oversized facility costly.

The dynamics have a great deal in common with venture debt covenants — the covenant package matters more than the headline rate.

3. Licensing Strategy Is a Financing Decision

How you get permission to operate determines your cost structure, your timeline and your risk profile.

Partner or sponsor bank model

You build the product and customer experience; a chartered bank provides the regulated capability. Fast to launch, low capital requirement, and the standard route for card issuing, deposit products and much lending.

The costs: revenue sharing, the bank's compliance oversight over your programme, and concentration risk. Your entire business depends on one partner. Regulatory scrutiny of bank-fintech arrangements has intensified in recent years, and programmes have been wound down at the bank's initiative. Having a second relationship, or a credible path to one, is genuine risk management and investors ask about it.

Owning licences

Money transmitter licences state by state, lending licences, broker-dealer registration, or ultimately a bank charter. Slow — multi-state licensing is measured in years — capital-intensive, and operationally demanding.

The benefit is removing the dependency, capturing more economics, and building something that is itself a barrier to entry.

Most companies start on a partner model and migrate selectively. Investors want to see that you have a considered view rather than a default.

4. The Unit Economics Investors Test

Which metrics matter depends entirely on the model.

Lending

• Cohort loss curves. The single most important artefact in the entire diligence. Losses by origination vintage, plotted over months on book. Investors compare vintages to see whether underwriting is improving or deteriorating as you scale.
• Net interest margin after cost of funds
• Contribution margin per loan after acquisition, servicing and losses
• Payback per cohort, and how sensitive it is to a downturn

Rapid origination growth with worsening vintages is a worse business than slower growth with stable ones, and experienced credit investors identify this immediately. Present the cohort curves yourself.

Payments

• Take rate and, more importantly, net revenue after interchange and network costs
• Gross margin after processing and fraud losses
• Volume retention and merchant churn

Neobanking and deposits

• Cost to acquire a funded account, not a signup
• Deposit balances per user and their stability
• Revenue per user, and how much depends on interchange versus interest
• Interest rate sensitivity — businesses earning primarily on deposit spread have revenue that moves with rates and no control over it

Everywhere

• Fraud losses, presented honestly, with the trend
• Compliance cost per transaction or account, which does not scale as gracefully as software costs

5. What Diligence Includes That Software Diligence Does Not

• Regulatory posture. Which licences you hold, which you rely on a partner for, what your examination history looks like, and any regulatory correspondence.
• The compliance function. Is there a real compliance officer with authority, or a founder with a checklist? Investors ask to meet them.
• BSA and anti-money-laundering programme. Written policies, transaction monitoring, sanctions screening, suspicious activity reporting. Weakness here is a deal-stopper, not a negotiating point.
• The partner bank agreement. Term, termination rights, exclusivity, economics, and what happens to your customers if it ends.
• Consumer protection compliance — disclosure requirements, fair lending, complaint handling, and marketing review.
• Data security and privacy, including the specific requirements applying to financial data.
• Model governance, if underwriting uses machine learning. Explainability and fair lending analysis are expected, not optional — an issue shared with the broader questions in our guide to AI company diligence.

All of this belongs in your data room from the start. Assembling it under time pressure signals exactly the wrong thing about a regulated business.

6. How the Milestones Sequence

Fintech rounds are gated on specific proof points rather than on revenue growth alone, and knowing which one you are working toward determines what you should be spending money on.

Pre-seed — can you legally operate? The proof point is a regulatory path: a signed or credible partner bank relationship, or a licensing plan with counsel behind it. Investors at this stage are underwriting whether the product can exist at all. Founders who raise on product vision with no regulatory analysis are frequently surprised by how quickly the conversation turns to it.

Seed — does anyone want it, and can you originate? Live customers, real transactions, and the first small volume of assets on book. The purpose of this stage is not scale; it is producing the earliest cohort data. Deliberately small, controlled origination with clean records is worth more here than volume you cannot explain.

Series A — do the unit economics work, and can you fund them? Twelve or more months of cohort performance showing losses stabilising, contribution margin positive after acquisition and servicing, and — critically — a first warehouse facility in place. That facility is frequently the actual gate: an investor is underwriting your ability to fund growth without their equity, and a company with strong metrics and no credit relationship is a harder Series A than the reverse.

Series B — does it scale without breaking? Multiple vintages performing consistently, a larger facility at improved advance rates, a compliance function with genuine authority, and evidence that acquisition cost has not deteriorated as you scaled. This is also where investors test concentration — one partner bank, one funding source or one acquisition channel all become live concerns.

Growth — can you fund cheaply and survive a cycle? Forward flow arrangements or securitisation, a materially lower cost of funds, and cohort data through some period of stress. The question at this stage is no longer whether the model works but what happens to it when unemployment rises or rates move, and the answer has to be shown in data rather than asserted.

The sequencing point that matters most: your facility milestones and your equity milestones are on different clocks and both need managing. A first warehouse takes months to negotiate and requires performance history you have to create deliberately. Founders who begin those conversations when they need capacity find the Series A gate closed for a reason they could have anticipated a year earlier.

7. Who Funds Fintech

• Specialist fintech venture funds, which understand the two-stack structure and will not be surprised by your capital plan
• Generalist venture funds, which will fund the equity but may not grasp the facility side — be prepared to explain it
• Bank and financial institution corporate venture arms, which bring distribution and credibility, and sometimes a partner relationship. Watch for exclusivity terms.
• Credit funds and specialty lenders for the facility side. Entirely different institutions, different diligence, different relationships — and worth building well before you need capacity. Our guide to private credit covers this lender universe.
• Payment networks and processors, which invest strategically in the ecosystem

8. Practical Sequencing

1. Decide the licensing path early. It determines timeline, cost and what you can promise investors.
2. Raise equity for the company, not the assets. Say this explicitly in the round — investors respect a founder who has separated the two.
3. Start facility conversations far earlier than feels necessary. Credit relationships take months and your first facility is the hardest.
4. Generate cohort data deliberately. Small, controlled origination that produces clean performance history is worth more than large volume you cannot explain.
5. Build compliance before you need it. Retrofitting it after a regulatory question is far more expensive.
6. Model the downturn. Investors will ask what happens to your cohorts and your covenants in a recession. Have the answer.

Frequently Asked Questions

Can we launch without any licences?

Frequently yes, through a partner bank or a licensed processor. What you cannot do is assume the question does not apply to you. Get a regulatory analysis from specialist counsel before launch, not after — our guide to choosing counsel covers finding the right kind.

How much equity does a lending business consume?

It depends almost entirely on your advance rate. At an 80% advance you fund 20% of every loan from your own capital, so growth consumes equity proportionally. Improving the advance rate, or moving to forward flow where a buyer takes the asset, is what breaks the constraint — and it is why facility terms matter as much as your equity round.

Do fintechs get valued like software companies?

Only when they look like software companies. Infrastructure and SaaS-model fintechs earn software multiples. Balance-sheet lenders are frequently valued closer to specialty finance, on book value and return on equity. Knowing which comparison an investor is applying explains most valuation disagreements in the sector.

What about crypto and digital assets?

Adjacent but a distinct regulatory perimeter, with its own licensing, custody and securities questions that vary by jurisdiction and asset. Investor appetite is more cyclical, and some funds are prohibited from the category entirely by their LP agreements.

Is embedded finance a better model?

Distributing through partners who already own the customer relationship lowers acquisition cost substantially, which is genuinely attractive. The trade-offs are revenue sharing, concentration in a small number of partners, and less control over the customer experience. It is a good model with a different risk profile, not a free one.

What happens if our partner bank ends the relationship?

It is disruptive and it happens, sometimes for reasons entirely unconnected to your conduct — a change in the bank's risk appetite, a regulatory finding against their programme generally, or a strategic exit from sponsorship. Read the termination provisions and the wind-down period before you sign, and understand what happens to open accounts, in-flight transactions and customer data. The practical mitigation is a second relationship in progress before you need it, which is slow to establish and is exactly why investors ask about it.

How much regulatory capital do we actually need?

It depends entirely on which licences you hold. Money transmitter licensing carries net worth and surety bond requirements that vary by state and accumulate as you add jurisdictions; lending licences and broker-dealer registration have their own. This is real cash that must sit unused, and it is a genuine line in a funding plan that founders on partner-bank models frequently forget when they later decide to own licences. Get the numbers from counsel before committing to a licensing path.

Do we need a compliance officer before we have revenue?

Earlier than most founders want, yes. A named compliance officer with genuine authority is what a partner bank requires, what a regulator expects, and what an investor asks to meet in diligence. It does not have to be full-time at the earliest stage — fractional and outsourced compliance officers exist and are widely used — but it does have to be a real person with a real mandate rather than a founder wearing a third hat.

What breaks fintech companies most often?

Three things, in roughly this order. Cohort deterioration that triggers a facility performance covenant, cutting off funding precisely when you need it. Loss of a partner bank with no alternative in place. And acquisition costs rising faster than lifetime value as the easy customers are exhausted, which is a slow failure rather than a sudden one. All three are visible in the data months before they become critical, which is the argument for cohort discipline and honest reporting to your board.

The Bottom Line

If you carry assets, raise two kinds of capital and never confuse them. Decide the licensing path deliberately, because it sets your cost structure for years. And present cohort loss curves before anyone asks — in this sector, that single artefact carries more weight than the rest of the deck.

Global Capital Network connects fintech founders with venture and credit investors across our network and events. Get in touch.

This article is general information, not legal or financial advice. Financial services regulation is complex and jurisdiction-specific. Engage specialist counsel before launching a regulated product.