Chapter 9

Where the loan capital comes from

FinVolution facilitates loans it mostly does not fund itself; the money belongs to banks, consumer-finance companies and trusts it introduces borrowers to. That supply side is the half of the platform the rest of this report has not examined, and it carries its own durability test. FinVolution has already rebuilt its entire funding base once, after regulators legislated its original source out of existence, and it now draws on 115 Chinese and 18 overseas institutions on non-exclusive terms at a cost that has been falling. The counter-case is that none of those partners is contractually locked in.

Source: cumulative partner counts and non-exclusivity from FY2025 Form 20-F [1]; leverage from the Q3 2025 earnings call [2]; China take rate from the Q2 2025 earnings call [3].

From retail money to institutional money

FinVolution began in June 2007 as an online consumer-finance platform, later operating as PPDAI [4]. For its first decade the loans were funded by individual investors — retail savers lending through the marketplace, the peer-to-peer model that defined the Chinese online-lending boom and then its clean-up. That funding source was removed by regulation: FinVolution ceased facilitating new loans with individual-investor money in October 2019, and stopped funding through third-party online-lending intermediaries the following month [5]. The business (Business and Discount) took the FinVolution name in the same window.

What matters for the funding question is what came next. Since 2020, every new loan the company facilitates in China has been funded by institutional partners or its own licensed micro-loan companies, not by retail money [6]. The platform kept originating at scale through the switch: China loan volume was ¥186.4 billion in 2023, ¥196.1 billion in 2024 and ¥186.3 billion (US$26.6 billion) in 2025 [6]. A platform that can lose 100% of its funding source and rebuild volume within a year has demonstrated what a facilitation model most needs to prove — that its value to borrowers survives a change in who holds the paper. For a reader who will not own a business that could go bankrupt, that history is the most direct evidence the funding side is not a single point of failure.

A diversified, non-exclusive partner base

The rebuilt base is broad and still widening. Cumulative Chinese funding partners rose from 88 institutions by the third quarter of 2023 to 110 at the end of 2024 and 115 by the end of 2025; overseas partners reached 18 [7] [8] [1].

Source: Q3 2023 count from the Q3 2023 earnings call [7]; year-end 2024 from the FY2024 Form 20-F [8]; Q1 2025 from the Q1 2025 earnings call [9]; year-end 2025 from the FY2025 Form 20-F [6].

The partners span commercial banks, internet and private banks, consumer-finance companies, micro-loan companies and trust-management companies [1]. Breadth cuts the risk that any one institution's withdrawal strands the platform, and FinVolution does not disclose — nor, on these numbers, does it appear to run — a dependence on a single dominant funder.

The offsetting fact sits in the same filing: the cooperation is not exclusive [1]. A partner that finds a cheaper or lower-risk channel can move borrowers there, and the same non-exclusivity that lets FinVolution add 27 partners in two years lets any of them leave. Diversification lowers the cost of one departure; it does not create a contractual lock. The company's stated defence is commercial rather than legal — offering attractive risk-adjusted returns and integration tools to keep partners engaged [8].

Falling funding cost is doing the work in the take rate

The economics of a facilitated loan sit inside a fixed ceiling. FinVolution's average China borrowing rate has held near 22% — 22.2% in the second quarter of 2024 — because the industry operates under interest-rate caps that leave little room to charge borrowers more [10]. Within that ceiling, the company's revenue take rate — what it keeps after funding cost, credit cost and expenses — has climbed from 3.1% in the second quarter of 2024 to 3.4% by mid-2025 [10] [3].

Source: Q2 2024 take rate from the Q2 2024 earnings call [10]; Q4 2024 to Q2 2025 from the Q1 2025 and Q2 2025 earnings calls [9] [3].

With the borrower rate fixed, that improvement came from the funding side, not from the customer. Funding cost fell 90 basis points in the second quarter of 2024 and continued in roughly 10-basis-point steps through 2025 [10] [9]. Management's explanation is that lenders re-price the platform's assets as they accumulate evidence of quality: as more institutions recognise FinVolution's asset quality, they lend to it more cheaply [11]. That is a real advantage of scale and track record, but a modest one — the gains are measured in basis points and the take rate has plateaued at 3.4%. The lever cuts both ways: in a tighter liquidity environment the same partners would demand a higher return, and with the borrower rate capped, that cost would land on FinVolution's take rate rather than the borrower's bill [12].

Two funding models, and a book that keeps shrinking against equity

The 115 partners are funded under two different arrangements, and the distinction governs how much risk sits on FinVolution's own balance sheet. For a substantial majority of the loans, FinVolution provides a quality-assurance commitment — it guarantees repayment to the funding partner and therefore keeps the credit risk, the mechanism dissected in Guarantee Economics [8]. Alongside it runs a capital-light model, launched in 2020, under which FinVolution acts purely as a technology enabler, earns a service fee set as a percentage of the partner's loan pricing, and assumes no credit risk at all [13].

The mix has been shifting toward the lighter model, and the company's own leverage measure shows it. FinVolution tracks a leverage ratio defined as risk-bearing loans divided by shareholders' equity; it fell from 4.3x in the second quarter of 2023 to 2.7x by the first quarter of 2025 and to a stated historical low of 2.4x by the third quarter of 2025, alongside a 570% provision-coverage ratio and ¥7 billion of cash and short-term investments [14] [15] [2].

Source: Q2 2023 leverage from the Q2 2023 earnings call [14]; Q1 2025 from the Q1 2025 earnings call [15]; Q3 2025 from the Q3 2025 earnings call [2].

A falling ratio means the risk-bearing book is shrinking relative to the equity standing behind it — the balance sheet is carrying less of the platform's volume, not more. That is consistent with the deleveraging read the closing chapter reaches for the solvency question, and it is the mechanical reason the platform can keep growing facilitated volume while the on-balance-sheet risk recedes.

The regulatory frame around the funding relationship

The relationship between platform and funder is itself regulated, and the rules moved in 2025. On 1 April 2025 China's financial regulator issued a notice tightening the management of commercial banks' internet loan-facilitation business, effective 1 October 2025; management described it as the first regulation to clearly define the loan-facilitation model — loans issued by licensed financial institutions through external third-party platforms [15]. The near-term response was defensive: in the fourth quarter of 2025 FinVolution prioritised risk control over origination, tightening underwriting and accepting a moderation in China loan volume [16].

A longer-standing rule cuts closer to the guarantee-based half of the model. Circular 141 bars banks, trusts and consumer-finance companies participating in loan facilitation from accepting credit-enhancement services — including commitments to assume default risk — from third parties [17]. FinVolution channels its quality-assurance commitment through licensed financing-guarantee subsidiaries to stay inside such restrictions, but the direction of regulation is toward separating the funder's credit decision from the platform's guarantee. Taken to its conclusion, that pressure pushes the model toward the pure capital-light service-fee arrangement — lower risk, but also a lower take rate than the guarantee model earns.

The weight of the evidence is that FinVolution's funding base is more resilient than a value-trap reading of the discount assumes: it survived the total loss of its original source, rebuilt into a broad, non-exclusive, cheapening institutional pool, and now runs its risk-bearing book at a record-low multiple of equity. The strongest fact against that read is that the resilience is commercial, not contractual — no partner is locked in, and the October 2025 re-regulation of bank loan-facilitation reshapes the very relationship the model depends on. What would change the read is direct and observable in the quarterly disclosure: partner attrition, or a reversal in funding cost showing up as a falling take rate, would signal the supply side turning from tailwind to constraint.