Author: Mario Stefanidis, Head of Research at Artemis Analytics; Source: Artemis; Compilation: Shaw Golden Finance

Introduction

According to data from the International Finance Institute (IIF), the global debt market reached a record high of $348 trillion by the end of 2025. Of this, government debt is about $107 trillion, corporate debt $101 trillion, household debt $65 trillion, and financial-sector debt $76 trillion. The share of digital and financial technology lending platforms in total debt ranges between $590 billion and $680 billion—equivalent to less than 0.2%.

This is the largest credit market in human history, and it still runs on infrastructure designed decades ago (FICO launched in 1989, MERS went live in 1995). According to data from the Mortgage Bankers Association, the average cost to originate a single mortgage loan in the U.S. is about $11k. Despite huge technological progress and the widespread adoption of AI, this cost remains twice what it was in the early 2010s.

Source: Freddie Mac

The clearing and settlement of a standard wire transfer still takes about 28 hours, while most banks’ credit-approval decisions still rely on a committee process and depend on a black-box scoring model built from 20 to 30 variables. These are all public facts. But less obvious is: in what way are solutions actually being implemented?

The credit industry is not being reshaped by a Silicon Valley-style romantic disruption model—no startup can simply replace systemically important global banks like JPMorgan Chase in one fell swoop. The real change is more subtle and more structural: the end-to-end credit workflow that was once vertically integrated within banks—loan origination, distribution, risk review, funding, and underlying infrastructure— is being dismantled into a horizontal, modular architecture, with each stage controlled by specialized institutions.

This architectural shift mirrors the transformation in cloud computing from monolithic systems to microservices, and the media industry’s shift from the studio model to streaming and creator ecosystems. Now, this transformation has finally arrived in credit.

In this wave of re-integration, the winners are not the largest institutions by balance sheet size, but the core-layer companies that occupy key bottlenecks—positions that other participants cannot bypass. There are two categories whose importance far exceeds the rest: first, the intelligent decision layer, where AI-driven risk review and risk scoring determine where capital flows and the terms of lending; second, the clearing-and-settlement channel layer, where blockchain infrastructure is compressing loan origination costs and settlement time by orders of magnitude.

As long as you occupy these two “water-seller” style core positions, other lenders will pay you usage fees. If neither is occupied, then you can only fight a price war in a commoditized market—where $3.5 trillion of private credit capital is already chasing yield.

Here, Artemis maps 40 companies across 15 sub-segments, classifying them into five tiers to analyze where structural value is consolidating.

The five tiers of the new credit architecture

First tier: Loan origination

The loan origination layer is the source of credit business, covering consumer loans, mortgages, small-and-micro business loans, as well as loans collateralized by crypto assets. This area is also becoming increasingly commoditized. Nowadays, having the ability to originate loans is no longer a competitive moat—just the basic entry requirement. The key factor that separates winners from other participants lies in loan origination costs and approval rates.

SoFi, valued at about $24 billion, and the rocket company (rocket mortgage lender), with a market cap of $48 billion, both have large-scale loan origination operations. But the core logic of their profits is how to complete lending at lower cost. Figure, with a market cap of $6 billion, relies on its Provenance blockchain-native origination of home equity credit lines (HELOCs) and first-lien mortgages, removing the multi-layer intermediary steps that make traditional mortgage origination slow and costly.

In the crypto space, Aave (market cap $2.7 billion) and MakerDAO/Sky (market cap $1.6 billion) completely blur the boundary between fintech and decentralized finance (DeFi) in the loan-origination layer.

Second tier: Channel distribution

The distribution layer is the demand aggregation stage, where embedded finance and “buy now, pay later” (BNPL) are reshaping this space. The embedded finance market is expected to grow from $156 billion in 2026 to $454 billion by 2031, a CAGR of 24%. The BNPL model is expected to cover 13% of digital transactions, up sharply from 6% in 2021.

Affirm ($15 billion market cap) and Klarna ($5 billion) are well-known companies in the industry, but the real structural trend is this: credit servicing has been deeply embedded into the checkout flow, software platforms, and merchants’ consumer experiences. Even though both companies’ stock prices have fallen significantly from historical highs, they are not “water-seller” type businesses that can win mainstream market share. Lenders that borrowers can’t even perceive are often the ultimate winners.

Today, major software companies are adding financial products. Shopify, Amazon, Square, and Stripe all need API infrastructure layers—and institutions providing these kinds of services will extract fees from the scale of each new transaction.

Third tier: Risk review and risk pricing

This is the first core stage in the entire credit architecture. Whoever controls borrowers’ credit scoring controls how the profits across the entire credit value chain are distributed.

Currently, the credit bureau and credit reporting field in lending is dominated by an oligopoly of three giants: Experian, TransUnion, and Equifax. Together, the three generate about $18 billion in annual revenue by scoring borrowers using 20–30 variables.

AI risk models can evaluate more than 1,600 variables (data from Upstart). Upstart’s disclosed data also shows that, while maintaining the same bad-loan rate as traditional models, its approval volume increases by 44%, default rate drops by 53%, and the annual percentage rate (APR) decreases by 36%. With mortgage rates surging to nearly 7% today, every basis point matters greatly for first-time homebuyers.

As of now, 92% of Upstart’s lending decisions are fully automated, with approvals completed within minutes, while traditional risk reviews take 3 to 5 days. The U.S. Consumer Financial Protection Bureau (CFPB) is pushing for alternative FICO scores that are less discriminatory. The EU’s AI Act also lists credit scoring as a high-risk scenario, requiring explainability. These regulatory shifts all benefit explainable machine-learning models, which are more advantageous than traditional credit bureaus using black-box models.

The value of this tier is extremely high, because whoever controls the scoring engine controls the profit curve across the entire upstream-to-downstream chain. But at the same time, the moat in this area still needs ongoing validation—rapid advances in AI mean that, with sufficient resources and time, “any institution” can build scoring models.

Fourth tier: Capital and funding supply

In the post-pandemic era, overall capital is abundant. Although the current environment is challenging, the scale of private credit management has expanded to $3.5 trillion, and Morgan Stanley expects it to reach $5 trillion by 2029. The total value locked (TVL) in DeFi lending protocols ranges from $5 billion to $78 billion, accounting for about half of all DeFi activity. The scale of non-tradable perpetual assets (NPE) has grown from zero growth in 2021 to over $200 billion.

In a world where capital is abundant, the most core capability is intelligent allocation—directing where capital flows. Therefore, despite the enormous size of the funding layer, its structural position still comes after the intelligent decision layer above and the infrastructure layer below.

Private credit institutions such as Ares, Blue Owl, and Golub are important allocators of capital, but they heavily rely on upstream scoring systems and downstream clearing channels to execute efficient lending. In DeFi, Ape has an absolute dominant position in liquidity, holding more than half of the lending market; while protocols such as Maker, Morpho, Maple, and Kamino fight for the remaining market share.

Fifth tier: Infrastructure

Infrastructure is the second core stage in the entire architecture. Whoever controls financial licenses or clearing-and-settlement channels has everyone paying “tolls”. According to management disclosures, the bank license held by SoFi lowers its cost of funds by 170 basis points, reducing annualized interest expense by more than $500 million. Figure, relying on its Provenance blockchain, has processed more than $50 billion in total transaction volume; the cost to originate a single loan is below $1,000, while the average cost of traditional channels is about $11k. Final settlement confirmation on blockchain takes only seconds, while traditional wire transfers take about 28 hours.

SoFi’s Galileo and Technisys technology stack, along with platforms such as Blend Labs, form the remaining lending-as-a-service (LaaS) underlying technical support. Cross River Bank, as an “invisible” correspondent bank behind dozens of fintech companies, has already facilitated the issuance of more than 96 million loans in partnership, with a total value exceeding $140 billion.

Companies that can win long-term either occupy a key bottleneck and become indispensable to all participants, or vertically integrate multiple tiers to form compound competitive advantages. And companies that lose will be trapped in commoditized business layers, lacking structural voice and bargaining power—leaving only price competition until profits approach zero.

Winners: Core-tier controllers and multi-tier compound-advantage companies

SoFi: An all-stack compound tool

SoFi is the only company that covers four of the five tiers:

• Directly originates consumer loans and mortgage loans.

• Through the Galileo platform, exports lending infrastructure to third parties, supporting about 160 million activated accounts.

• Conducts loan underwriting using its in-house risk-control models, with core evaluation dimensions including repayment willingness, repayment capacity, and stability.

• Holds a bank license, and in the infrastructure layer has Galileo and Technisys’ core banking technology stack.

SoFi set a historical revenue record of $3.6 billion in 2025, up 38%. The platform has 13.7 million members and a $20.2 billion scale of financial products. Management guidance expects revenue of $4.7 billion in 2026, and EBITDA of $1.6 billion. This business not only has strong revenue growth, but also excellent profitability, with a profit margin of 34%. Just the bank license alone enables SoFi to finance loans through deposits rather than the wholesale market, directly lowering funding costs by 170 basis points.

SoFi is building the “Amazon Web Services (AWS)” of lending— a platform that both competes with other lenders and empowers them. Galileo itself has already been built into a billion-dollar revenue engine. Technisys, acquired for $1.1 billion in 2022, provides the core banking systems layer for third-party institutions. A bank license is a structural moat that most fintech lenders cannot replicate. Although the industry has rushed to imitate it: the U.S. Office of the Comptroller of the Currency (OCC) received 14 applications for new bank licenses in 2025 alone, signaling that the battle over the infrastructure layer is accelerating.

Upstart and Pagaya: The intelligent decision layer

Ironically, winning in lending may not require doing the lending business yourself. Both Upstart and Pagaya center around a risk-review engine; their risk-control performance is better than the in-house models of lending institutions, and they do not need to run the business on their own balance sheets. This is exactly the “water-seller” logic applied to the credit decision space.

Compared with traditional FICO-based risk models, Upstart’s model can approve 44% more borrowers at the same bad-loan rate, reduce the default rate by 53%, and offer borrowers a significantly lower annual percentage rate. On the platform, nearly all new loan origination is now fully automated, greatly reducing human intervention. This differs fundamentally from traditional consumer credit risk-control models.

Pagaya is in the same track, but faces harsher market realities. The company does not directly originate loans; instead, it authorizes banks to use its AI risk engine. Since its founding in 2016, Pagaya has assessed loan applications totaling about $3.48M across 31 partner banks. Its structural positioning is very clear: it doesn’t need borrowers to know the brand—it only needs banks to rely on its scoring system. But the current market has not validated this logic. In Q4 2025, online business volume grew only 3% year over year; revenue missed market consensus expectations, and forward guidance also came in below expectations. The stock price plunged by nearly a quarter in a single day. The value of the intelligent decision layer is entirely constrained by the credit cycle; when bad-loan rates rise across the partner network, even excellent AI cannot withstand the pressure of deteriorating asset quality.

But the core logic still holds: FICO forms single-slice scores based only on a small set of historical variables. As consumers’ financial situations become more complex and diversified, AI risk-control systems will become increasingly critical. Unlike FICO, these systems continuously learn and optimize every time they complete a scoring.

Figure: The next-generation clearing and settlement channel

By originating a single loan through traditional channels and the Mortgage Electronic Registration System (MERS), the cost is $11,000. With Figure’s technology stack, which includes the Provenance blockchain and the DART system, that cost can be reduced to $717. This type of new-channel infrastructure enables loan costs to drop by orders of magnitude.

Figure has used the Provenance blockchain to originate more than $1.07M in home equity products (primarily home equity lines of credit). On-chain, it has processed a total transaction value exceeding $50 billion. In Q4 2025, loan origination volume reached $2.7 billion, up 131% year over year. The company holds more than 180 lending licenses and U.S. SEC-registered broker-dealer qualifications, providing a compliant foundation for scaled operations. It also has more than 300 white-label lending partners; since filing its S-1 for listing last September, it has been adding partners at a pace of one per day on average. Its revenue grew from $28.5 million annualized in Q1 2023 to $146.8 million today.

Figure’s core business is not closely tied to crypto assets, but its stock price movement is highly similar to Bitcoin’s. Its settlement system reflects the logic of restructuring cost structures: final settlement confirmation takes only seconds, whereas traditional methods take more than a day; origination cost is only a fraction of the traditional model. Across the entire loan lifecycle, costs related to securitization are saved by over 100 basis points— in an annual securitization market of $3 trillion, this implies potential cost reductions of more than $30 billion.

Aave: The core controller in the DeFi space

Aave holds more than half of the share in the DeFi lending market. Liquidity breeds more liquidity: borrowers continually cluster toward the platform with the deepest liquidity pool (network effects). Its cumulative loan origination volume has exceeded $1 trillion, and the protocol officially crossed the $1 trillion mark in cumulative lending just last month.

Besides its dominance in DeFi, the most compelling part of Aave structurally is its institutional lending business line, Horizon. Horizon has attracted $580 million in deposits, aiming to surpass $1 billion by 2026. It acts as a bridge connecting DeFi liquidity with traditional credit demand. If Aave can bring on-chain funds into institutional-grade lending products, it would become the capital supply layer for traditional lenders, unlocking a potential total addressable market (TAM) far larger than the retail DeFi market.

DeFi lending also has an underappreciated structural risk advantage. In DeFi, the typical overcollateralization ratio is usually between 150%–180%, whereas traditional peer-to-peer lending is only 50%–70%. In DeFi, bad loans mainly come from oracles or technical failures, not from creditworthiness defaults.

Affirm: Channel entrenchment

Affirm has established a leading position in the buy now, pay later (BNPL) sector by deeply embedding merchant payment settlement infrastructure. Critics focus on its consumer credit risk, but overlook the core structural logic: Affirm is not a consumer-lending institution in the traditional sense, but a credit distribution channel for the point of sale. Its moat is integration with merchants’ systems. Given that BNPL is expected to cover 13% of all digital transactions, large-scale platforms embedded into checkout flows will collect structural “channel fees” from the commercial transactions themselves.

The losing pattern: four structural failure modes

We deliberately do not name the companies that fit these patterns. If you are an investor or operator in the credit space, you naturally know who they are. More important than specific names is understanding why these structural positions are destined to fail—because in the next cycle, the same patterns will produce new casualties.

Loan lenders focused only on the balance sheet

The only competitive advantage of these businesses is that they can obtain funding. They issue loans using traditional risk models, provide funding with their own balance sheets, and have no dedicated technology layer. They are simply “mindless pipes” for capital.

In a world where private credit management has already reached $3.5 trillion and is moving toward $5 trillion, capital is not scarce—the scarce resources are intelligent decision-making and infrastructure. These businesses can only compete on price, compressing profits to zero in each interest-rate cycle, while forcing them to take excessive risk. In the end, these lenders will extend credit to high-risk borrowers and suffer losses when the cycle turns.

These participants are often traditional consumer-lending companies, small-scale banks, and fintech lenders that have never built a technology moat beyond their initial loan products. When capital becomes commoditized, and when there is no technology advantage and lending relies solely on their own balance sheets, it is tantamount to slowly handing shareholders’ equity over to borrowers.

CeFi lending casualties

The centralized crypto lending platforms (CeFi) that collapsed dramatically in 2022 were not victims of the bear market. They fell due to the oldest failure mode in the credit industry: maturity mismatch, misuse of customer funds, lending collateralized by non-liquid assets, and lack of transparent risk management.

Decentralized finance (DeFi) protocols that automatically enforce collateral discipline through smart contracts and make on-chain collateral ratios visible did not blow up. What actually went wrong was those CeFi platforms that relied on human judgment and had opaque balance sheets. Any lending platform—whether in crypto or traditional finance—if it only asks you to trust its balance sheet, but does not show you the collateral, is simply repeating the same structural path that has already failed.

Ghost protocols

There is a class of DeFi lending protocols that are still alive technically but dead structurally. After going live, they attracted initial locked capital through token incentives, but once those incentives faded, they stalled. The code can run, and total value locked (TVL) is not zero, but utilization curves are flat or continue to decline, and there is no clear path for natural demand growth.

The reason is that DeFi lending exhibits extreme power-law distribution characteristics: liquidity concentrates in platforms with network effects—Aave’s absolute dominance of market share is proof. Protocols that cannot break through critical scale end up in a structural “no-man’s land”: too small to attract natural liquidity and complementary integrations; but not so small that they can shut down decently. As profit-seeking capital flows to the top platforms, their locked-value slowly bleeds away and the process is irreversible. These are zombie protocols that can only barely be sustained by the sunk costs of governance tokens.

Lenders that missed the platformization transition

Some companies built strong loan origination capabilities in the previous cycle, but never developed platform capabilities. They have no API distribution channels, no embedded-finance partnerships, and no licensing model for technology. They can originate loans extremely well, but cannot export capabilities outward.

As the credit industry moves toward modularization, whether you can become a component within someone else’s system is just as important as directly originating loans. Companies that can only lend directly to end borrowers will have growth constrained by the coverage of their own channels. In contrast, companies that can provide lending capability support to other institutions have an unlimited potential market space (TAM). Pure loan originators often have solid unit economics per customer, but their growth curve is flat because the reachable market is limited to their own brand and channels. In a modular architecture, being an excellent lender is a necessary condition, but being an excellent lender that other lenders can plug into is the real winning position.

Investable opportunities worth watching

The winning companies above have become market consensus or close to consensus, while the following companies are not. They possess structural traits to become controllers of core tiers, but they have not yet been validated at a scalable level. These are targets worth tracking continuously.

Morpho

Morpho’s total value locked (TVL) is already $6.6 billion, up 164%, and its market cap exceeds $800 million. Its structural logic is completely different from Aave’s: Aave is like a commercial bank within decentralized finance (using a unified lending liquidity pool model), while Morpho is building a modular lending layer that lets institutional participants tailor their own lending markets based on their risk parameters, collateral types, and interest-rate models. If the lending system truly moves toward modularization, Morpho will become a lending-as-a-service protocol on-chain.

Maple Finance

In 2025, Maple originated total loans of $11.3 billion, serving 65 active borrowers. Its assets under management (AUM) surged from $516 million to $4.6 billion, a growth of 767%. The company’s goal is to achieve $100 million in annual recurring revenue (ARR) in 2026. Maple is one of the few protocols truly committed to bringing real-world business lending to blockchain infrastructure—connecting institutional credit demand with on-chain capital and settlement systems to run the business. The explosive growth in its AUM indicates that institutional interest in on-chain credit markets is shifting from theoretical concepts to practical execution.

Cross River Bank

Since 2008, Cross River has, through partnerships, issued more than 96 million loans totaling over $140 billion. It is the partner bank behind Affirm, Upstart, and dozens of other fintech lenders. Reports indicate that the bank is preparing for an IPO. Cross River is a “stealth bank” that supports the operation of a significant portion of fintech lending in the infrastructure layer. As the correspondent-bank model matures, the leverage from its market position is something no single fintech lender can replicate. Its key to winning lies in making it so fintech companies cannot proceed with lending without its support.

License battle

The U.S. Office of the Comptroller of the Currency (OCC) received 14 applications for new bank charters in 2025 alone—nearly equal to the total for the previous four years. The total number of charter applications submitted by fintech institutions has hit a historical high of 20. Affirm, Stripe, and Nubank are all actively applying for licenses. These companies view licenses as the core competitive advantage for the end-game of rebuilding the credit business.

Companies that started as technology service providers are now capturing economic value across the entire industry chain by obtaining regulatory credentials. In lending, the position of a bank license is comparable to regional nodes in cloud computing, because:

• Building costs are extremely high;

• Industry participants cannot bypass them;

• Once acquired, it creates a permanent structural advantage.

The business logic is straightforward: for every 1 basis point improvement in the cost of funds, pre-tax return on net assets (ROE) can increase by several percentage points. For scaled enterprises, the advantages brought by a license are significant. But for small and mid-sized institutions, a license can actually become a trap: they must bear all compliance costs, regulatory inspection pressure, and capital requirements, without enough operating scale to cover these expenses. Only companies that already have massive business volumes can make the license into a growth accelerator.

Credit architecture in 2030

If you want to remember one core analytical framework from this article, it is the following three questions. They apply to all lending businesses, whether public or private, or on-chain.

First: What tier does the company occupy? Loan origination and commoditized capital supply are red-ocean tracks, where profit margins will be continuously compressed with industry cycles. AI risk control, blockchain settlement, and bank licenses are core choke points, where value compounds and accumulates. If a company is stuck in a red-ocean track and cannot enter core tiers, no matter how strong the team is, its long-term profitability will be eroded continuously.

Second: Is it a platform or a single product? A single product that serves end borrowers grows linearly with the reach of its own channels. A platform empowers other lenders, and growth depends on the scale of the entire ecosystem, not just its own business. SoFi has attributes of both, while Pagaya is a pure platform-type company. For enterprises that only lend directly to their own customers, growth has a ceiling; platform-type enterprises have no such limitation.

Third: Does it have a regulatory moat? This includes bank licenses, 180 state lending licenses, or programmatic compliance implemented through smart contracts. In the lending industry, regulation is not an extra cost—it is core infrastructure. Companies that recognize this early will build advantages that competitors would need years and huge capital to catch up with.

By 2030, the lending industry will no longer resemble traditional banking so much as the cloud computing industry. A small number of full-stack platforms will cover multiple tiers and create compounded advantages across each stage. The most typical representative in traditional finance is SoFi; in the on-chain space, it is Aave. Around these core platforms, many specialized tier service providers will connect via APIs and on-chain channels, each deeply focusing on specific functions and charging service fees.

In the global $348 trillion debt market, fintech penetration is still below 0.2%. This market is not for dividing between hundreds or thousands of lenders; it will be dominated by a dozen-plus platforms and become the underlying support for the entire industry.