The Onchain Credit Flywheel

How originators can grow both supply and demand for quality loans.

The proof of concept already exists.

Figure Technologies started with a question: why does it cost $11,000 to originate a home equity line of credit?^[1][2] CEO Mike Cagney’s prior company, SoFi, had shown that digital-first lending could disrupt student loans and personal credit. Figure went further, building proprietary blockchain infrastructure to collapse costs that even the best fintechs couldn’t eliminate. The result is roughly $730 origination costs per loan, a 93% reduction, and a 5-day close versus the industry’s 45.^[2]

Figure has now originated $19B+ in HELOCs through 2025, at 754 average FICO scores and 50% average combined loan-to-value: investment-grade collateral by any standard. The company reports ~55% EBITDA margins on annualized origination volume exceeding $12B,^[2] making it the largest non-bank HELOC provider by a large margin.

The critical insight is that Figure bootstrapped on-chain lending with off-chain demand. Rather than courting DeFi yield farmers, it produced HELOCs so cheaply and quickly that traditional institutional buyers (banks, credit unions, insurance companies, asset managers) were eager to purchase them through conventional channels. The loan production economics were compelling enough to attract capital without needing a crypto-native buyer base.^[3]

Three layers, not one breakthrough. And it’s not all blockchain.

The $10,270 gap between Figure’s ~$730 cost and the industry’s ~$11,000 doesn’t come from a single innovation. It compounds across three distinct layers of technology, and understanding which layers require blockchain is essential to evaluating where this model transfers. Roughly 35–40% of the savings come from standard digitization any fintech could adopt, which tells us that Cagney’s background in digital lending matters nearly as much as the onchain infrastructure. The remaining 50–60% depend on blockchain infrastructure that took years and hundreds of millions of dollars to build.

Cheaper production grows the market.

Let’s break down the unit economics of a $50K HELOC. Assuming an origination fee of 3% (Figure charges up to 4.99%), Figure generates $1,500 in origination fee against a $730 cost base. That’s 51% margin on the origination fee alone. Compare this to a traditional lender’s $11,000 origination cost base, which would put the lender in a ~$9,500 hole.

Figure breaks even at ~$25K loan size; a traditional lender needs ~$367K. That $342K gap defines a market that traditional lenders structurally struggle to serve, but tech-forward lenders can serve profitably.

Figure then monetizes the loan across four events in its lifecycle:

The origination fee alone yields a 73% gross margin on Figure’s average $90K HELOC, and the additional revenue streams (gain-on-sale, warehouse carry, servicing) stack on top. Currently, after G&A and overhead across all revenue lines, the company reports ~55% EBITDA margins at its current $12B+ annual run rate.^[2] However, the disruption is structural and highlights upwards room on margin. Further, cheaper and faster loan production expands the pool of economically viable loans, growing supply into segments traditional lenders cannot reach. Rates stay competitive; the innovation is about growing the market, not just capturing more of it.

Investors get a better product. Demand grows too.

Cheaper, faster loan production is compelling on its own. But credit is a marketplace, and the thesis has a second leg: on-chain infrastructure creates real, quantifiable advantages for the investors buying these loans. Even when borrower rates are competitive rather than dramatically lower, the investor’s net effective yield improves because the structural costs of owning and trading the asset are compressed. Three forces grow investor demand.

Each advantage is individually modest: settlement speed might be worth 15–20bps, intermediary cost reduction contributes another 50–100bps, and transparency effects (harder to quantify, but market proxies suggest 10–30bps) round out the picture. Taken together, the structural improvement lands in the range of 100–200bps of net effective yield, delivering a better product to investors and growing the demand pool for on-chain loans.

The investor math. Equal or better yield, lower structural cost.

Even if borrower rates are lower on-chain, are investor yields lower too? Based on available data, they are not. The investor’s net yield is equal or better on-chain because the structural costs between gross coupon and net return are dramatically compressed.

A few notes on methodology. The gross coupon range for traditional HELOC ABS reflects recent AAA–A tranche pricing for prime collateral.^[4] Figure’s range reflects the Democratized Prime marketplace, where investors bid hourly via Dutch auction and recent clearing rates have been in the 8.5–9% range.^[6] The excess spread line is lower on-chain because Figure’s collateral (754 FICO, 50% CLTV) has shown lower loss rates than blended traditional pools, and on-chain transparency reduces the uncertainty premium.

Quality product first. Then blockchain innovation.

Figure’s model works because it anchors to a quality product and uses blockchain to innovate at the technology level. The contrast with the 2020–2022 crypto credit cycle is instructive.

The early instinct was simple: tokenize debt, sell it to yield-hungry DeFi capital, disintermediate banks. When traditional lenders wouldn’t touch an asset class, crypto would. The result was uncollateralized loans to trading firms and emerging-market consumer debt with thin documentation – textbook adverse selection where the assets that arrived first were the ones traditional capital markets had already rejected.^[9]

Celsius, Voyager, and BlockFi collapsed in 2022 with billions in under-collateralized exposure, aggregate damage exceeding $25 billion.^[9] The lesson is that crypto capital doesn’t change underwriting standards, and marketing to a different buyer base doesn’t reduce your cost of capital.

The quality correction

After the blowups, capital didn’t leave on-chain credit. It moved upmarket. Dramatically.

The composition tells the quality correction story. Treasuries account for almost half of distributed assets. Private credit now sits at $2.8B, mainly due to the success of Maple, an onchain private credit fund. And of course, both of these categories are dwarfed by the ~$300B of tokenized US dollars, more commonly referred to as stablecoins. Note that most data providers exclude Figure from the data here given the instruments live on Figure’s blockchain, Provenance, which is permissioned.

Tokenization more broadly scales by starting with assets where demand is clearest and blockchain infrastructure creates structural advantages or a distribution edge. Stablecoins first. Then treasuries. Then investment-grade and/or secured consumer lending backed by real assets, which would be where Figure sits.

Blockchain cannot make a bad asset good, but it can make a good asset cheaper to produce, faster to trade, and more transparent to own. Figure absorbed that lesson; the last cycle didn’t.

Who does what. And where Figure’s moat lives.

On-chain credit is a stack of specialized infrastructure layers. Figure controls all six through vertical integration, which is a key reason its economics are so strong. Most other participants specialize in one or two layers. The map below shows what each layer does, who operates in it, and where Figure’s ownership creates its moat.

Figure’s advantage comes from controlling all six layers simultaneously. Every dollar of value that would leak to intermediaries in a disaggregated model (title companies, warehouse banks, I-banks, sub-servicers) stays internal. Vertical integration eliminates that leakage, producing 55% EBITDA margins while charging competitive rates.

Outside of Figure, origination, issuance and servicing remain dominated by traditional players because these layers require licenses, relationships, and operational infrastructure. For example, the largest issuer of tokenized treasuries is Blackrock. Tokenization services and capital layers are where DeFi-native protocols have built real traction. The lien registry layer is the thinnest market, with only DART operating at meaningful scale for consumer credit. These gaps define the opportunity for new entrants and the stack positions worth owning.

Where else does this transfer? Five dimensions to evaluate.

Figure proved the model in HELOCs. The natural next question: which other credit products offer comparable transformation potential? The strongest candidates — where both supply and demand improve — score well across all five dimensions below.

This space seems to be at an inflection point, with many interesting new credit (and credit-adjacent) pools also being put onchain, such as USDai’s loans to AI businesses and Daylight’s tokenized electricity revenue streams. Still, there remain undertapped markets worth watching including first-lien mortgages (largest absolute cost delta), equipment and SMB lending (UCC filing as a DART analog), and trade finance (short duration, high friction, proven on-chain traction via Centrifuge and Huma). Each opportunity set has a distinct profile across these five dimensions.

The flywheel is already spinning.

$19B+ originated at 93% cost reduction, 120–200bps investor yield advantage, and a 5-day close. When high-quality credit is produced cheaper and faster, supply grows; when investors get a better product, demand grows with it. Figure Technologies is proof that anchoring to quality and innovating at the technology level creates a flywheel that compounds.

We research this at Inversion because we obsess over financial infrastructure. What Figure has done to HELOCs is a glimpse of what’s possible across high-quality, liquid credit markets, and a signal of the market shift still ahead.