Lessons Learned Building and Running a $1.9B On-Chain Credit Fund
Kevin Miao is one of the most insightful practitioners at the intersection of traditional and on-chain finance.
Until recently, Kevin was General Partner of BlockTower Credit, a $1.9 billion AUM private credit fund that leveraged public blockchains, stablecoins, and tokenization to maximize capital efficiency.
Prior to BlockTower, Kevin managed a Structured Credit trading desk at Citigroup, focusing on consumer credit, esoterics, and distressed crisis-era securitizations.
Currently, Kevin leads growth at Steakhouse Financial, a DeFi-native advisory and technology business that helps manage the biggest DAOs in web3.
In this conversation, Kevin and Will Beeson explore the theory and practice of running an on-chain credit fund, the benefits and challenges versus the traditional model, current limitations in on-chain investment management, and potential paths to upgrading finance with blockchain rails.
Will:
Kevin Miao, it's great to connect with you.
We’ve been having a number of parallel conversations on different topics—mostly around stablecoins and the work both of us have been involved in within that space.
But the reason I wanted to connect today—and the material I think we could dive into, which I find really interesting—is your most recent professional experience as Managing Director and Head of Credit at BlockTower.
You wrote an essay in 2022 for Packy McCormick, and to paraphrase a line from that piece: you, Kevin, can predict the future. You were echoing a Jeff Bezos quote—essentially saying that borrowers will always want a lower interest rate. So, no matter what else changes, if you can deliver that, you’ll always have a market.
How does that insight frame the founding story of BlockTower Credit?
Kevin:
I definitely don’t want to start this off by saying I can predict the future in any way—but I do really love that Bezos quote. He talks about how, when deciding where to allocate time and resources, he focuses on things that don’t change. For Amazon, that means customers will always want more selection, faster delivery, and lower prices. So anything they do to improve on those three fronts is a worthwhile investment.
When you think about BlockTower Credit—or lending in general—interest rates, of course, are a function of broader macroeconomic conditions. But the spread between the rate a borrower receives and the risk-free benchmark rate is something I believe can—and should—tighten over time, ideally approaching zero.
The way to do that is by improving market structure. That means increasing transparency, but also eliminating a lot of the data silos and manual processes that still exist throughout the traditional lending value chain. Intermediaries add friction—and cost.
So the founding story behind BlockTower Credit was really about applying disintermediating technology—in this case, blockchain-based systems of record—to that exact problem. And not just for record-keeping, but for enabling more sophisticated functions on top of it, like securitization.
By February 2022, the technology had matured to the point where we could prove to people that you could move billions of dollars through this infrastructure with guaranteed accuracy and transparency. That proof of concept is an important first step toward narrowing the gap between what borrowers pay and what they should pay.
Before this, I was running a securitization desk at Citigroup. And when you look at that world—say, someone takes out a mortgage at 6.5%, and the capital backing it is being funded at 5%—that spread isn’t just market dynamics. It’s intermediaries, like me in my old job, stripping out millions (or billions) of dollars into our own pockets.
BlockTower Credit was built to show that you can eliminate a lot of that. That’s the goal.
Will:
Alright, so over the course of this conversation, let’s dive into a lot of the points you just mentioned—and maybe explore the broader arc of the story more generally.
Picking up on one of your last points—your background in securitization at Citi. Can you talk about that? What were you doing at Citigroup?
Kevin:
At Citigroup, I had a few different roles, all primarily focused on what I’d describe as "Big Short–era" credit.
So, for anyone who’s seen the movie The Big Short and has a general sense of how mortgage-backed securities work, the 101 version is: you have tens of thousands of raw mortgage loans—say you want to take out a mortgage—and a bank like Citigroup or one of the other major players will package those loans together.
They then tranche them—breaking the capital structure into senior and junior slices—so some investors can buy the safer parts, and others can take on more risk. After that, the whole structure gets effectively IPO’d to the market.
Who buys these tranches? The senior debt is usually bought by more conservative institutions—pension funds, insurance companies, etc. The riskier pieces—the high-octane stuff—are bought by private credit funds that are running 2-and-20 vehicles. These funds are paid to evaluate and select which pools of risk they want leveraged exposure to.
At Citi, we were structuring and selling those deals, trading them in secondary markets, and running what was essentially a proprietary hedge fund within the broker-dealer. We traded ABS, CDOs, subprime mortgage credit, credit default swaps—basically everything that became highly distressed during the financial crisis.
And it wasn't just mortgages—we traded a broad spectrum of credit: drug royalties, railcar leases, credit cards, auto loans, residential mortgages, you name it. Everything from highly structured products to plain vanilla debt.
That was my experience in TradFi, at least at Citigroup.
I left in 2021 to go run operations at a fintech startup called Capchase. From there, I had a number of experiences that ultimately led me to BlockTower Credit—but more broadly, to the realization that blockchain technology was the area I personally needed to go deeper on.
So, I’ll pause there.
Will:
So if I think about everything you just described from a traditional finance standpoint, there are a few distinct cash flow streams or revenue pools that are associated with the process.
And we might as well walk through all of them, since they’re probably relevant to your work at BlockTower Credit and where this conversation is going.
First, there’s the origination side—when these mortgages or credit products are first created. There’s capital being lent, typically by a bank, and that capital has a cost associated with it. So the interest rate on the mortgage needs to account for the cost of capital, the credit risk taken on by the lender, and of course, some amount of profit. It also has to cover the operational costs of managing the lending business, which—especially in legacy institutions—can be substantial.
That’s just one part of it. And I assume this isn’t only true for mortgages, but also for the other credit products you mentioned earlier.
Then, on the securitization side, you’ve got a portfolio of yield-generating assets—loans, in this case—and an investment bank that’s structuring a deal. That bank is taking a fee for putting the deal together. The cash flows from that loan portfolio get tranched and distributed to different investors: senior, mezzanine, and junior tranche holders.
Senior tranche holders get safer paper with lower yield, while junior tranche holders take on more risk but receive correspondingly higher returns. And throughout all of this, I imagine a portion of those cash flows—either fixed or percentage-based—is being taken out to cover operational and management costs.
Does that generally describe how you see it as well? I just want to start by establishing a baseline before we get into how blockchain—and how BlockTower—is thinking about this space differently.
Kevin:
Yeah, I think for bank-originated debt, that’s pretty spot on.
And by bank-originated, I mean an entity that not only has the license to originate debt but also has the balance sheet to warehouse those loans. Over the past 10 to 15 years, though, we’ve seen non-bank lenders take a significant amount of market share.
These are companies like LendingClub and Prosper—the first wave of fintech lenders that emerged around 2007–2008. They started winning over a lot of consumer borrowing demand, but they didn’t have balance sheets of their own.
Another intermediary in this space would be a bank like Citigroup, which provides warehouse debt capital to companies like LendingClub or Prosper. That capital allows these fintechs to lend to customers on the front end, while also giving them time—maybe six months to a year—to aggregate a meaningful volume of loans under their banner. Once they hit that scale, they can securitize and syndicate those loans to investors.
That aggregation period is important. Especially when you're a hedge fund taking a leveraged position in the junior tranche of a securitization, you want diversification. You’re relying on the law of large numbers to protect you from being overly exposed to a small, correlated pocket of risk in the portfolio.
Other than that, I'd say your summary was pretty exhaustive. Very impressed you pulled all that off the top.
There’s definitely a broader network of intermediaries and service providers in the middle—but honestly, once you go too far down that rabbit hole, it gets pretty granular… and pretty gross.
Will:
So when you think about the cost of capital—and then the yield on these loans—the spread between those two is where everything lives.
In your mind, what percentage of that spread is truly attributable to credit risk? Maybe there’s also an illiquidity premium baked in. And of course, there’s the cost of running a lending business, plus the profit margin the lender expects for putting capital to work.
Given all that, what’s the actual opportunity to optimize with better technology—specifically, to reduce borrowing costs?
Kevin:
That's a really good point—and a really good question.
One of the first pieces of research I saw that tried to quantify this came from Figure Technologies, Mike Cagney’s company. The number they often quote for on-chain securitizations is around 100 basis points of cost savings from end-to-end origination through execution.
That gives us a kind of benchmark—although, to be honest, I think that number is a bit amorphous. It’s hard to pin down with precision.
Let’s go back to mortgages for a moment, specifically the TBA (To Be Announced) market. These are mortgages purchased by Fannie Mae and Freddie Mac, aggregated into securitizations, and sold to investors. There’s no credit risk in the traditional sense because the government guarantees the principal.
Now, if you look at the average spread between the yield to maturity for an investor and the underlying weighted average coupon of the mortgages (what the homeowner is actually paying), it’s typically around 125 to 150 basis points.
So, where does that spread come from?
A portion of it comes from servicing fees. Quicken Loans—or whoever is originating the mortgage—takes 12 to 25 basis points as an ongoing fee. Their job is to move money from the borrower to the investor, handle payment collections, keep records for audits, and navigate state and federal compliance. Essentially, they’re absorbing a lot of middle- and back-office overhead, and they need a buffer to do that.
Then you’ve got the government’s guarantee fee—about 49 basis points—which builds up a reserve to protect principal. That’s not waste; it’s insurance. But it’s still part of the cost.
The rest—call it 76 basis points—comes down to the lender's funding and operational structure. Using Quicken again as an example, they need to borrow money—likely at SOFR plus 250 basis points—and they have to hold that funding long enough to ramp up origination volume before they can securitize and sell the loans.
But they don’t know what the market is going to look like three, six, or nine months down the line. That uncertainty forces them to build in a cushion.
Plus, launching a securitization itself costs money. At Citi, we’d see that as a 1.5% notional cost. That includes legal documentation, modeling, packaging the deal, paying rating agencies—it’s a significant upfront capital outlay. These lenders have to bake that cost into what they charge today to cover themselves down the road.
And we haven’t even talked about the borrower-facing fees. A consumer might pay 1.5% to 2.5% just in origination fees to get a mortgage. Why? Because the lender is trying to build that reserve in case the market blows up in six months and they need to keep funding loans while waiting it out.
There are so many layers here. It’s a really complex question—and honestly, I’d love to dig into it more with you at some point, maybe even write something on it together. I’m not sure I fully unpacked it clearly just now, but if I had to boil it down:
There’s too much time between you taking out a mortgage and me—the investor—receiving it. And there are just too many hands in the middle.
Will:
Yeah, that’s really interesting—that timing aspect—because it’s not something I had really considered before.
In many ways, time is risk. I’d been thinking more about cost and operational efficiency, and therefore cost optimization. But what you’re pointing out is that time itself is also a major contributing factor.
So how does that translate into the founding thesis behind BlockTower Credit? What were you aiming to solve, given that insight?
Kevin:
I’d start by saying that our initial thesis was: we can replace a lot of the manual intervention and record-keeping that happens in traditional securitization processes.
Anyone who’s worked in a bank—especially in middle or back office roles—knows how time-consuming and resource-intensive those functions can be. We believed that if we pointed everyone toward a shared ledger and replaced trustees, administrators, and operational teams with well-defined smart contracts and clear relationships, we could run securitizations at a fraction of the cost.
In traditional finance, a securitization might cost $400,000 to $500,000 per year to manage. We believed we could do it for around $20,000—and at BlockTower Credit, we actually did.
We published a case study with Centrifuge, a partner of ours, showing that in the context of securitization alone, we achieved around a 97% cost reduction. In 2023, we ran over $250 million worth of securitizations on Ethereum mainnet, and we paid less than $40,000—reporting and compliance included.
That was the core of the original thesis: we could eliminate massive operational overhead.
Now, regarding the time component—this is a longer-term view. Our thinking was: if we’ve already moved the final stage of the lending value chain (securitization) on-chain, can we work backward from there? Can we eventually get all the way upstream, to what companies like Quicken Loans are doing—originating loans directly?
We had started building a solution internally—codenamed Orbit—that was meant to complete that journey and support on-chain loan origination.
If you think about a securitization as a company, the tranches represent the debt side of the capital structure. Investors are buying into that pool of loans, trading capital now for a stream of future cash flows. That capital flows downstream—through intermediaries like investment banks or loan originators—before it reaches the borrower.
At BlockTower, we had already moved the liability side of the structure on-chain. But what we realized in doing that is: unless the entire process is moved on-chain—end to end—it doesn’t solve the core issue of timing.
Yes, we can save 97% of the cost on the liability side, but we’re not going to save any time if we still have to wait on data from originators like Quicken Loans, and wait for upstream parties to send us the assets.
That’s what we wanted to change.
But to be honest, we also came to the realization that the industry is probably still a bit too early to do everything we imagined. We're getting there—but the full vision will take time.
Will:
How did you bridge the off-chain origination side with the on-chain liability side?
Kevin:
Great question. So I’m sure you’re familiar with the concept of oracles—essentially, mechanisms for bringing off-chain data into the blockchain ecosystem. That could be price feeds or other external inputs. This is what companies like Chainlink and Chronicle do.
At its core, an oracle is just a way of getting structured data on-chain so that smart contracts can interact with it. For us, bridging off-chain origination with on-chain liabilities followed that same logic.
We were taking information from our custodians and loan originators—things like Excel spreadsheets where each row represented an individual loan and each column included attributes like FICO score, address, notional balance, interest rate, etc.
We’d take that data and plug it into a platform called Centrifuge.
On Centrifuge, every row in that spreadsheet was effectively turned into an asset-level NFT. And that NFT was really nothing more than a digital version of that loan record—it could be transferred, tracked, and assigned ownership, just like any other on-chain asset.
We would then group those NFTs into an on-chain securitization structure—also built by Centrifuge. Let’s say we had 50 loans, but only wanted to include 15 or 20 in a given structure—we’d simply transfer those NFTs into a smart contract representing an on-chain SPV (Special Purpose Vehicle).
From there, any repayments we received from borrowers would come in off-chain—say via wire transfer from our banking partners. For example: “These 25 borrowers sent us $50,000 this month.” We would take that money, convert it into stablecoins, and route it back through the individual NFTs. Each NFT’s balance would update according to the principal and interest paid.
Now obviously, that’s not as efficient as a fully on-chain system. Ideally, you’d have those payments happening directly on-chain from end to end—but this was the best available bridge for now.
Once inside the on-chain SPV, we’d implement typical securitization logic: senior interest gets paid first, then junior interest, senior principal, and so on. Standard waterfall structure.
We also had a $150 million debt facility from a DAO called MakerDAO, and all of the covenants you’d expect from a traditional warehouse facility were enforced on-chain. For example, our fund had to maintain a 30% equity buffer above Maker’s position. If we dropped below that—say due to a loss—and fell to 28%, we’d be accelerated. That meant no more cash flow, no further draws from the facility. That enforcement was coded in.
And look—it may sound simple at a high level, but there are fintech startups today trying to build just one part of this stack. Companies like Finley CMS, for example, are focusing only on the tail end—and they may end up being a unicorn for that single piece.
But when you start with a blockchain-based system of record and build your architecture around that, you begin to realize: you can collapse all of this into one unified system. It was a really interesting experience.
Will:
Well, I doubt that anyone—apart from deep securitization guys like yourself—would call any of what you just described simple. In fact, it sounds incredibly complex.
I’d love to dig into a few of the pieces you mentioned.
You said “on-chain SPV”—is that basically just a smart contract?
Kevin:
Yes. Just a smart contract.
Will:
And you mentioned a 30% equity buffer. Is that held within the SPV itself? So the structure would be something like: up to $150 million of Maker capital, and then 30% of your own equity sitting alongside that within the SPV?
Is that right?
Kevin:
That’s exactly right.
The senior and junior assets were represented as ERC-20 tokens. For our investors—which were very traditional, institutional investors—we were investing in securitized credit. They could have invested directly into the token, held the token, and received the exact same cash flows as if they’d invested through a traditional vehicle.
That said, we still built a conventional fund structure—a Cayman feeder fund, etc.—primarily for tax and regulatory reasons. But also to abstract away the blockchain UX. We didn’t want our LPs needing to spin up a Coinbase wallet just to participate.
Will:
And by “hold the token,” you mean owning the individual NFTs that represent the loans you described earlier?
Kevin:
Not exactly. The token represents ownership of the right to the cash flows from those NFTs.
In a traditional securitization, the underlying assets are, say, mortgages. An investor like PIMCO doesn’t hold the mortgages themselves—they hold the QIP (Qualified Institutional Placement), which gives them the right to receive payments from that pool of loans, according to a defined waterfall.
In our structure, those QIPs were represented on-chain as tokens. So if you look at something like Apollo’s private credit fund on-chain, it’s essentially a QIP—backed by a pool of loans. We took it a step further and actually put the individual loans themselves on-chain as NFTs.
That’s how we start building real transparency—and eventually, the ability to perform much more complex financial transformations than what I’m even describing here. But it’s all abstracted away from the end user.
Will:
Okay, so the token is the QIP, and that QIP corresponds to the cash flows being routed through the SPV—right?
Kevin:
Yep, exactly.
Will:
And the QIPs are tranched? There’s a senior and junior structure? Or does everyone receive the same exposure?
Kevin:
Yeah, it's tranched. In our case, we used a fairly simple senior-junior structure. But Centrifuge’s technology actually allows you to build any kind of capital structure you want—it’s fully customizable.
Will:
Got it.
So in a sense, the entire backend of the capital structure was on-chain—but you were still relying on traditional, off-chain origination. A loan would be created by a traditional lender or custodial partner, and then you'd bring the data on-chain through an oracle.
And when the cash flows came in—when borrowers made payments to the custodian—you’d take the fiat, use something like Circle to on-ramp it into stablecoins, and from there, everything became automated?
So basically, from the point that both the data and the stablecoins were on-chain, everything downstream was handled automatically?
Kevin:
Exactly. Once those two inputs were on-chain—data and stablecoins—everything downstream happened immediately.
And that was kind of the wake-up call for us.
We saw what happens when you enable parallel processing across the liability side of the structure—things that, in my previous career, were handled entirely linearly.
Traditionally, the servicer—say, Quicken Loans—would send a huge wire along with an Excel file to the trustee, saying, “Here are the 20,000 loans that repaid $50 million this month.” The trustee would then have to reconcile those figures and pass them to the administrator. The administrator would then initiate wire transfers through DTC. DTC would wire funds to each individual CUSIP holder. Then, each CUSIP holder’s middle office team would check that the trustee's and servicer’s numbers matched what actually landed in their account.
We were able to do all of that—end to end—in real time.
Not “could do.” We did do it. And that was a huge wake-up call in terms of what’s possible when the entire liability side is digitized and automated on-chain.
Will:
So you said you were nonetheless too early? Most specifically. I guess it has to do with the origination side of it. Can you talk more about that?
Kevin:
I’ll circle back on the origination side in a sec, but first—when we were building this, crypto was essentially a criminalized industry. And honestly, it deserved a lot of the scrutiny it got. But Operation Choke Point and the broader regulatory environment made that really real for us.
We were constantly dealing with bank failures. We were customers at Signature Bank. We were getting offboarded left and right. We couldn’t build redundancy across on-ramps, off-ramps, or even basic service providers.
So that was the first major issue. I think it’s irrational to try to build a real company in an environment where you can’t even get access to basic infrastructure—and I’m not talking about dev tooling. I mean things like getting a bank account.
The second challenge was market comfort—specifically, where the Overton window was in terms of investor willingness to deploy capital into these kinds of structures. Credit investors, even the most technically capable ones, aren’t paid to take much risk. All you get is your principal and interest back. So by definition, they’re not going to be first movers.
We were effectively the only on-chain securitized credit investor in existence. That’s fine if you’re trying to run a hedge fund. But if you’re trying to scale, to become an originator, you need other capital markets participants involved. You need redundancy. Without that, it’s not viable to build a business—you’re not going to be able to sell the loans you originate.
Now, you’re starting to see players like Apollo and Hamilton Lane get involved. They’re doing something that looks more like what we were doing. They want to create on-chain CUSIPs so they can siphon crypto liquidity into real-world assets and earn yield. And that’s great. Hopefully that gets them comfortable enough to eventually go in the other direction, like we did.
But the reality is: it’s still too early.
On the origination side, I think one of the biggest lessons I’ve learned is that you can’t build a company purely off ideology.
I deeply believe in economic justice. I believe people should have a stake in the systems that coordinate and control their lives. That’s why I’m in crypto. And yes, I talk a lot about efficiency and stripping out cost—but only because the outcome of that is better consumer outcomes. Lower interest rates. Cheaper mortgages. That’s what I wanted to help build.
In functional terms, what we’re doing is a 1,000x improvement in efficiency. But then you go try to sell that to an Apollo or a Fidelity and they say, “Yeah, but we don’t really care.” That’s why we wanted to focus on the origination side—direct-to-consumer. But after thinking more deeply about it over the last year, I’m not sure consumers are particularly rational either.
How often do people even shop for mortgages?
Even I—someone very familiar with all of this—applied for a mortgage and only talked to one broker who checked with two banks. That was it. So I’m not convinced that consumers would actually value saving 40 or 50 basis points through blockchain-enabled efficiencies. I believe that’s where the industry should go—but I’m not sure it’s where it will go.
And that made me rethink the entire approach. What’s something that’s not just better, but completely impossible in the traditional world?
That’s where I want to focus going forward.
Let’s talk about underwriting. Before you even get to Quicken Loans, there’s already so much data collected about you—credit scores, bureau data, etc. That data is sold to lenders for billions of dollars a year. Lenders then pass those costs back to consumers through higher interest rates and origination fees.
But on blockchain, data and value move on the same rails. So what if, instead of just saving consumers money, we could help them monetize their own data?
When you pay off your loan, that repayment data could be ledged—privately or quasi-privately—on-chain. And if you borrow through our platform, we’d have that full repayment history. We could collectively bargain on your behalf and return some of that value to you.
That’s the longer-term vision. It’s a massive idea. It’s not possible today in the way I want to build it—but it’s not that far off.
That’s where I want to go over the next 12 to 18 months.
Will:
Well, I love that vision. It’s extremely powerful. And I also really like how you’ve framed the question—not just, “What can blockchain do better?” but, “What’s actually impossible to do in the traditional space?”
Kevin:
Exactly.
If you look at some of the most successful projects in our space recently—Athena is a great example—it’s just not possible to offer 15–20% market-neutral returns outside of crypto.
Crypto has a unique market structure. It’s effectively long-only, because of the nature of the investors involved. I remember talking to Goldman maybe a year ago, and they said, “This trade has infinite demand—because it doesn’t exist in the traditional world.”
And that’s really what sets Athena apart. When I think about why they’ve succeeded—and why BlackRock’s tokenized money market fund is only about a twelfth the size of Athena’s, despite launching around the same time—it’s because Athena gave people access to something that was previously impossible.
There’s a huge lesson in that for builders across crypto, especially for people like me who don’t come from a pure finance background.
We often assume that if we just make something more efficient, people will love it. But that’s not how it works—especially not with consumers. What people really respond to is access to something they couldn’t get before.
That’s what hit me. It was a huge wake-up call.
Will:
Yeah, that’s a reflection I think we should all keep in mind.
You mentioned Athena, which is super interesting—and it kind of ties into the last component of all this from a BlockTower Credit standpoint that we haven’t really talked about yet: the Maker facility. The leverage you were able to access.
At the end of the day, that’s pretty fascinating, right? Because on the other side of that leverage is retail. There’s no investment bank providing traditional financing to Maker, which it then turns around and lends on-chain.
How did you approach capital sourcing on-chain?
You talked earlier about the challenge of placing the equity tranche with traditional investors—the appetite just isn’t there yet for on-chain exposure. But it does seem like there’s a significant opportunity to source liquidity that might not currently be optimally deployed on-chain.
Can you talk a bit more about that?
Kevin:
Yeah, absolutely.
It’s funny—I’ve since left BlockTower Credit and am now with Steakhouse Financial. One of the co-founders there is Seb Devereux, a French guy who used to run the real-world asset (RWA) efforts at MakerDAO. This was way back in 2018–2019, before most people were even thinking about any of this.
In 2020, Maker deployed about $100 million to a regulated bank in Pennsylvania called Huntingdon Valley Bank. The deal gave them pari passu exposure to small business loans being originated in the region.
Now, whether or not that was a good deal for Maker is debatable—but to me, when I was still at HBS, it was mind-blowing. A DAO—a decentralized organization—had deployed capital to a regulated U.S. bank. And both sides understood each other. They were speaking the same language.
That was what gave me the conviction to drop out and pursue this. There was a real precedent.
For me, the need was long-term capital. Across the DeFi ecosystem, the illiquidity premium is huge. Athena’s model works because it’s wrapped in a stablecoin—people can move in and out. But there are very few entities that can deploy real size over long periods of time. Maker may be the only one.
That Huntingdon Valley deal was a proof of concept. And we thought, “Okay, if we can give Maker what they want—namely, protection and value—we can get access to that capital.”
So we subordinated our equity capital below theirs. And beyond that, we weren’t just credit guys—we were also some of the best engineers in the space. So we said, “If you want to collateralize your stablecoin with real-world assets like credit card receivables, we’ll figure out how to do that.”
That’s how the relationship with MakerDAO came together.
That said, DeFi—and the DAO ecosystem more broadly—is very political. So the way we went about it was basically self-promotion in Reddit-style governance forums. That’s not something most traditional companies are used to doing. But in crypto-native contexts, that’s just how it works.
I learned a lot from that experience, especially about informal power structures within DAOs. It was valuable—but definitely not scalable in a traditional sense.
Will:
So would you say that was a one-off, point-in-time opportunity to source that capital from Maker on favorable terms? Or do you think it reflects a broader category of opportunities?
Kevin:
That question deserves a nuanced answer.
To do exactly what we did, again, in the same way? Probably not. It was a very specific moment.
This is an open-source ecosystem where transparency is paramount. And we were very open with the Maker community that the deal was—at least in part—a subsidy to help us get off the ground so we could, in turn, help build additional infrastructure for them.
For example, we’re probably still the largest single on-chain holder of tokenized treasuries. At BlockTower Credit, we ran $1.5 billion of treasuries for MakerDAO. We did that as a trade: they gave us startup capital to launch a private credit fund, and we helped them build out the asset side of their balance sheet.
It was collaborative—we weren’t trying to be an extractive counterparty. We genuinely wanted to help grow both Maker and DeFi more broadly.
I think the big change now is that a lot of traditional finance players entering the space are purely extractive. Their main business is still in TradFi, and this is just another revenue channel. They’re not here to build the ecosystem in the same way.
That’s a big shift.
But from a structural standpoint, we’re starting to see more projects try to replicate what Maker did at a meta level. Maker essentially wrapped illiquid debt and liquid assets into their stablecoin—DAI—to make it feel liquid. And it is—DAI is one of the most liquid stablecoins today.
Now you’ve got new players, like Maple with their Syrup token, trying to create liquidity for long-duration assets. I’m advising a number of companies now at Steakhouse that are trying to do similar things.
This model—taking semi-liquid assets, creating a liquidity layer around them, and issuing tokens that can be traded—looks a lot like credit ETFs in traditional finance. You’ve got underlying instruments that are relatively illiquid, and authorized participants offering intraday liquidity via token creation/redemption.
That, to me, is one of the real growth areas for crypto and RWAs going forward. It’s a path worth following.
And frankly, I think we—alongside MakerDAO—accidentally pioneered it. Sometimes I don’t even think Maker fully realizes how innovative what we built together was. But now, it’s becoming the playbook for everyone building in this space.
I’m excited to see where it goes.
Will:
What sort of returns were you able to generate on the fund and how would that compare with a traditional credit fund?
Kevin:
There were really two components to the returns in our fund.
The first was the levered credit strategy. We were very intentional—both strategic and tactical—about what we deployed into. Primarily, we were buying investment-grade asset-backed securities. Things blew out a bit at the end of 2022, so the yield to maturity on our book was in the range of 9.5% to 10.5%.
We were leveraging that at a 70% loan-to-value, with a cost of capital around 4%, which was provided by Maker. Now, once you factor in the expenses of on- and off-ramping and all the other operational friction, maybe the effective cost of capital was closer to 5%.
Even with that, we were generating 19% to 20% gross returns—off of investment-grade assets yielding around 10%. In traditional markets, a private credit fund or a securitized credit fund at the time would have been borrowing closer to 7%. So just on cost of capital alone, we had a 200–300 basis point advantage, which translated to an extra 5% to 6% in gross return.
That was one component.
The second component was what I’d describe as a form of yield farming. Because we were bringing significant capital and visibility to some of our partners—like Centrifuge—and they were invested in our success, our fund received substantial exposure to their tokens and warrants.
So for our earliest investors, the total gross return—combining both the levered credit and token exposure—was likely in excess of 35%. And that was all from taking investment-grade ABS risk.
I’m really proud of what we built with that fund. I think we were excellent credit investors. Of course, the jury’s still out, but I believe we did a great job.
But the biggest lesson was that investing—especially in a frontier space like this—is so much about timing and luck. We were incredibly fortunate in terms of when we launched.
The skill, I think, came from imagination and our willingness to explore a new paradigm—rather than just building the 15th version of Angelo Gordon.
No offense—those guys are great. But we were trying to do something different.
Will:
Two final points I’d love to cover.
First—it sounds like BlockTower Credit is no longer your full-time focus. No need to get into the full backstory, but what’s the short version of what’s happening with BlockTower Credit now, and what might its future look like?
Kevin:
We returned capital from BlockTower Credit to investors after what I’d consider a really successful run.
There’s this quote from The Dark Knight that always sticks with me: “You either die a hero, or you live long enough to become the villain.” And I think that captures how I felt toward the end. The biggest opportunities in this space aren’t just in running hedge funds—they’re in building and reimagining everything. It’s not about solving just one piece of the puzzle.
So in Q4 of last year, we made the decision to wind it down and return the capital. We executed that by the end of the year.
Since then, I’ve started a new role as Head of Growth at Steakhouse Financial, which partners with a broader suite of real-world asset issuers, hedge funds entering the space, and stablecoin projects. Steakhouse also runs some really interesting DeFi initiatives directly. So I’m incredibly excited about the work we’re doing there.
As for BlockTower Credit’s future—I think, like most hedge funds, it’ll eventually fade into the background. That’s honestly something I struggled with. One of the things that draws people to crypto—especially the builders—is this desire to create something that outlasts you. You’re writing smart contracts that might literally run forever.
So to have poured so much energy into something, to have learned so much, and then to accept that it won’t be that enduring legacy—it was tough. But it also gave me an amazing foundation for what comes next.
And right now, I’m exactly where I need to be.
Will:
Final question for you.
Let’s imagine a future iteration—BlockTower Credit 2.0. In order for it to fully realize the vision you originally had, what would need to change?
How specifically would the market need to mature to make that possible?
Kevin:
Great question. I can definitely get a bit nerdy here—but that’s not really a departure from how BlockTower Credit was built.
Putting my ideology aside—things like economic justice and building systems that empower people—if you just look at BlockTower Credit as a fund, then our core mandate was to (1) grow, (2) drive down operational costs, and (3) gain capital efficiency through regulatory arbitrage.
That third pillar is actually the most important. It’s why Apollo owns Athene. It’s the Berkshire Hathaway model using Geico float. In my view, most of the real innovation in buy-side finance hasn’t happened on the asset side—it’s all on the liability side.
Even pod shops like Millennium—what do they really do? They run low-volatility strategies and lever them up. Their alpha isn’t necessarily the strategy—it’s their ability to access leverage from banks at favorable terms.
So, for BlockTower Credit 2.0 to succeed, I think a few major things need to change.
1. Market Maturity
First, more people need to get comfortable with this technology. That’s happening slowly. Ethereum’s continued uptime helps. BlackRock’s move into the space—particularly with BUIDL—is a massive step toward shifting the Overton window of what’s acceptable in institutional finance.
2. Operational Efficiency
There’s still a lot of friction. We were operating on Ethereum mainnet, which is essentially the worst technology that will ever exist in this space. And it’s already improving. With Layer 2s and other infrastructure coming online, we’ll see transaction costs trend toward zero.
But that’s only one half of the equation. The cost of moving in and out of the blockchain ecosystem—fiat to digital and back—must also go to zero.
Today, if you want to use a stablecoin payment orchestrator like Bridge, you’re paying something like $8,000/month plus 25 basis points per transaction. That’s not sustainable, especially when you're competing against free ACH payments subsidized by banks that hold float.
Even stablecoin issuers like Circle and Coinbase have financial incentives to keep funds in their ecosystem. They’ll charge you exit fees—say, 10 basis points if you off-ramp $10 million in a month. That friction needs to disappear.
And I think it will. Banks are already moving toward something closer to what Silvergate and Signature offered us a few years ago—instant, costless transfers from your bank account to an EVM-compatible chain. Eventually, that functionality will just be part of a traditional bank’s core product suite.
I don’t believe on/off-ramps can exist as standalone businesses. Someone—probably a bank—is going to bundle them into a broader offering, and that will force the cost of bridging fiat and crypto to zero.
3. Regulatory Reform
The third piece is regulatory alignment—specifically, changes like repealing SAB 121, which currently requires digital assets held in custody to be counted on a custodian's balance sheet. That makes it difficult for institutions to hold blockchain-native assets.
We worked with S&P, Moody’s, and other rating agencies to get our on-chain securitizations evaluated. If you get an investment-grade rating, then insurance companies and other low-cost capital providers can hold your product. That’s critical.
Why? Because 70–80% of any securitization capital stack is going to be a senior note yielding 4.5%–5%. Who’s buying that? Not a hedge fund. But an insurance company will—if they can hold it on their balance sheet and apply favorable risk capital treatment.
That’s the real game: the liability side. It’s where the quiet leverage lives. Apollo and others have figured that out. It’s opaque, but it’s also the blunt-force object that allows firms with access to low-cost capital to outcompete everyone else.
Even if I save 100 basis points on a securitization—like Figure does—if I’m raising capital at 15% from hedge funds and someone else is raising at 4% from insurance companies, I simply can’t compete.
So yeah—I went full nerd there.
But I really do believe this is going to change. And when it does, everything we built with BlockTower Credit will make even more sense in hindsight.
Will:
Just to round out that point about the repeal of SAB 121—there’s also a related issue with how the Bank for International Settlements (BIS) classifies digital assets.
Right now, they’re essentially treating all digital assets as if they’re equivalent to highly volatile cryptocurrencies, and assigning the highest possible risk weighting to them. That has massive implications.
Take the BlackRock tokenized money market fund, for example. If a bank owns that fund in tokenized form, it isn’t viewed as high-quality liquid assets or Tier 1 capital. Instead, under current BIS guidance, it’s treated as the equivalent of holding speculative crypto—despite the fact that it’s a money market fund managed by BlackRock.
That classification, frankly, is shocking. It completely misrepresents the nature of how these tokenized instruments are actually structured and settled.
Sure, you could make the argument that there’s some technology risk in using Ethereum as a delivery mechanism. But ultimately, BlackRock’s BUIDL fund is issued as a token on Ethereum for convenience. The actual fund has off-chain books and records. If Ethereum disappeared tomorrow, BlackRock could still reconstitute the full fund and repay investors.
None of that nuance seems to be reflected in current bank regulatory guidance—and I agree, it’s one of the biggest remaining pinch points for adoption.
Kevin:
Yeah, I think you nailed it.
At the highest level, native digital assets are bearer assets—but the reality is, the tokens we created, and even BlackRock’s BUIDL fund, aren’t actually bearer assets. Under today’s ATS (Alternative Trading System) regime in the U.S., even firms like Securitize and others with broker-dealer licenses are still operating off of paper ledgers.
Hopefully that changes in the future. But as of now, there are off-chain backups and records—and you captured that nuance exactly right.
Of course, what really drives institutions to go deep and engage with that nuance is profit. And it has to be significant, because these are massive organizations.
What I’ve observed over the past few years is a shift—from institutions entering the space out of fear (“I don’t get this, but I don’t want to be left behind”)—to entering out of genuine interest, because they’re starting to see real opportunity.
When players like BlackRock get involved—who, let’s be honest, aren’t in the game out of curiosity but out of profit-seeking—it signals to everyone else: “Wait, maybe there’s something here.”
And that kind of motivation will lead more institutions to get into the weeds, to understand the structure, the nuance—exactly what you just laid out. Because you’re right: that misunderstanding is holding back a lot of infrastructure development and the broader ability to treat the on-chain economy with the same seriousness as off-chain.
Will:
Yeah, I totally agree.
There’s so much institutional energy right now—especially since the shift in the U.S. political landscape. It's honestly night and day compared to just a few months ago.
So I’m personally very optimistic about the future of on-chain finance.
Kevin, thank you so much. This was an incredible conversation. Really glad we got to go deep on all of this.
Kevin:
Absolutely. Thank you for your time.