In Part 1 of this series, we reviewed the history of debt in bitcoin mining, examined key principles of debt, and looked at some of the most common structures available to bitcoin miners. Now that we understand the landscape, we will take a look at the considerations for borrower and lender alike, the effect of leverage on mining returns, and discuss how the future of the market might look.
Many of the lenders in bitcoin mining have been viewed as predatory, gouging hard-working hashers at usurious interest rates and driving miners into bankruptcy. But your humble author believes this critique to be a bit unfair. The last 6 months suggest that bitcoin mining lenders were perhaps not conservative enough in their underwriting practices, and perhaps should have demanded even more protection or higher interest rates (remember the mantra, “there’s no bad risk, just a bad price”).
So let’s examine the key considerations for a lender to better understand where they are coming from. When evaluating the below considerations, instead of saying certain aspects of a loan make it “better” or “worse,” we will instead discuss them as having more or less “risk.”
Primary Goal: the lender’s primary goal is to earn attractive risk-adjusted returns (“return on capital”). This means AT LEAST receiving their principal back (“return OF capital”), and this latter concern is the motivating force behind most lender considerations.
Now that we understand the considerations for lenders, we will examine how miners should approach these structural features. To simplify things, we will consider a loan to be “more attractive” from a borrower’s perspective if it comes with less restrictions thereby providing them more flexibility.
The following section illustrates the effect of leverage on the returns of a bitcoin mining operation. The outputs were created using a simple cash flow model which can be accessed here. The curious reader is invited to download a copy to play around with the model and explore the impact of leverage on various mining scenarios.
Before we dive into the details of the case study, we will state a few simplifying assumptions that are unrealistic, but helpful for purposes of forecasting.
The operation assumptions are fairly simple:
Now, we will consider the effect of adding an ASIC-backed loan with a ~50% LTV ($750k of debt proceeds, prior to a 2% origination fee). We assume this $765k loan has an interest rate of 15% and an 18 month term.
For three hash price scenarios we will examine the following output. The top half of the analysis shows the monthly cash flows, and the bottom half shows the returns over a 3 year forecast period. The left and right columns show the results of the miner taking on that leverage (LHS) vs. the results if the miner never took on leverage in the first place (RHS). Notably, when looking at the monthly cash flows for the miner that did not take on debt (“unlevered”), the reader can see the cash flow profile of the levered miner after the debt is paid off.
As the “Base Case” we will look at the approximate 3 month trailing average hash price (~6.5 cents / TH / s / day). Under these conditions, the returns from a mining project with the above specifications are marginal. It takes nearly the entire 3 year period for the miner to recover their initial equity investment. This helps illustrate the brutality of recent bitcoin mining conditions.
Now, if hash price fell to 5.0 cents (the approximate hash price if bitcoin fell to $15.3k and the network held at 280 EH/s), the debt service exceeds the operating profit, resulting in the miner hemorrhaging cash until the debt is paid off. The returns in this scenario are even worse, with the levered miner never recovering more than 14% of its invested capital, while the unlevered miner earns only 45% of the invested capital. This highlights the role that leverage can play in harming the cash flow profile of a miner.
In reality, however, it is unlikely a miner would continue to hemorrhage cash by servicing this debt until maturity. Typically, one or more of the following would occur: restructuring, default, or bankruptcy. The field of distressed debt is highly complicated and nuanced but below is an oversimplified discussion of each (disclaimer: the only true experts in this subject are practitioners with years of experience in navigating restructurings and bankruptcies).
The reason many miners took such risks, however, was the effect leverage can have on returns (though some might claim ignorance played a role as well). To better understand, let us consider the returns if bitcoin pumped back toward its previous all time high of $69,420 the day after the loan was issued and the capital invested. This would result in ~23 cent hash price (assuming network hash rate stayed flat).
In this scenario, the levered miner earns a return on their invested capital 3x higher than that of the unlevered miner. And the operating cash flow so vastly exceeds the debt service, that the monthly payments are inconsequential. Looking at this scenario, it’s not surprising that so many miners aped into machines in 2021. Few other physical asset classes have the potential to offer returns like these (though it’s worth noting that the price of machines and infrastructure would likely inflate as well if mining conditions improved so drastically, but this is intended to be a simple model, so we try to minimize the number of changing assumptions).
The TLDR is perhaps unsurprising: leverage offers miners the ability to amplify returns and expand faster, but it comes at the cost of fragility during bear markets. Those seeking to understand the effect of different assumptions and added complexities on returns are invited to play around with the open-sourced model.
Two years ago, if you, dear reader, had asked me, your humble author (a former asset-backed securities investment banker and ASIC-backed lender) how debt capital markets in bitcoin mining would evolve, I would have said debt in bitcoin mining would evolve just like many other asset-backed credit markets with specialty lenders.
In these markets, when a new asset class emerges, originations start with smaller lenders offering new debt products at high interest rates. Over time people get more comfortable with the asset class and interest rates decline (as underwriting prowess increases and perceived risk decreases). Eventually other financial institutions agree to purchase these loans (back-end financing), changing the business model of the lenders to become “origination platforms,” collecting a fee on the loans they originate and then sell. Banks then facilitate the issuance of “term” asset-backed securities to other investors (largely insurance companies, seeking a fixed return over a fixed term). So the “origination platform” creates the loan and takes a spread, the bank takes another spread for packaging the asset-backed security, and the investor enjoys the rest of the loan’s economics. This is how securitization markets work. And given the parallels to other securitized asset classes (e.g., traditional equipment finance), I would have said that ASIC-backed debt would follow this same path.
But such securitization markets take years to develop, and the boom and bust of ASIC-backed debt in the last 12 - 24 months has highlighted many shortcomings of the structure that your humble author and many lenders did not predict. These shortcomings call into question the long-term viability of the structure, and are as follows:
The cost of capital for ASIC-backed debt also poses an issue. It is unclear if the interest rates of ASIC-backed debt will ever be competitive with bitcoin-backed debt. Perhaps this is unsurprising, as bitcoin is the best form of collateral in human history. For this reason, it seems unlikely that bitcoin-backed debt will ever go away entirely (in fact, the author expects this market to expand by many orders of magnitude over the next few decades). But even setting other structures aside, the macro backdrop of rising interest rates may hit ASIC-backed debt harder than bitcoin-backed debt, as a ~5% increase in ASIC-backed debt could take interest rates as high as 20-30%.
The above reasons cast doubt as to whether ASIC-backed debt will reemerge as the dominant structure when conditions improve. But what will replace it? While the exact structures are uncertain, the cost of capital will remain king. If miners do decide to employ leverage, they will continue to seek the cheapest and least onerous form of debt.
Perhaps instead of ASIC-backed debt lenders will make a push for large corporate debt facilities with all-asset liens (i.e., bitcoin, ASICs, infrastructure, and any PPAs) or even site-specific loans. Such structures may offer a lower cost of capital perspective, as in an event of default, a lender could foreclose on an entire mining operation and run it themselves. The lender would then earn all associated free cash flow while they look to liquidate the site (if they want to liquidate at all). In such a scenario, a lender-cum-miner would have far more flexibility than they would if they were forced to liquidate a portfolio of offline ASICs. The desire for “self-sufficient” collateral may mean that hosted data centers are the best lenders of ASIC-backed debt for their customers, as foreclosure on collateral is trivial (the lender just needs to redirect the ASIC to the mining pool of their choice).
One other interesting manifestation of this theme is that retail miners with good credit scores (who often do not have access to ASIC-backed debt) may be able to take out personal loans to finance machines, which are sometimes priced at lower interest rates than ASIC-backed debt (if ASIC-backed debt is available to them at all).
The quest for interest rate arbitrage could also give an advantage to would-be-miners from other industries that have lower cost of capital. For example, an energy super major has far more debt products available to them at a lower cost of capital. Energy companies could use this to their advantage if and when they enter into bitcoin mining in earnest.
From a lender’s perspective, cost of capital is equally important. Only lenders with access to cheap capital will be able to compete in offering debt financing to miners. Imagine two lenders: one is a venture-backed startup the other a large bank. There is little chance a venture-backed startup will be able to provide loans at a lower interest rate than banks with access to the fiat debt capital markets (even when subsidized by fiat VC funding). The only way to win in financing businesses is to have the lowest cost of capital.
Perhaps instead miner financing will be dominated by entirely novel structures. One debt-like product that might play a role is hash rate-based financing. While the market has yet to settle on one dominant structure, the crux of hash rate financing is that miners should be able to sell their future production similar to other commodity producers. The benefit is twofold:
Famously, hash rate marketplaces have been 6 - 12 months away for the last 3 or so years, but Q4 2022 saw a new push of such products. This form of financing would be entirely new to bitcoin mining, and, at time of writing, it remains unclear whether there is sufficient demand for hash rate to allow this market to flourish.
But regardless of what debt products dominate bitcoin mining in the future, at time of writing (January 2023), it seems likely that we will see far less debt in the near term, as miners and lenders alike have been burned in this downturn. Moving forward, miners will likely be more hesitant to use leverage and thereby increase the fragility of their businesses. Likewise, many lenders have left the asset class either due to poor loan performance or because they’ve gone out of business altogether (e.g., BlockFi and Celsius). Those lenders who do remain will likely be far more conservative moving forward (though the next bull run may see a return to similarly aggressive lending practices from new entrants).
This highlights one last key similarity with other credit markets. During bull markets, things are fine and everyone makes money; but during bear markets and recessions, many borrowers, lenders, and investors get burned, resulting in a decrease in originations and an increase in interest rates. This will persist for a while, until conditions improve. Optimism is regained. And the cycle repeats again.
Like other industries, there will be a diversity of attitudes toward debt. Looking back to the energy sector again, many energy companies choose to stay away from debt altogether because of volatile commodity prices. On the other hand, some energy companies binge on debt, a decision which can be wildly successful or catastrophic depending on market timing and execution. Only time will tell which strategies will succeed.
Check out Part 1 of this seris if you haven't already done so.
This article was written for the Braiins blog by Drew Armstrong. Drew is the President and COO of Cathedra Bitcoin, a company that believes sound money and abundant energy are the keys to human flourishing. Prior to joining Cathedra, he was a founding member of Galaxy Digital’s bitcoin mining team and helped build out Galaxy's mining equipment finance product. Drew began his career at Barclays' investment bank, where he focused on the origination of esoteric securitized products, such as data center securitizations and collateralized fund obligations. His views here do not reflect those of any of his past, present, or future employers. Follow Drew on Twitter.
Bitcoin mining software company: Braiins Pool, Braiins OS+ & Stratum V2.
By miners, for miners.