Shareholder Letters

2024 Letter to Shareholders

To the Shareholders of Cathedra Bitcoin Inc.:

When we wrote you last in Q1 2022, the world was a different place.

We were laser-focused on mining bitcoin off-grid with waste methane. At the time this made sense; the fuel source was cheap and abundant (in fact, it still is). Supply chain bottlenecks led to unacceptable lead times for building new on-grid sites. The off-grid opportunity offered a (relatively) untapped market to anyone that could build and deploy infrastructure.

We hired a best-in-class team to attack this strategy. We developed manufacturing operations in New Hampshire to build our own modular data centers for the oil patch (“Rovers”). We even began hatching plans to vertically integrate, dreaming of buying oil and gas assets ourselves to become a full-fledged energy company. But as the twentieth century philosopher Mike Tyson once said, “everyone has a plan ‘til they get punched in the mouth.”

The European energy crisis upended the economics of off-grid bitcoin mining. Oil and gas prices sky-rocketed, a trend only exacerbated by Russia’s invasion of Ukraine. Due to increased demand from the oilfield, the cost of buying or renting generators to convert gas to electricity on-site became uncompetitive with other power sources. In 2020 and 2021, such gensets could be leased for 3 to 3.5 cents per kilowatt-hour. In 2022, this ballooned to as much as 8 cents. These rates have yet to come down meaningfully.

The onslaught of COVID-era supply chain disruptions plus monetary and fiscal stimulus caught up to us; significant price inflation reared its head for the first time in decades, leading to the most acute increase in interest rates the U.S. has seen since the Volcker era.

When writing our last letter to shareholders, we believed there was still juice left in the 2021 bitcoin bull market. We were wrong. Instead, hash price continued to deteriorate over the course of 2022. Many of the largest institutions in crypto would go on to collapse, sending bitcoin price below US$16k, while network hash rate continued to climb.

In May 2022, we realized the bitcoin mining landscape was shifting beneath our feet. So we pivoted. We sold all of our bitcoin at ~US$30k to pay down debt in anticipation of a prolonged bear market. Operationally, we optimized for free cash flow, deploying machines at third-party on-grid data centers, cutting costs (including our own salaries), utilizing aftermarket firmware to improve fleet efficiency, and only expanding in a capex-lite manner. There is a reason we celebrated the cockroach during the bear market: it survives.

Reflections on Bitcoin Mining

Once we battened down the hatches, we had ample opportunity to reevaluate our prior assumptions and reflect on the business of bitcoin mining. A few of our conclusions follow:

• “Permissionless”: not always good for business. Bitcoin mining is commodity production. But unlike other forms of commodity production, unmined bitcoin is not a physical resource one can secure rights to or control. And the key input–energy–is more evenly distributed than other commodities geographically, inviting worldwide participation. As knowledge of bitcoin mining increases, the opportunity is being arbitraged away. In short, the problem with pure-play bitcoin mining is that someone else is always plugging in an ASIC. This is good for Bitcoin and bad for shareholders.
• ASIC efficiency improvement. The efficiency of bitcoin mining ASICs improves over time. When we took the helm of Cathedra in 2021, we regurgitated talking points about “the end of Moore’s Law.” The end may still be near, but the bleeding edge of bitcoin mining hardware is now 50% more energy-efficient than it was then. Even if subsequent models do not follow the step-change efficiency gains of the early ASIC era, we do expect the trend of increasing efficiency to continue. This means, all else equal, network hash rate will continue to increase even in the absence of additional power capacity being dedicated to mining. And bitcoin miners must continue to invest in latest-generation machines just to maintain their share of mining rewards. In fact, the primary source of mining margins to date has been the silicon bottleneck–the machine manufacturers have only been able to secure so many chips. As a result, they extract much of the economic rent from the bitcoin mining value chain when they mark up machines following improvements in mining conditions. Even if a limit on ASIC efficiency is reached and manufacturers focus instead on maximizing output while minimizing cost, the effect on mining economics would be the same.
• Diminishing block reward. This hash rate growth, paired with bitcoin’s finite supply and diminishing issuance rate, means it becomes more difficult and expensive to mine a given quantity of bitcoin over time, as measured by bitcoin produced per unit of hash rate.
• Volatile revenues. The market price of the commodity in question (bitcoin) is also extraordinarily volatile, rendering planning, capitalizing, and sustaining a bitcoin mining business uniquely challenging. The rising dependence on variable transaction fees will compound this volatility. In this epoch alone, we have seen transaction fees vary from less than 4% of the block subsidy to over 400%. This volatility is the new normal.
• Hash price decays. This revenue volatility means bitcoin mining margins fluctuate within market cycles, but given continued ASIC efficiency gains and the diminishing issuance rate, it is likely that hash price will trend toward zero over time. Appreciation in the price of bitcoin will delay, but not fundamentally alter, this effect.
• Non-economic actors. Exacerbating these issues, there are also certain market participants who are content to mine bitcoin at a loss. Nation-states might mine to diversify their GDP by monetizing state-owned energy assets, to evade sanctions, or just to line their own pockets by appropriating power from the private sector. Hobbyists might use ASICs to heat their homes or to accumulate bitcoin in a private manner. Bitcoin mining businesses that are accountable to profit-and-loss statements must compete with these economically irrational actors. Meanwhile, many publicly traded bitcoin miners have been rewarded for ignoring metrics like return on invested capital, instead raising equity whenever markets allow so they can buy new machines and increase hash rate. This growth-at-all-costs model has proven successful at raising stock prices in recent years, notwithstanding the faulty economic logic and negative effect on per-share financial metrics. We call this phenomenon–financing purchases of the latest-and-greatest machines with dilution to achieve growth for its own sake–the “ASIC Hamster Wheel.” However, the arrival of dozens of spot bitcoin ETFs may usher in a new capital markets paradigm. No longer are bitcoin miners the only game in town for investors seeking bitcoin beta in their brokerage accounts. We suspect the proliferation of such accessible and efficient ETFs to discipline bitcoin miners’ capital allocation over time, as investors begin to focus on returns on capital rather than merely top-line hash rate growth.
• Conclusion. These economic realities illustrate the difficulty of pure-play bitcoin mining and suggest that, in the long run, the industry will deliver low returns on invested capital. Perhaps perfectly timing ASIC purchases or bitcoin sales is enough to combat these headwinds, but it is not something we as a management team feel we can predictably do over the next 5, 10, or 50 years. It also seems to us that these dynamics are not well understood by the bitcoin mining investment community at large.

If our thinking is sound, then to consistently produce attractive returns on capital, a bitcoin miner must enjoy “unfair advantages.” Examples of such advantages include:

• Privileged access to latest-generation bitcoin mining machines.
• Lower cost of capital due to scale and significant liquidity of publicly traded securities.
• Ancillary revenue streams which are uncorrelated to bitcoin mining economics (and ideally which are synergistic with bitcoin mining itself: power production, power trading, or multi-purpose data centers, for example). Such ancillary revenue streams are already employed by some miners, but in future years, they will become table stakes.

The pursuit of these unfair advantages is natural. After all, competition is for losers. In the absence of such advantages, we believe the purchase of ASICs ought to be viewed as a trade on short-term market conditions rather than a way to predictably compound capital at attractive rates over many years.

The task before us now is to build the business anew, guided by this sober appreciation of the essential characteristics of bitcoin mining and fueled by our determination to develop unfair advantages of our own. In doing so, we will continue to rely on a few core beliefs which have not changed in recent years:

• The importance of Bitcoin. We remain convinced that bitcoin will emerge as a dominant global reserve asset in the 21^st century and beyond, due to its unique properties (finite supply, political neutrality, digital nature, etc.). Prevailing macroeconomic and geopolitical conditions will aid in its adoption–specifically, overleveraged balance sheets globally; deglobalization amid the end of a unipolar geopolitical paradigm; the rise of populism and political instability globally; and the return of significant monetary and consumer price inflation.
• The proliferation of bitcoin mining. We believe bitcoin mining’s total addressable market will increase, even if that market generates low returns on capital for investors. This will be due in part to its saturation of the energy sector, as companies seek a secondary market for their power. Such a secondary market may be necessary to survive the volatility brought on by recent Malthusian energy policies in the West, which promote intermittent energy resources at the expense of baseload generation, resulting in increasingly volatile electricity prices. The energy sink of bitcoin mining will absorb some of this volatility and help salvage malinvestment. Meanwhile, stranded energy resources producing low-cost power will use Bitcoin as an alternative market. Many miners are going out further on the geopolitical risk-curve to utilize such resources.

• The growth in compute. More broadly, the proliferation of digital technologies–including bitcoin and artificial intelligence–will continue, requiring more compute and related infrastructure in the coming decades. Given the promise of its capabilities, we may see general-purpose compute exhibit a similar “arms race” dynamic to bitcoin mining. Amazon’s recent purchase of a 960-MW data center campus collocated at a nuclear power plant is a key data point in this trend (as well as the premium that will be placed on baseload generation). These voracious technologies will continue to demand more energy, resulting in more data centers seeking more reliable power. 

Business Update

Beyond introspection, 2023 was about reorienting the business in light of the lessons learned from the last market cycle.

We deployed the last machines from our 2021 futures orders, increasing fleetwide hash rate from 203 PH/s to 404 PH/s with minimal capex. We also continued to optimize our free cash flow by dynamically changing our machines’ power draw using our custom after-market firmware, CathedraOS, which we subsequently released to the market as a new product. With these margins, we began stacking bitcoin on our balance sheet again; we hold approximately 48 bitcoin worth US$3 million at time of writing.

Finally, and perhaps most importantly, we took steps to restructure our balance sheet, settling C$13.2 million of debt for shares throughout the year and extending the remaining C$5.7m outstanding until Q4 2025. It was this debt restructuring which allowed us to begin a more comprehensive strategic review of the Company and begin discussions with potential partners  concerning a business combination.

The Kungsleden Transaction

In March, we announced a proposed transaction to merge with Kungsleden, Inc., a U.S.-based developer and operator of data centers which primarily provide hosting services to bitcoin miners. Kungsleden owns and operates three data centers in Tennessee and Kentucky, totaling 30 MW of active hosting capacity. They are also a 25% partner in a 60-MW data center in North Dakota that is currently under construction; Kungsleden is responsible for developing and managing the full 60-MW site, and their 25% ownership interest translates to 15 MW of owned capacity. Upon completion of the North Dakota data center (expected by end of July), they will manage 90 MW of power capacity across their portfolio of four facilities. In our view, the term “data center” does not do these facilities justice; they are sites with fit-for-purpose infrastructure designed solely to deliver high-density compute rather than store data.

Each of these sites boasts roughly similar unit economics: Kungsleden enters into multi-year power purchase agreements for electricity at approximately US$50/MWh, and subsequently resells the power with hosting services to high-quality bitcoin miner tenants (who we believe enjoy one or more of the “unfair advantages” we describe above) running latest generation machines at a weighted-average rate of US$73/MWh.

Kungsleden has developed these sites at industry-low cost (approximately US$170k/MW, versus recent Marathon Digital acquisitions priced at US$450-500k/MW) and fast deployment times. At these historical costs and an assumed US$23/MWh hosting margin, the expected payback on a new site is less than a year.

The Kungsleden team consists of talented bitcoin mining infrastructure developers with years of experience, hundreds of megawatts of data centers developed for Kungsleden and other entities, and the capabilities to expand into new verticals. In fact, Kungsleden has already identified over 100 MW of potential growth opportunities which could be used for bitcoin mining or other verticals.

Done properly–prudently structuring power and hosting agreements and selecting creditworthy and high-quality tenants–we believe the bitcoin mining hosting business offers more stable economics and higher risk-adjusted returns on capital than pure-play bitcoin mining. We believe someone will always be looking to deploy latest-generation machines quickly and with minimal capex. If the miner defaults on the hosting agreement due to the machines becoming unprofitable, Kungsleden as the host can take possession of them to mine for themselves, effectively moving down the cost curve by an increment equal to their hosting margin. To quantify this: the breakeven hash price for a Bitmain S21 machine (17.5 J/TH efficiency) hosted at a rate of US$73/MWh is US$30.66/PH/s/day; at a rate of US$50/MWh, that breakeven hash price drops 32% to US$21.00/PH/s/day. While such levels are only ~50% below current hash price of ~$50/PH/s/day (which is already near all-time lows), in bitcoin mining as in all commodity production, you don’t need to be faster than the bear–you just need to be faster than the guy next to you. Beyond hosting, we have plans to diversify the business into markets uncorrelated to hash price–more on this below.

The proposed terms of the merger would result in Kungsleden shareholders owning approximately 72.5% of the pro forma Company. We would continue to serve as CEO and President, respectively. Tom Masiero, Gavin Qu, and Matthew Kita would join Cathedra’s existing board of directors as it expands from four to seven.

The scale of the merger is substantial. We believe it presents an attractive deal for Cathedra shareholders for several reasons:

• Diversifying revenue streams. The introduction of a hosting business diversifies Cathedra’s revenue streams away from pure-play bitcoin mining and allows us to more predictably compound capital at attractive rates over the long term. It also gets us off the ASIC Hamster Wheel; we will not be forced to rely on perpetual dilution to maintain our margins or grow our asset base. Instead, we intend to develop low-cost hosting infrastructure for bitcoin mining and, in time, other compute-intensive use cases, which offers attractive cash yields and fast paybacks. We will grow by reinvesting cash flows and raising external financing selectively at attractive terms. And we will utilize the fixed-margin cash flow produced by the hosting business to accumulate bitcoin on our balance sheet as a reserve asset.
• Finding a niche. Together, Cathedra and Kungsleden can focus on our comparative advantage. We can’t compete with Marathon Digital to raise capital and grow hash rate. But with Kungsleden’s capabilities, we can develop and operate facilities at below-market cost and with fast time-to-market.
• Long rack space. “Rack space” (reliable power and data center capacity in which to run bitcoin mining machines), no longer ASICs, is now the scarce commodity in bitcoin mining. Publicly traded bitcoin miners alone have announced machine purchases amounting to 100 EH/s since October 2023^[1], 63 EH/s of which is expected to be delivered in or after Q2 2024^[2]. Assuming latest generation power efficiency (17.5 J/TH for Bitmain’s new S21), 100 EH/s implies ~1,750 MW of incremental rack space, most of which does not exist today. The machine manufacturers continue to surprise to the upside in their ability to secure foundry space and deliver large-scale orders. There are also many new hardware manufacturers seeking to breach the silicon bottleneck. Anyone with capital can order these mining machines. We believe developing the rack space to accommodate those machines is a much more difficult–and therefore valuable–piece of the bitcoin mining stack. Kungsleden has proven expertise in securing land; cultivating relationships with utilities and power producers (who are notoriously slow-moving and justifiably skeptical of our industry); negotiating power agreements; designing fit-for-purpose data centers; building out supply chains for the requisite construction, electrical, and networking components at competitive cost; and working with local labor to construct and deliver the project in a time- and cost-effective manner. The market prices being paid by large-scale miners for completed sites (US$450-500k/MW) reflect the difficulty of this process, which gives us comfort that Kungsleden’s expertise provides a durable competitive moat.
• Continued focus on geographic diversification. The age of the hyperscale grid-connected bitcoin mining farm is over. We believe a confluence of political, social, and economic factors will consign the GW-scale West Texas bitcoin mine to the annals of history. In the future, market share will go to smaller scale (10-100 MW) sites located closer to the energy resource. Kungsleden has pioneered this trend: their first three 10-MW sites are adjacent to underutilized substations in rural Appalachia, where they utilize excess power to the benefit of local utilities and ratepayers. Their fourth is behind-the-meter at a power generation facility in North Dakota, allowing them to interface directly with the power producer for greater reliability.
• Flexibility. Furthermore, Kungsleden’s development model offers us flexibility in tailoring our exposure to hosting and mining economics. Given the challenges inherent to bitcoin mining that we have identified above, we are reluctant to purchase new bitcoin mining machines under normal circumstances, particularly when the opportunity cost of doing so is developing additional rack space via Kungsleden. However, if particularly advantageous conditions present themselves, we reserve the option to opportunistically increase our hash rate exposure, or even acquire deployed hash rate from one of Kungsleden’s existing tenants which can be immediately redirected to Cathedra’s mining pool account (“turnkey hash rate”). Ultimately, this flexibility derives from Kungsleden’s development capabilities and our own discipline in refusing to participate in the mining arms race, pursuing hash rate growth for its own sake.
• Pipeline for growth. These capabilities paired with Kungsleden’s relationships result in a strong pipeline of high-return deals within the bitcoin mining vertical. As mentioned above, the Kungsleden team has identified over 100 MW of potential expansion opportunities for similar hosting sites. The stable margins these assets produce will allow for the pro forma Company to grow without relying on dilution alone.
• New verticals. Looking to the future, we are confident in the pro forma Company’s ability to evaluate and develop infrastructure for high-density compute across other verticals, specifically artificial intelligence. Bitcoin is one of a suite of digital technologies that will shape the course of the 21^st century. Procuring power, developing electrical infrastructure, and building and operating data centers cheaply and quickly will be a very valuable set of capabilities in the coming decades. It is natural for bitcoin miners to consider these higher value uses of energy, and more and more people are talking about this. The future calls for more compute.
• Team. Equally important to their attractive assets is the Kungsleden team. Co-founders Tom Masiero and Gavin Qu bring with them hundreds of megawatts of on- and off-grid data center development experience. They have global networks and supply chains, allowing us to procure and develop quality infrastructure at a lower cost than nearly anyone else in the industry.

In short, the transaction brings Cathedra a more stable, predictable, and scalable business model in hosting, along with the team to execute and a platform for accretive growth. As a bonus, the combined Company will also have the scale to pursue a U.S. listing, which we intend to do imminently upon closing the merger.

Outlook

This transaction has even greater strategic importance for the Company in the long term. When we look at the future of bitcoin mining and compute more generally, we foresee a bottleneck in power infrastructure.

We have already discussed why we believe bitcoin will continue its ascendence as a monetary asset. This will inevitably lead to further hash rate growth. But there are now new competing uses for power as well, some of which are proving to be more lucrative than bitcoin mining on a revenue per kilowatt-hour basis. The advent of highly-effective large language models (LLMs) and other forms of artificial intelligence are changing the landscape of the internet as we know it. In years past, we had been skeptical of promises of the “exploding market for high-performance compute,” but recent experience suggests the demand is finally real.

In Q1, our team began a high-performance compute pilot project, operating a small cluster of legacy GPUs and leasing out compute via an online marketplace. The results of this pilot have been promising. Notwithstanding the usual hiccups to be expected in any new endeavor, it has produced revenue equivalent to $0.42/kWh. Compare this to the $0.12/kWh produced by an S21 mining machine under current conditions (as well as the significant age gap between the two machines) and one begins to understand the hype surrounding the AI data center opportunity.

We are not oblivious to the fact that the AI boom of the last 12-24 months shares many characteristics of typical market bubbles. We will adopt a prudent, disciplined approach to exploring the vertical in hopes of finding another high value market for our kilowatt-hours.

Bitcoin mining and AI are alike in that the demand for compute can experience exponential growth quite quickly. But each also depends on physical infrastructure which is subject to much longer lead-times and the laws of physics. This latency between new demand and new supply (compute) will be made more acute by the current scale of these technologies. The low-hanging, low-cost power is being eaten up. We may soon hit a point where the insatiable demand for compute overwhelms available power resources. And it will take a long time to build the power plants and transmission infrastructure needed to bring the market back to equilibrium.

The world is going to need more power and more compute. The New Cathedra is aimed directly at this intersection. We will develop infrastructure for this digital economy, which we see being built with Bitcoin at its foundation.

Conclusion

The last two years have been long and arduous. We know they have not been easy for shareholders. But we feel convicted in this new direction and the capabilities of the combined team to execute against it. We are focused on developing unfair advantages and ignoring vanity metrics. We as management feel very strongly that this merger will transform the company to the benefit of all shareholders.

We will hold a shareholder meeting on July 22 to seek approval for the merger (among other matters) and expect the deal to close shortly thereafter. An information circular containing additional details on the transaction has been distributed to shareholders, and we will also offer a virtual investor presentation alongside Kungsleden management to the public on July 9. In the meantime, we will continue to provide periodic updates on the business. We thank you again for your continued patience and support.

AJ Scalia
Chief Executive Officer

Drew Armstrong
President & Chief Operating Officer

27 June 2024
Block height: 849,650

Disclaimer Regarding Forward-Looking Information:

This shareholder letter has been issued as a matter of interest to investors and other followers of Cathedra Bitcoin Inc.  This shareholder letter contains forward-looking information and Cathedra cautions readers that forward-looking information is based on certain assumptions and risk factors that could cause actual results to differ materially from the expectations of Cathedra.  Readers should not place undue reliance on forward-looking information.  Please refer to those risks set out in the public documents of Cathedra filed on www.sedarplus.ca.  Securities regulators including the TSX Venture Exchange have not reviewed the information disclosed in this shareholder letter and no securities regulator accepts responsibility for the adequacy or accuracy of this content.