Summary We believe that IREN is a policy change away from becoming a viable investment. This article explains how IREN's no-HODL is holding back profitability, destroying shareholder equity, and potentially rendering its thousand expansion efforts irrelevant. IREN's cost basis with the newer fleets and HPC can rival CLSK, but its profitability profile remains inferior because CLSK can sell Bitcoin at potentially higher prices in the future. Overall, IREN's no-HODL policy is a dealbreaker for us as it is detrimental to its profitability prospects and valuation. Introduction There's a saying: A single mistake can undo a thousand good deeds We think that this saying reflects our thesis for Iris Energy ( IREN ) quite accurately. IREN was an underdog just two years ago. It grew its capacity from 0.8 EH/s to 21 EH/s (September update) while maintaining operation cost efficiency around the sector average (Fig 1). Fig 1. All-in Cost Basis of Major Bitcoin Miners. It includes power cost, depreciation SG&A, financial costs, and other costs not related to impairment. One-off expenses/income such as BITF's 2024Q2 tax recovery are pro-rated to previous periods. (Author) On the growth side of the equation, IREN remains as the fastest-growing Bitcoin mining company in the sector now and into the near future. IREN is still keeping its momentum with expansion to 30 EH/s fully funded and a visible pathway to 50 EH/s via purchasing options . As a result, as of 2024Q2 (not the monthly update), IREN gained 233% more network share (relative Bitcoin network) compared to 2022Q4 figures, while Bitfarms ( BITF ) and Riot Platforms ( RIOT ) lost a quarter of their market share (Fig 2, Fig 3). Table 1 also shows that IREN is expected to outgrow competitors in the near future. Table 1. Expansion Guidance of Major Bitcoin Mining Companies Company BITF CLSK IREN MARA RIOT Latest Capacity 11.30 27.6 21 36.9 28.2 Guided Capacity 21.00 50.00 40.00 50.00 41.00 Guided Growth (%) 86% 81% 90% 36% 45% Guidance Timeline 2024Q4 2025Q4 2025H1 2024Q4 2025Q4 This rate of capacity growth is important to gain network share because the Bitcoin reward (production) is based on one's capacity relative to the sum of capacity of all miners. Fig 2. Normalized Change (%) of Relative Network Share of Major Bitcoin Miners. The Relative Network Hash Rate was used instead of the Bitcoin Total Network Hash Rate for comps in order to better present the growth between the miners under consideration. All revenue-related computations still uses the Total Bitcoin Network Hash Rate. (Author) Fig 3. Relative Network Share of Major Bitcoin Miners (Author) On the cost side of the equation, although the all-in cost basis per Bitcoin throughout the sector has increased after the halving event in April 2024, IREN still maintains its sector average cost efficiency, not as good as CleanSpark, but better than the rest, namely Riot Platforms ((RIOT)), Marathon Holdings ( MARA ), and Bitfarms ((BITF)). Overall, IREN looks promising and has gained positive attention among investors. However, we identified a fundamental problem we perceive as a non-negotiable dealbreaker for investing in IREN. The good news is this problem could be easily reversible by the board. Hence, we remain hopeful. The Mistake That Undoes A Thousand Efforts IREN's business model revolves around liquidating all Bitcoins mined and shareholder dilution to fund operations and expansions. Many analysts and investors will attribute the advantage of such a policy to being debt-free. This pursuit of a debt-free policy could be a side effect of a past shadow , where it lost 3.6 EH/s worth of operation due to a loan default. Whether this is true is not relevant. What's relevant is that the adverse consequences of this policy outweigh the benefit of being debt-free. Irreversible Realized Losses The first adverse consequence of a no-HODL policy is the irreversible losses from Bitcoin mining. IREN reported a net loss of $24.2mil in 2024Q2 or a $29,500 loss per Bitcoin mined (divided by 821 Bitcoins mined during the period). Although all major Bitcoin mining companies made losses in 2024Q2, IREN made different types of losses. IREN's losses differ from those of other Bitcoin mining companies because its losses are fully realized and irreversible. We highlighted this problem in our buy thesis for MARA , where we looked into whether an unprofitable Bitcoin mining company can justify a price-to-book value ratio ('PBR') greater than 1. We highly recommend you go through our rationale for full context. In short, we explained that PBR>1 indicates an underlying expectation for profitability. Yet, PBR>1 is okay for unprofitable Bitcoin miners because they can HODL the Bitcoins produced and sell them when the price appreciates above the cost incurred for their production. As long as the Bitcoins are not sold for cash, the profits and losses are not realized and has the opportunity to turn a profit. It is like an active trade position in Bitcoin. For example, if IREN HODLs Bitcoin, IREN can wait until the Bitcoin price appreciates above $92,500 (its all-in cost basis per Bitcoin in 2024Q2) before selling the 801 Bitcoins mined in 2024Q2. This essentially reverses the reported losses in 2024Q2 into gains on digital assets. Unfortunately, IREN's no-HODL policy inhibits it from doing so. Depending on our Bitcoin outlook, this policy may pose a significant problem. If you expect Bitcoin to trade above its cost basis (like in the past), which is $92,500, IREN's no-HODL policy shouldn't concern you. However, based on our expectations (peaking at $100,000 before an 85% drop during the bear market ), IREN's all-in cost basis of $92,500 per Bitcoin is unfeasible. This unprofitability can lead to a bigger problem: IREN's aggressive 50 EH/s expansion may become irrelevant, essentially undoing its thousand-effort. The Undoing of its Thousand Efforts We stressed that it is not about having the largest absolute mining capacity ('EH/s'); it is the growth rate that counts. In this aspect, IREN is putting tremendous effort into what counts, which saw it increase network share by 233% compared to the major Bitcoin miners and Bitcoin production. However, these growth efforts could be irrelevant if Bitcoin cannot be produced profitably because producing more Bitcoins at a loss will only worsen profitability. Some might argue that IREN's expansion will include newer mining fleets with better efficiency. Still, our findings indicate that the improved efficiency is insufficient for IREN to turn profitable with a no-HODL policy. IREN expects its mining fleet efficiency to improve by 40% from the current 25 J/TH to 15 J/TH with the new mining fleets. This will reduce its cost of revenue (electricity) down from $25.7mil (2024Q2) to $15.4mil, which will reduce total cost from $75.9mil (or $92,500 per Bitcoin) to $65.6mil (or $80,000 per Bitcoin). IREN remains unprofitable even if the full cost savings from the newer mining fleets are captured. Hence, IREN's effort to expand may yield no fruit (profit) because of the no-HODL policy. Then what about the HPC business segment? Can it help solve the irreversible loss problem? Contributions of HPC Our findings show that IREN's HPC can be a cost-saving measure. Since IREN is hosting its HPC operations and Bitcoin mining operations under the same roof, we don't expect SG&A costs to scale as much as HPC capacity. So, the two cost drivers would be electricity and depreciation. Electricity is a non-issue because the hardware margin is 98%. For depreciation, assuming a 5-year useful life on the $76mil ( $ 10mil + $22mil on H100 + $44mil on H200) GPU investments, the annual depreciation cost per quarter would be $15.2mil. So after depreciation, expected annual profits will drop from $34mil to $19mil (or $4.75mil per quarter). Based on 2024Q2 Bitcoin production, this translates to $5,800 cost savings per Bitcoin ($4.75mil divide by 821 Bitcoin). Assuming the expected $80,000 cost basis per Bitcoin, IREN's HPC segment can further push cost down to $74,200 per Bitcoin. It remains to be seen whether this is enough to make IREN the most efficient in the sector because CLSK, the reigning champion of mining efficiency, will also update its mining fleets with newer models. Nevertheless, IREN's HPC segment is expected to help it achieve similar cost basis as CLSK, which is the best in the sector (Fig 1). However, their profitability profile differs drastically because CLSK has the opportunity to sell Bitcoin at potentially higher prices in the future thanks to its HODL policy. Therefore, we can see that IREN's no-HODL policy is lowering profitability, potentially to a point of rendering its expansion efforts irrelevant. Unrecoverable Shareholder Equity Another problem with IREN's no-HODL policy is about shareholder dilution. Our previous thesis mentioned that any form of margin of safety (based on book value) is expected to be temporal due to the lack of un-depreciable assets (Bitcoin). On the flip side, IREN's book value also does not have room for appreciation due to the lack of appreciable assets (Bitcoin). For context, all these major Bitcoin miners increase outstanding shares almost proportionally (aside from CLSK) over the past 3 years. But only MARA and RIOT managed to recover their shareholder equity (Fig 4). Note that during the Bitcoin bear market (Fig 4), every major Bitcoin miners lost shareholder equity during the Bitcoin bear market. What MARA and RIOT had in common was their large Bitcoin reserves. This is also a contributing factor for our bullish thesis for MARA . Fig 4. Adjusted Book Value Per Share of Major Bitcoin Miners (Author) Fig 5. Proportion (%) of Bitcoin over Total Asset (Author) (Author) Fig 6. Normalized Change (%) Outstanding Shares of Major Bitcoin Miners (Author) Unrecoverable book value is also highly detrimental to valuation. By referring to Fig 7, we observe an ongoing sector-wide mean reversion on PBR where IREN is now trading at the same PBR as the likes of MARA, which has high proportion of Bitcoin on Balance Sheet. This is where the same water that hardens the egg, softens the potato. This mean reversion offers fundamental upside for MARA but significant downside risk for IREN. We explained that PBR>1 implies expectations for an appreciable book value or a high profitability prospect, both of which IREN does not currently possess, again, due to its no-HODL policy. Fig 7. Sector Mean Reversion on PBR (Author) Can HPC help with valuation? Unlikely as per IREN's own guidance. IREN only expects a 2.5x earnings multiple on the GPUs. IREN stated that it'll take 2 years to recoup the CAPEX on GPUs while the GPUs only have an expected 5-year useful life. The 2.5x earnings multiple on a $19mil profit only contributes $47.5mil of market cap or 3% of IREN's current market cap. Also, many people might argue HPC is a profit-guarantee business. Who would charge a rate lower than the cost? Well, we only have to look as far as RIOT for making significant losses QoQ in its hosting segment. Hence, the HPC segment is not without its risks and does not contribute to IREN's valuation. Verdict IREN is performing exceptionally well on what matters: growing capacity fast while keeping cost basis low. What we think is lacking is a decision to abolish its no-HODL policy. We think that IREN, with a HODL policy, can become one of the most exciting and attractive investments in Bitcoin mining. In this article, we showed how IREN's no-HODL is holding back profitability, destroying shareholder equity, and potentially rendering its 'thousand efforts' (to expand) irrelevant. It is also why IREN's no-HODL is a dealbreaker for us. Therefore, investors of IREN should pay close attention to this particular catalyst, the day IREN HODLs Bitcoin. Meanwhile, IREN could still pump more than other miners on the back of exciting headline numbers (capacity, HPC, etc), but we find little that can support a buy thesis when things go south under current circumstances. Again, we think that all it takes is for IREN to turnaround is a decision from the board. Until then, IREN will only live in our watchlist. For now, we're only looking to tilt our portfolio into MARA and CLSK when we receive our 2 confirmations for a potential bull run by 17th October.
