Summary I see IREN as a high-upside, favorable risk/reward play in AI and HPC, similar to early SuperMicro, with >100% upside potential. IREN's unique advantages—massive low-cost power contracts, strong cash flow from BTC mining, and operational flexibility—give it leverage over even the best neo-cloud companies like CRWV and NBIS. Despite a technical knowledge gap, IREN's profitability and power advantage should allow it to close the competency gap and achieve a much higher valuation. Key risks include rapid AI tech shifts reducing power demand, but current low valuation and strong EBITDA from BTC mining provide downside protection. As someone who wrote up SuperMicro when the stock was at $10, I believe that IREN now represents a similar opportunity in terms of favorable risk/reward caused by AI spending. At the time SuperMicro’s valuation was also about $5B, similar to IREN today. The core of my thesis is that IREN has a distinct advantage in contracted power, cash flow, and flexibility, that give it leverage over nearly all other players in the HPC colocation/neo-cloud space, and the chance that IREN therefore emerges as a long term neo-cloud winner is significant. This means that IREN should probably trade >100 % higher. The IREN advantages IREN enjoys important advantages compared to Coreweave ( CRWV ) and Nebius ( NBIS ), IREN has more power, at lower cost, and a large BTC cash flow tailwind. IREN also enjoys tremendous flexibility because it has a lower cost structure and can allocate resources to either btc mining, colocation, or its cloud business, as is most favorable to them. CRWV and NBIS have none of these flexibility advantages. The value of flexibility is enormous as the environment is extremely fast moving and hard to predict. CRWV and NBIS are potentially forced to make large investments in hardware that may prove suboptimal, IREN has no such pressure. How did IREN arrive at these advantages? IREN is one of the most unique stories in the market. Through a series of prescient foresights and some luck, Dan Roberts cobbled together 3+GW’s of mostly renewable energy power contracts, 50 exahash of bitcoin miners, and so far about 10K NVDA GPU’s (but this is likely to grow quickly). This puts them in a position where they are profitable at roughly 8x EBITDA with a growing ~$1B revenue stream that comes from a mix of BTC mining and renting GPU’s. Dan Roberts worked in renewable energy infrastructure investment banking and noticed tremendous amounts of renewable energy being built to satisfy government incentives without population centers nearby to consume it. Taking some time to think through how this unused resource could be allocated he came up with bitcoin mining, and then HPC. In the late 2010’s he founded IREN and started contracting large amounts of renewable energy. During this time it was not yet evident that AI would cause a massive power crunch by the year 2025, so they focused on building a large bitcoin mining business. Coming from behind as the other bitcoin miners started before they did, IREN managed to become the best publicly traded bitcoin miner as it has the lowest cost/btc of any company with more than 50 Exahash, and the other bitcoin miners mostly do not have any AI diversification, with the exception of Core Scientific ( CORZ ) and Terawulf ( WULF ) , who have much higher costs/btc. The takeaway is that IREN used its contracted power advantage to become arguably the best BTC mining company out of nowhere. The question is, are they now going to use their contracted power advantage and their btc cash flow advantage to achieve a position in the neo-cloud industry similar to their advantage in the bitcoin industry? I believe it’s worth betting on as the upside is significant, and likely to happen in my opinion, and the downside is protected by the demand for colocation that the WULF and CORZ deals demonstrate. Briefly, the upside is that IREN becomes more valuable than Nebius ( NBIS ). I believe this is very possible because IREN is much more profitable currently, and has power contracts for >3x more power than NBIS giving them a significant margin advantage as they grow their AI business. The downside is covered by the need for AI colocation contracts similar to the ones announced by CORZ and WULF, with bitcoin mining as added downside coverage, which would leave IREN around the current share price or probably higher as IREN has so much more capacity then CORZ and WULF combined. Why is IREN so cheap? IREN has a significant technical knowledge disadvantage vs NBIS or CRWV. This is not an issue regarding bitcoin mining or HPC colo. However it does add some uncertainty regarding IREN’s ability to become a neo-cloud. However the value discrepancy is so large currently that it much more than discounts this relative weakness. With IREN’s profitability advantage compared to NBIS or CRWV they can afford to overpay to acquire technical talent and still come out ahead. Another reason IREN is so cheap is because they have only started to seriously consider the neo-cloud business as a primary strategic objective in the past few months, so the investment markets have simply not yet reacted to the change. With Jane Street accumulating a significant stake in CRWV, the neo-cloud industry looks like it’s here to stay, which was not clear to everyone even a few months ago. SemiAnalysis ClusterMAX Rating System The key factor holding back IREN’s valuation is its inexperience, smaller size, and reduced technical knowledge compared to other neo-clouds. This is reflected in SemiAnalysis’s ratings. I believe that IREN’s cash flow and contracted power advantaged are large enough to allow them to change this. It is extremely early days for the industry, and a low technical capability should be viewed as more of a growing pain rather than a permanent disadvantage. By this very rating system Google Cloud itself is only rated Bronze. SemiAnalysis noted when they released the rating system that “CoreWeave is the only non-hyperscaler currently that is experienced at operating large-scale 10k+ H100 clusters reliably.” IREN now has approximately 10k GPUs after the announcement this week. IREN vs CRWV I am not going to argue that IREN should be valued to similarly to CRWV as it would be arguing for a 10x in IREN’s stock price. However IREN has 3GW power vs 2.2GW for CRWV, dramatically less debt, better flexibility, and better cash flow. In defense of a much higher valuation for CRWV, they are making tremendous investments that IREN has not even sort of begun to scale into, IREN’s AI investment are orders of magnitude less than CRWV’s currently. I do believe that IREN has some advantages relative to the much larger CRWV and that should be reflected in IREN’s valuation. One extremely important thing to understand is that Nvidia ( NVDA ) is a significant shareholder of CRWV. This is a major strategic advantage that IREN does not currently have. IREN vs NBIS This is an extremely striking comparison. NBIS has a market cap of $17B, and is worse on every financial metric than IREN. The market values NBIS at such a premium because they are viewed as having the competency to make a neo-cloud (they received a ClusterMax rating of gold). They also have a very large technical team with more than 1000 highly technical employees from Yandex. Similar to CRWV, NVDA is an investor in NBIS. My belief is that IREN’s strategic and financial advantages are so large that the competency gap will close over a few years and IREN should enjoy a similar valuation as NBIS, or 3x higher than the current market cap. Some striking facts about NBIS; NBIS is still in the process of securing 1GW by the end of 2026 , so IREN has more than 3x as much secured power. NBIS is guiding to ~$1B of ARR by the end of 2025, and recently achieved adjusted ebitda positive, while IREN is already at $1B revenue currently if you include BTC mining and has very high adjusted EBITDA margins. Their recently announced purchase of 4200 Blackwells will bring total GPU ARR to nine figures by year end, assuming no more GPU based deals, which there probably will be. It’s even conceivable than IREN will always have higher revenue than NBIS, although I wouldn’t bet on it. For that to happen IREN will have to have a colocation deal for at least >300MW plus an additional >10k GPUs of cloud business, or possibly a large rise in BTC. That’s possible but less likely to happen because IREN’s growth will be constrained by their technical knowledge disadvantage in 2026, but in the longer term their contracted power and flexibility advantage could be more important. IREN vs WULF Terawulf is valued just under $5B (adjusting for the dilution from the google deal). They have ~1 gigawatt to contract out, a bit less than 500 MW of which is currently contracted. Their megawatts are extremely high quality because they are some of the only available capacity that is predominately renewable, and large scale, close to NYC. They are doing a pure HPC colocation strategy, with btc operations that are roughly 4x smaller than IREN’s arguably too small to be material to their valuation. They have much less flexibility and upside optionality than IREN as they have no pathway to neo-cloud type operations, and no pathway to over $500m of BTC EBITDA. I would argue that the various qualitative puts and takes mean that IREN should trade at least on par with them on a $/MW basis, which would mean that IREN would trade at $15B. Management Background IREN’s management team comes from a background in managing infrastructure assets, ports, airports, solar farms, etc. The management teams of the other players come from more of a technology background. It may be that IREN’s tremendous outperformance vs all other BTC players, and superior position in HPC, come from this difference. Risks The big risk is an AI technological development that reduces the need for power. This would massively reduce the value of IREN stock, and IREN should definitely be considered speculative for that reason. However, it’s important to emphasize that IREN trades at a low multiple of EBITDA because it has tremendous EBITDA coming from BTC mining, this partially mitigates the risk. But this is a very fast moving space and there is significant risk. I believe the potential for a 3x return compensates you for taking this risk.
