Summary Bitfarms is pivoting from crypto mining to HPC/AI data centers, focusing on the Panther Creek campus as its main growth driver. BITF's Panther Creek project could generate significant revenue, but higher energy costs and a lack of a competitive moat may limit profitability. Valuation analysis suggests most future profits are already priced in, making BITF appear overvalued compared to peers like Applied Digital. Given speculative AI sector hype, I rate BITF as a hold—exiting positions but not recommending buying or shorting at current levels. Recap of the January 2025 Bear Case In January, I posted a bear case thesis for Bitfarms Ltd. (BITF). The bearish perspective was built on the thesis that crypto mining at that time was unsustainable due to the decline of the ''hashprice'' and that the company was constantly selling its mined crypto to cover its operating losses. In my opinion, that was a poor decision due to Bitcoin slowly appreciating over time. In the article, I also noted that when the company pivots to a data center play, this could mark an inflection point and that at that time I would close my short position. I did that but missed the opportunity to buy the stock, so in this article, I plan to once again look into Bitfarms and evaluate if it is still a good buying opportunity. Pivot to HPC Due to challenges with crypto mining being unsustainable, as I outlined in the previous article, Bitfarms is pivoting to HPC AI, which is currently in a ‘’gold rush''. In the Q2 2025 earnings, management stated that no further Bitcoin mining growth is planned and described the mining business as a source of cash flow to drive HPC/AI. They also sold their Paraguay mining operations to generate cash for investments in the USA and discontinued their operations at a mining facility in Argentina. Thus, management is basically ending its operations outside of the USA and is slowly exiting the mining business. The centerpiece of Bitfarms' transformation is the development of its Panther Creek campus in Pennsylvania into a large-scale HPC/AI data center. The Panther Creek campus is expected to deliver 50 MW by the end of 2026 and up to 350 MW by 2027. This project, acquired through the Stronghold merger , now represents the primary engine for future growth. I also wanted to note that Bitfarms has already secured financing by issuing senior convertible notes to develop the Panther Creek site. With cash secured, Bitfarms announced a partnership with T5 Data Centers to build its HPC data center. I want to go deeper into this project and try to estimate how profitable such a data center would be, based on the data from peers. I believe understanding this part of the business is crucial to being able to provide a fair valuation of the company. Valuation: Panther Creek, Pennsylvania The Panther Creek campus is planned to open in 2 phases: Phase 1 will bring 50 MW of capacity online by the end of 2026, and Phase 2 will add the remaining 300 MW as early as 2027. But building the infrastructure is only half the battle. Bitfarms must then go through a competitive process of securing long-term contracts with the likes of CoreWeave, Inc. (CRWV). One such contract happened recently when Applied Digital Corporation ( APLD ) secured a 15-year contract with CoreWeave for its Forge 1 campus. The deal for a 400 MW datacenter has an anticipated contracted lease revenue of $11B. This amounts to $27.5M per MW. If we translate this to Panther Creek, a fair contracted lease value would be around $10B for the whole campus when completed at 350 MW. I should also mention that Applied Digital uses cheap wind power from its wind farms, while Bitfarms needs to use waste coal or buy electricity from the grid. The problem I see here is that the power plants that Bitfarms has in Pennsylvania, the Scrubgrass Plant with 85 MW nameplate capacity and the Panther Creek Plant with 80 MW nameplate capacity, are not enough to cover their energy needs, and the rest of the power will need to be bought. Plus, I believe mining waste coal or buying coal is not nearly as cheap as operating wind farms and getting cheap wind energy like Applied Digital does. In my opinion, the contracted price for Bitfarms’ datacenters will not be nearly as high per MW as in Applied Digital’s case. How much lower? Hard to say. I will estimate the contracted price per MW to be 30% lower than the Forge 1 campus deal. This would mean a total contracted lease value of $7B. Next is the net margin. If the project’s net margin is 30%, which is really generous in my opinion, Bitfarms will be able to get $2.1B of net profits from the project in the period of 15 years. Current Bitfarms’ market cap is 2.8B. This means profits from their most important project will not even be able to “pay” the current market cap. One could argue that they will do additional projects, which is true, but the problem is that they do not construct their data centers (they have a partnership with T5 centers); additionally, they don’t have any power plants and will have to buy electricity, nor do they have a well-known GPU-as-a-service platform like CoreWeave does. So, what moat do they have? In my opinion, none. Once there is enough data center supply, their margins should be low. I still want to use another method of evaluating the value of the project. According to the CBRE report , Phoenix data centers’ rent pricing was stable year-over-year at an average of $190/kW/month following rapid gains in prior years. I believe that due to the price stabilization, supply might be catching up with the demand; thus, I will estimate that in 2027, when Bitfarms’ project is completed, rates will remain at a similar price or maybe even lower. This would mean that their campus at 350 MW would bring in $66M per month, which is around $800M annually. At a 30% net margin, that would mean 240M net profit annually. Once again, a 30% net margin for a company that has little moat and thus pricing power is really generous. At PE 15, which is, in my opinion, reasonable due to the high degradation of AI chips and them being outdated after 3-5 years , a fair valuation for the company would be 3.6B. If I am a bit more conservative and use a net margin of 15%, the company should be valued at 1.7B. Once again, the current market cap is 2.8B. If I sum it all up, I believe most of the potential profits are already priced in, and thus, I do not see much upside. I believe that the comparison of Bitfarms with Applied Digital’s CoreWeave deal is a better valuation method, and according to that method, Bitfarms is currently overvalued. Plan As stated above, I believe the company is currently overvalued, but on the other hand, lots of companies in the AI sector are experiencing a lot of hype. Thus, the stock might just keep rising due to speculative investing. As such, I would not dare to short this stock or do it with a small amount and hold it until the current mania wears off, which at some point certainly will happen. But as Warren Buffett said, "Markets can remain irrational longer than you can remain solvent." Thus, I will rate it as a hold, which for me means exit the position but would not buy or short.