Mining is not as good as calculating AI! Bitfarms abandons Bitcoin CEO: The truth behind the 25 times revenue transformation.

Bitfarms, one of the largest Bitcoin mining farms in North America, announced that it will cease Bitcoin mining in the next two years and fully invest in high-performance computing (HPC) and artificial intelligence infrastructure services. The company reported a loss of $46 million in Q3 2025, with the cost of mining a single Bitcoin Cash reaching $82,400. In contrast, AI hosting contracts can be valued at up to $6 million per MW, with revenue potential several times higher than mining.

Mining Profit Collapse Drives Transformation Wave

(Source: Bitfarms)

The volatility of Bitcoin prices and the increase in mining difficulty have made it more challenging to profit from “digital gold.” The direct reason for Bitfarms' transformation is the reduced opportunity cost of Bitcoin mining revenues. The company reported a loss of $46 million in the third quarter of 2025, with internal estimates indicating that the cash cost to mine a single Bitcoin has reached $82,400. When Bitcoin prices fluctuate around $100,000, this cost structure means that profit margins are severely compressed.

The surge in mining costs stems from multiple factors. First is the continuous increase in Bitcoin network difficulty; as more computing power joins the network, the share of Bitcoin rewards each miner receives continuously decreases. Second is the fluctuation of energy costs; although mining companies typically choose regions with lower electricity prices, the overall rise in global energy prices still translates into mining costs. Third is the pressure of hardware depreciation and upgrades; the technological iteration of ASIC miners is accelerating, and the efficiency advantage of old equipment is rapidly lost.

In contrast, the surge of generative AI has driven up the demand for GPU computing power. Industry reports indicate that the revenue potential of AI data centers per megawatt (MW) can be several times higher than that of Mining, while the value of AI hosting contracts can reach 6 million USD per MW. This significant revenue gap makes the transformation decisions for mining companies reasonable and urgent. CEO Ben Gagnon emphasized that although Washington's Mining Farms account for less than 1% of the total developable mining sites, transitioning to GPU as a service could generate a higher net operating income than traditional mining operations.

Bitfarms is not the first mining company to do this. Peers like Cipher and Terawulf are also collaborating with tech giants such as SoftBank and Google to layout AI, with about 70% of large miners having initiated diversified revenue plans. This ratio indicates that the mining industry is undergoing a structural transformation, shifting from a fragile business model that solely relies on Bitcoin prices to more diversified and predictable sources of income.

Technical Details of Washington State Mining Farm Renovation

According to reports, Bitfarms' first phase of transformation will focus on an 18 MW Mining Farm in Washington State, expected to be completed by December 2026, and has secured $128 million in funding. The technical specifications of this project demonstrate the seriousness of the mining company in transitioning to AI computing power supply. The hardware core introduces the Nvidia GB300 GPU, which is Nvidia's latest generation high-performance GPU designed specifically for AI training and inference.

The company will adopt a liquid cooling system, with a rack power density of up to 190 kW, and will control the PUE (Power Usage Effectiveness) between 1.2 and 1.3. These technical indicators are at a leading level in the data center industry. Traditional air-cooled data center PUE is usually above 1.5, while liquid cooling systems can dissipate heat more effectively, reducing energy waste. A rack power density of 190 kW means that more computing power can be deployed in the same space, improving space utilization efficiency.

Bitfarms Washington State Mining Farm Transformation Core Indicators

Capacity Scale: 18 MW (approximately less than 1% of the total development portfolio)

Investment Amount: 128 million USD

Core Hardware: Nvidia GB300 GPU

Cooling System: Liquid cooling technology, rack power density 190 kW

Energy Efficiency: PUE controlled between 1.2 and 1.3

Completion Time: Before December 2026

Faced with the future shift of the business model from power consumers to power suppliers, Bitfarms will offer GPUaaS (GPU as a Service) and hosting services. This business model is similar to cloud services, where customers can rent GPU computing power on demand for AI model training or inference, without the need to purchase and maintain expensive hardware themselves. The company estimates that the overall investment is less than 1% of the total development portfolio, yet it can release tremendous production capacity, demonstrating that the inherent energy and data center advantages of the Mining Farm are sufficient to support AI upgrades.

The Natural Advantages and Challenges of Mining Farm Transformation to AI

The transformation of the mining industry to AI computing power supply is not a coincidence, but based on several natural advantages of mining farms. First is the power infrastructure; mining farms are usually built in areas with abundant and low-cost electricity, and have established large-capacity power supply systems. These infrastructures can be directly used to support high-power GPU clusters. Second is the server room environment; mining farms already have good ventilation, temperature control, and security systems, all of which are equally critical for data center operations.

Third, there is operational experience. Mining companies have accumulated rich experience in managing large-scale hardware, optimizing energy usage, and maintaining system stability. These capabilities can be directly transferred to the operations of AI data centers. Fourth, there is geographical location. Many mining farms are located in remote areas where land costs are low and there is ample room for expansion, which is an important advantage for data centers that require a lot of physical space.

However, the transformation is not an easy task. The cost of 8 to 11 million USD per MW means that Bitfarms must manage its capital properly to avoid project delays or performance not meeting expectations. This high capital expenditure puts pressure on cash flow, especially during the transition period when the mining business has not fully exited and new businesses have not generated stable income. In addition, the company will compete directly with cloud service providers such as Amazon AWS, Microsoft Azure, and CoreWeave.

These cloud giants possess mature technology stacks, globally distributed data center networks, and strong brand influence. AI computing power suppliers transitioning from mining companies need to find differentiated competitive advantages, which may include offering lower prices (benefiting from lower electricity costs), more flexible contract terms, or focusing on AI applications in specific verticals.

After the announcement, along with the drop in Bitcoin prices, Bitfarms' stock price fell by 18.47% last night, reflecting the market's concerns about execution risks. Investors' doubts mainly focus on several aspects: whether the transformation can be completed on time, whether the new business can quickly acquire customers, whether it can survive in competition with cloud giants, and the financial pressure during the exit period of the traditional Mining business.

The Future and Insights of the Mining Industry

Bitfarms' transformation decision reveals the deep challenges faced by the Mining industry. The Bitcoin halving mechanism means that the Mining rewards are halved every four years, which will inevitably lead to a decline in Mining profits in the long run, unless the price of Bitcoin rises at the same or a faster rate. However, the volatility of Bitcoin prices makes this assumption highly uncertain. When Mining costs approach or even exceed the price of Bitcoin, the survival space for Mining companies is severely compressed.

The explosive demand for AI computing power has provided mining companies with a new outlet. The popularity of generative AI applications like ChatGPT has driven an exponential growth in GPU computing power demand, while the global GPU supply remains relatively tight. This supply-demand imbalance has created significant market opportunities. The electricity, space, and operational capabilities that mining companies possess align perfectly with the needs of AI data centers, making the transition a natural choice.

About 70% of large mining operators have initiated diversified revenue plans, and this figure itself illustrates the formation of industry consensus. Mining is no longer a sustainable single business model, but must be combined with other businesses that have high computing power demands. Future “mining companies” may more accurately be referred to as “computing power infrastructure providers,” which can flexibly allocate computing power for mining or AI services based on market conditions.

Bitfarms leverages its existing power and site advantages to transition from Bitcoin Mining to AI data centers. Although capital pressure, technology integration, and market competition cannot be overlooked, if the timeline can be adhered to, stable GPU contract revenues will bring higher predictable cash flow to the company, opening up greater imaginative space for the overall mining industry towards AI infrastructure.