Bitcoin Miners Embrace AI and High-Performance Computing Amid Revenue Uncertainty

The Bitcoin mining industry is undergoing a significant transformation as miners increasingly turn to artificial intelligence (AI) and high-performance computing (HPC) to diversify their revenue streams. According to a recent report by VanEck, this shift is driven by the growing uncertainty surrounding long-term on-chain revenue growth for Bitcoin miners. While network congestion can lead to temporary spikes in transaction fees, the rise of off-chain solutions—such as exchange-traded funds (ETFs), futures markets, layer-2 (L2) solutions, and centralized exchanges—has diminished the reliance on traditional on-chain revenue. As block rewards halve every four years, Bitcoin miners face a pressing need to adapt, as the Bitcoin price would need to double to maintain current revenue levels. This dynamic has highlighted the importance of pivoting to AI and HPC as key diversification strategies.

From Bitcoin Mining to AI and HPC: A Strategic Pivot

Successful Bitcoin miners are integrating alternative strategies to optimize their revenue streams and remain competitive in a rapidly evolving landscape. One approach involves using mining operations to subsidize grid expansion in remote energy markets, where access to affordable and reliable power is limited. Others are leveraging their existing power infrastructure to support AI and HPC workloads, which offer higher margins compared to traditional mining. These workloads, which include AI model training and iterative computational tasks, require dense compute clusters and significant power resources, making Bitcoin miners’ infrastructure ideal for such operations.

Innovations in liquid cooling systems, chip design, and co-location technologies are further enhancing operational efficiencies, enabling miners to allocate their resources more effectively. The shift toward AI and HPC gained significant momentum following CoreWeave’s 700MW AI/HPC agreement in 2024, which served as a catalyst for several Bitcoin miners to explore similar revenue streams. Companies like Bitfarms (BITF) have engaged AI/HPC consultants to assess the feasibility of integrating these workloads across their North American sites. Similarly, Bitdeer (BTDR) has entered into ongoing discussions with AI/HPC development partners after completing a data center consulting engagement.

Cipher Mining (CIFR), Riot Platforms (RIOT), and HIVE Digital Technologies (HIVE) are also making strides in this space. Cipher Mining secured a $50 million investment from SoftBank in January to support its HPC data center expansion, while Riot Platforms expanded its board of directors to include expertise in AI/HPC investment banking, data centers, and real estate. HIVE Digital Technologies appointed Craig Tavares as President and COO of Buzz HPC, positioning the firm for growth in HPC and GPU cloud services. These moves underscore the industry’s commitment to diversification and innovation.

Scaling Electrical Capacity for AI and HPC Workloads

The pivot to AI and HPC has added a new layer of complexity to Bitcoin miners’ operations, particularly in terms of scaling electrical capacity. The report assessed 13 public Bitcoin miners and found a collective operational capacity of 7.1 gigawatts (GW), with expansion plans projecting an increase to 11.7 GW by 2025, 15.9 GW by 2026, and 20.4 GW by 2027. This represents a compound annual growth rate of 42% over three years. Beyond 2028, an additional 7.3 GW pipeline is planned, though this is considered a conservative estimate given the competitive nature of power procurement in Bitcoin mining.

Scaling these operations will require significant investment. Assuming the expansion through 2027 employs a modernized Bitmain Antminer S21 Pros fleet, priced at $5,000 each, alongside $450,000 per MW of supporting infrastructure, the total capital expenditure is estimated at $24.8 billion. However, Bitcoin miners are unlikely to dedicate all their capacity to mining, as securing electrical capacity has become a key competency amid growing AI power demands. Goldman Sachs estimates that AI currently consumes approximately 7.7 GW of global data center power usage, accounting for 14% of the total. By 2027, this figure is projected to increase to 22.7 GW (27%). Given this trend, Bitcoin miners are expected to allocate 20-30% of their electrical capacity to AI and HPC workloads, reflecting a strategic pivot toward more sustainable and diversified revenue streams.

Iris Energy (IREN) is advancing its AI and HPC initiatives with a 75 megawatts (MW) liquid-cooled AI/HPC data center set to launch in Childress, Texas, by the second half of 2025. Additionally, the company plans to expand its Sweetwater site with another 600 MW, bringing the total capacity to 2 GW and making it one of North America’s largest AI/HPC-eligible sites. The report noted the rarity of such large-scale sites and emphasized their strategic advantage in providing dense compute clusters essential for AI model training and iterative workloads.

Navigating the Energy and Computational Demands of AI and HPC

The integration of AI and HPC workloads into Bitcoin mining operations presents both opportunities and challenges. While these workloads offer higher margins and new financing avenues, they also require significant power resources and advanced infrastructure. Bitcoin miners are uniquely positioned to meet these demands, given their existing expertise in managing large-scale power consumption and computational operations. However, the industry must also navigate the growing competition for power resources, as AI adoption continues to accelerate.

Goldman Sachs estimates that AI currently consumes approximately 7.7 GW of global data center power usage, accounting for 14% of