HIVE Digital Technologies has crossed a new threshold in its “dual engine” strategy — Bitcoin mining and high-performance computing (HPC) for Artificial Intelligence. The company announced that it has surpassed 22 Exahashes per second (EH/s) of global mining capacity, representing a 267% growth this year, and is accelerating in Boden (Sweden) the conversion of a Tier-1 data center into a Tier-3 facility with liquid cooling dedicated to AI loads and GPU cloud. The goal: to gain market entry time and transform, in months rather than years, energy-efficient infrastructures into platforms ready for the new AI economy.
The 22 EH/s figure summarizes HIVE’s organic expansion and, especially, the boost provided by its third green campus of 100 megawatts in Valenzuela (Paraguay), powered by the Itaipu hydroelectric dam. There, the company locates a critical part of its ASIC fleet with an explicit focus on energy efficiency and cost predictability.
More Efficient Mining and Sustained Margins
HIVE estimates its current production at 9.5 Bitcoin daily, with a fleet efficiency of 17.8 J/TH and a mining margin of 55% net of electricity costs, assuming a Hashprice of $47. The operational takeaway is clear: disciplined consumption and energy arbitrage between locations enable resilience against crypto market volatility.
The company also anticipates reaching 25 EH/s ahead of Thanksgiving in the United States, with a target efficiency of 17.5 J/TH as it integrates new ASIC equipment. Management emphasizes that growth in EH/s is material for revenue and cash flow, in a sector where valuation multiples often pivot on revenue and cash flow.
From Tier-1 to Tier-3: Speed for the AI Cloud in Europe
In Europe, the strategic move involves converting a Tier-1 Boden center into a Tier-3 HPC with liquid cooling and robust critical load for 2,000 NVIDIA GPUs. According to HIVE, engineering and design are already complete, and construction will begin this quarter. This aims to reduce the time to cash flow to about 9 months, compared to the roughly 3 years typically required for a greenfield project built from scratch.
This decision is supported by two years of AI operations in Stockholm, where HIVE tested demand with corporate clients seeking low latency and green energy in Northern Europe. The experience, the company claims, validates the rapid conversion of existing assets to offer enterprise-grade AI capacity within the EU.
The move from Tier-1 to Tier-3 entails more redundancy and greater availability, critical for training and inference workloads of AI models that are sensitive to outages and thermal fluctuations. Liquid cooling — already gaining ground in new-generation facilities — enables higher densities per rack, essential as the TDP per GPU increases generation after generation.
A Global GPU Pipeline Through 2026
HIVE’s capacity plan extends beyond Sweden. In Toronto (Canada), its BUZZ division projects 2,000 additional GPUs in 2026, while preparing for growth in New Brunswick. Among these locations, the company claims to have secured energy and land for next-generation HPC operations powered by renewables.
Thanks to a placement agreement with Bell Canada, HIVE boasts agile infrastructure deployment for GPU scaling according to business demand. Combining these fronts, the corporate forecast is to operate around 6,000 GPUs by 2026, serving both training and inference needs.
Leadership Messages
For Frank Holmes, co-founder and CEO, surpassing 22 EH/s confirms the ambition to build one of the most efficient and sustainable Bitcoin mining fleets in the market. Simultaneously, the accelerated conversion of Tier-1 centers to Tier-3 reinforces HIVE’s thesis as a “digital power” company with two engines, connecting blockchain and the AI supercycle.
Holmes also highlights the historical role of mining: “Bitcoin miners were the first builders of Tier-1 digital infrastructure networks — substation, fiber, energy optimization — which now underpin AI hyperscale data centers.” As an example of this moment, he cites Stargate, a large-scale HPC campus project in Texas valued at $500 billion, which, in his view, replicates on a large scale insights from the sector focused on surplus or underutilized renewable energy.
From an operational perspective, Aydin Kilic, President and CEO, emphasizes that reusing infrastructure enables faster cash flow than greenfield projects, and that the organization is prepared to grow both mining and AI cloud services in parallel, maintaining the core focus on renewable energy.
From a country perspective, Johanna Thornblad, HIVE’s Sweden president, interprets the Boden expansion as a reinforcement of Sweden’s leadership in sustainable digital infrastructure. She underscores that, built on lessons from Stockholm, the project shortens the timeline to supply enterprise AI capacity to the EU market.
Funding, Revenue Visibility, and Goals
In Paraguay, HIVE details that all hardware is financed and delivered, with a fixed-rate hydroelectric tariff providing predictability. In the HPC segment, the company continues modeling recurring annual revenues (ARR) from AI and GPU cloud contracts, a non-GAAP metric focused on growth in billing rather than profitability.
The roadmap points to 25 EH/s by late 2025 and 35 EH/s in 2026, aiming to multiply HPC division capacity fivefold within the same period. The common vectors are renewable energy and computational density.
Implications for the Sector
The convergence of Bitcoin mining and AI data centers is no longer hypothetical. Firms with expertise in energy acquisition, management, and optimization — substations, stable PPAs, proximity to major renewables — are reimagining part of their footprint as HPC for AI. HIVE’s focus on Sweden and Canada responds to favorable temperatures, decarbonized electric mix, and established digital ecosystems.
If executed on schedule, the move from Tier-1 to Tier-3 with liquid cooling will place capacity on the market in less than a year — a crucial factor as GPU demand strains supply chains and electrical grids. The multi-country approach and renewables coverage aim to buffer volatility in energy prices, hashprice, and investment cycles.
Metrics and Disclaimers
HIVE defines “Mining Margin” as the profit after energy costs divided by mining revenue, expressed as a percentage. Its indicative estimate assumes $0.05 per kWh. The ARR metric refers to annualized revenue and does not imply profitability.
As with all forward-looking statements, the company includes prospective statements subject to multiple factors: project completion and performance, cryptocurrency volatility, regulatory environments in Canada, the US, and other countries, energy availability and costs, network difficulty, access to capital, technological and security risks, among others. The firm reminds that actual results may differ from expectations and that estimates reflect market conditions as of September 2025.
About HIVE
HIVE Digital Technologies was founded in 2017 as the first listed company to mine digital assets with 100% green energy. Today, it designs and operates data centers for blockchain and AI HPC in Canada, Sweden, and Paraguay, with an “dual engine” infrastructure: Bitcoin mining and NVIDIA GPU-accelerated computing for enterprise clients. Its core focus revolves around scalability, sustainability, and operational efficiency.
Frequently Asked Questions
What does surpassing 22 EH/s in Bitcoin mining mean for HIVE?
It indicates that its aggregate calculation capacity for solving Bitcoin blocks exceeds 22 Exahashes per second. Higher EH/s results in more network rewards likelihood, which can lead to more income, provided Bitcoin price, network difficulty, and energy costs stay favorable.
Practically, how does a Tier-1 data center differ from a Tier-3 for AI workloads?
Simplified: Tier-3 offers greater redundancy, failover tolerance, and maintenance windows without service interruption, essential for training and inference of AI models that cannot tolerate outages. Liquid cooling also provides higher density per rack, crucial for modern GPUs.
Why are Sweden and Canada ideal hubs for HIVE’s AI cloud?
Both countries combine cool climates, decarbonized energy mix, and mature digital ecosystems. This improves thermal efficiency, provides renewable energy, and offers predictable regulatory environments for deploying and scaling GPU cloud capacity.
How does Bitcoin mining align with AI computing in the same group?
They share core skills: large-scale energy management, electricity and network infrastructure, 24/7 operations, and thermal optimization. Reusing and converting some of these assets into HPC for AI can accelerate market entry and diversify revenue sources.