The cryptocurrency mining industry, once laser-focused on Bitcoin's proof-of-work mechanism, is undergoing a profound transformation.
Seven of the ten largest publicly traded mining companies now generate revenue from artificial intelligence (AI) operations or high-performance computing (HPC), with the remaining three actively preparing similar initiatives.
This shift is substantiated by corporate earnings reports, SEC filings, and analyses from crypto media outlets like CoinDesk and The Block, highlighting a pivot from volatile crypto rewards to stable, high-margin data center services.
Miners are repurposing their vast data center infrastructures - originally built for energy-intensive hashing - to host graphics processing units (GPUs) and servers tailored for AI workloads. This diversification provides predictable cash flows, often surpassing mining profitability amid Bitcoin's price fluctuations and halving events.
As energy costs remain a dominant expense (typically 60-80% of operations), leveraging existing power contracts for AI hosting maximizes asset utilization without proportional capital outlays.
Benchmark-Setting Deals and Revenue Metrics
A prime example is TeraWulf, which has secured 10-year contracts with Fluidstack, a UK-based AI cloud platform. These agreements yield approximately $1.85 million in annual revenue per megawatt (MW) of deployed capacity - an industry benchmark cited in TeraWulf's Q2 2024 earnings call and investor presentations.
For context, at current Bitcoin prices around $90,000 and post-halving difficulty levels, traditional mining generates $1-1.6 million per MW annually, per models from firms like Compass Mining and Luxor Technologies. AI hosting, by contrast, can command up to $2 million per MW, driven by demand for GPU clusters in training large language models and inference tasks.
TeraWulf's deal underscores the economics: Fluidstack leases colocation space in TeraWulf's nuclear-powered facilities in New York and Maryland, benefiting from low-cost, zero-carbon energy. This not only stabilizes TeraWulf's balance sheet but also positions it as a hybrid energy-infrastructure player, with AI contributing over 20% of projected 2025 revenue.
Leading Players and Their AI Pivots
Core Scientific exemplifies aggressive expansion. Emerging from Chapter 11 bankruptcy in early 2024, the company has inked multiple HPC contracts, including a $3.5 billion deal with CoreWeave (an Nvidia-backed AI cloud provider) for 200 MW of infrastructure.
This is expected to deliver $150-200 million in annual revenue starting in 2025, dwarfing its mining segment. Core Scientific's facilities in Texas and Georgia now allocate 300+ MW to AI, with GPU hosting margins exceeding 70% versus mining's 40-50%.
Bitdeer, backed by Bitmain founder Jihan Wu, operates similarly in Asia and the U.S. Its Q3 2024 report revealed AI cloud services generating $20 million quarterly, up 300% year-over-year, from hosting Nvidia H100 GPUs for clients like startups in generative AI. Bitdeer plans to dedicate 500 MW to HPC by 2026, targeting $1 billion in cumulative AI revenue.
CleanSpark, known for sustainable mining, has converted idle capacity post-halving. In Nevada, it hosts AI servers for undisclosed partners, reporting $5-7 million in Q2 2024 HPC revenue - nearly 15% of total earnings.
Iris Energy, with hydro-powered sites in Canada and Texas, secured a $500 million GPU financing deal in 2024, projecting AI to contribute 40% of revenue by mid-2025.
Other notables include:
- MARA (Marathon Digital): Partnering with Applied Digital for 50 MW of AI colocation, expected to yield $80-100 million annually.
- Riot Platforms: Allocating 100 MW in Texas for HPC, with pilots generating $1.7 million per MW in early tests.
- Cipher Mining: Developing AI infrastructure in partnership with data center firms, aiming for 20% revenue diversification by 2025.
The three laggards - Hut 8, Hive Digital, and Bitfarms - are in preparation mode. Hut 8 announced a $200 million AI data center joint venture in October 2024, while Bitfarms explores GPU hosting in Paraguay's low-cost hydro regions.
Also read:
- Phoenix from the Ashes: Three Years Since the FTX Crash and Bitcoin's 'Bottom'
- The Rise and Fall of Wearable AI: Why the Future Isn't in Your Pocket
- Jack Conte's Latest Patreon Updates: Discovery Feeds, Quips, and the Social Media Paradox
- Earn Cryptocurrency While Browsing: A Simple Way to Make Quasacoin
Investor Sentiment and Valuation Shifts
Analysts from JPMorgan and Bernstein now incorporate "AI-adjusted MW" into valuation models, alongside traditional hashrate metrics. A miner's AI exposure correlates with higher multiples: companies with >20% HPC revenue trade at 1.5-2x premiums to pure-play miners, per Bloomberg data.
This reflects AI's demand surge - global GPU shortages have pushed hyperscalers like Microsoft and Google to seek third-party hosting, with miners offering turnkey, high-power-density solutions.
Challenges persist: GPU procurement requires upfront capital, and contracts often include revenue-sharing or power-curtailment clauses. Regulatory scrutiny on energy use could cap growth, though miners' expertise in grid management provides an edge.
In summary, AI is not a side hustle but a core strategy for mining giants. With Bitcoin mining yields compressing (post-2024 halving reduced block rewards to 3.125 BTC), AI hosting offers resilience.
As one executive noted in a Riot earnings call, "We're evolving from crypto miners to digital infrastructure providers." This hybrid model could sustain the sector through future cycles, blending blockchain's decentralization with AI's computational explosion.
Author: Slava Vasipenok
Founder and CEO of QUASA (quasa.io) - Daily insights on Web3, AI, Crypto, and Freelance. Stay updated on finance, technology trends, and creator tools - with sources and real value.
Innovative entrepreneur with over 20 years of experience in IT, fintech, and blockchain. Specializes in decentralized solutions for freelancing, helping to overcome the barriers of traditional finance, especially in developing regions.
This is not financial or investment advice. Always do your own research (DYOR).
