From Bitcoin Mining to AI Power: How Galaxy Built a $1B Business
Did you know a former Bitcoin mining facility in rural Texas is now powering artificial intelligence for one of the world’s largest GPU cloud providers? Galaxy Digital just completed the first phase of its Helios data center campus, delivering 133 megawatts of critical IT load to Coreweave—formalizing a $1 billion revenue path. This isn’t just a company pivot; it’s a signal of a massive structural shift in the crypto economy. For anyone tracking where crypto mining is heading, this story reveals how the lines between Bitcoin mining and AI infrastructure are blurring fast. This guide explains why miners are repurposing their sites, how these deals work, and what it means for the future of both industries.
Read time: 10-12 minutes
Understanding Crypto Mining to AI Conversion for Beginners
Crypto mining to AI conversion refers to the process of transforming facilities originally built for Bitcoin mining into data centers that power artificial intelligence computing. Think of it like converting a warehouse built for storing physical goods into a high-tech fulfillment center for online orders—the building stays, but everything inside gets upgraded for a different, more profitable purpose.
Why is this happening? Two main forces are driving the shift. First, Bitcoin mining margins have tightened significantly after the April 2024 halving, which cut block rewards in half. Second, AI companies are desperate for power and willing to pay premium rates for electricity. The facilities miners already own—with access to cheap power, open land, and substation connections—are exactly what AI tenants need.
A real-world example is Galaxy Digital’s Helios campus in West Texas. It started as a Bitcoin mining facility built by Argo Blockchain in 2021. Galaxy bought it for roughly $65 million during the 2022 crypto winter. Now, it’s anchoring one of the largest AI infrastructure leases ever signed, with Coreweave committing to 526 megawatts under a 15-year lease.
The Technical Details: How a Mining Facility Becomes an AI Data Center
Converting a Bitcoin mining site for AI is far more complex than just swapping machines. Here’s what actually happens:
1. Infrastructure Gutting: Crews remove immersion cooling tanks and ASIC racks built for mining. Bitcoin mining uses specialized chips (ASICs) that are efficient at solving cryptographic puzzles but useless for AI workloads.
2. Power System Upgrade: They install Tier III N+1 redundant power systems. This means there’s backup for every critical component, ensuring 99.982% uptime—essential for AI training jobs that can run for weeks.
3. High-Density Electrical Infrastructure: AI requires denser power delivery than Bitcoin mining. GPU clusters need more power per square foot than ASIC miners, requiring new electrical distribution systems.
4. Fiber Network Installation: More than 100 miles of fiber laterals connect the site to long-haul routes. AI workloads need massive data throughput, unlike Bitcoin mining which only transmits block solutions.
5. Cooling System Transformation: Immersion cooling tanks (where miners submerged ASICs in liquid) get replaced with systems designed for the extreme heat output of Nvidia H100 and B200 GPUs.
Why this structure matters: The high upfront cost—Galaxy used a $1.4 billion debt facility and $350 million in equity—is justified by the long-term revenue stability. AI tenants sign 10-15 year leases with contracted rates, unlike Bitcoin mining revenue that fluctuates with BTC’s price.
Current Market Context: Why This Matters Now
As of July 2026, the crypto-to-AI conversion trend has become a defining narrative in the mining sector. Galaxy Digital’s Helios campus delivered its first 133 MW to Coreweave on July 6, 2026, formalizing a revenue path exceeding $1 billion annually when fully built out.
The timing is crucial. Bitcoin mining margins have been squeezed by three factors since the 2024 halving: higher power costs, increased network difficulty, and falling transaction fees. Meanwhile, AI companies are in a power acquisition frenzy. Coreweave alone committed to 526 MW across three phases at Helios, covering the full 800 MW of gross power currently approved at the site.
Mike Novogratz, Galaxy’s founder and CEO, captured the sentiment perfectly: “The demand for high-density, AI-ready power is not a cycle; it is a structural shift, and Galaxy is built to meet it.” The economics back this up—Phase I alone is projected to generate about $4.5 billion in lease payments over 15 years, or roughly $300 million annually, with site-level EBITDA margins near 90 percent.
Competitive Landscape: How Major Miners Compare in AI Conversion
| Feature | Galaxy Digital (Helios) | Hut 8 | Bit Digital |
|---|---|---|---|
| Committed AI Capacity | 526 MW (Coreweave lease) | ~300 MW (various deals) | ~100 MW (GPU rental) |
| Revenue Model | Build-to-suit lease (pass-through costs) | Colocation + direct GPU services | Direct GPU leasing |
| Anchor Tenant | Coreweave (AI cloud provider) | Multiple AI startups | Institutional funds |
| Site Location | West Texas (Helios campus) | Alberta, Canada | New York, USA |
| Total Power Capacity | 3.6 GW target (1.63 GW approved) | ~500 MW total | ~200 MW total |
| Key Advantage | Scale, proximity to power, 15-year lease | Low power costs, regulatory clarity | High-margin GPU services |
| Risk Factor | Construction delays, tenant concentration | Crypto price exposure still high | Small scale, competitive GPU market |
Why this matters: Galaxy’s deal with Coreweave is the largest of its kind. The 15-year lease provides revenue visibility that pure Bitcoin mining can’t match. However, it requires massive upfront capital—Galaxy raised $1.4 billion in debt and $460 million from an outside asset manager to fund construction.
Practical Applications: Real-World Use Cases
How are these converted facilities actually used?
- AI Training Infrastructure: Coreweave runs its own GPUs (Nvidia H100, B200) at Helios, training large language models and computer vision systems for clients like OpenAI and Stability AI. The benefit? Stable, contracted power costs instead of volatile crypto mining expenses.
- Inference Processing: Beyond training, these data centers handle AI inference—the real-time processing of user requests (like generating images or answering questions). Inference needs consistent, low-latency power, which these facilities provide.
- Cloud Gaming: GPU-heavy sites also host cloud gaming services. Companies like NVIDIA’s GeForce NOW could lease capacity, competing with traditional data center providers.
- Scientific Computing: Academic and pharmaceutical researchers use these GPU clusters for protein folding, drug discovery, and climate modeling—workloads that require massive parallel processing.
Who benefits most? AI cloud providers gain access to power-constrained regions where new data center construction takes 3-5 years. Miners get steady, contracted revenue without Bitcoin price exposure. Local communities gain jobs and tax revenue from repurposed industrial sites.
Risk Analysis: Expert Perspective
Primary Risks:
1. Execution Risk: Converting a mining facility to AI standards requires massive capital and expertise. Galaxy used $1.4 billion in debt financing—if AI demand softens or tenants default, the debt burden could become crushing.
2. Tenant Concentration: Coreweave is Galaxy’s anchor tenant for 526 MW. If Coreweave faces financial difficulties or shifts strategy, Galaxy has limited diversification.
3. Technology Obsolescence: AI hardware evolves rapidly. A data center optimized for H100 GPUs today might need expensive retrofits for next-generation chips in 2-3 years.
4. Power Cost Volatility: While AI tenants pay contracted rates, if electricity prices spike due to natural gas shortages or grid instability, margins could compress.
Mitigation Strategies:
- Galaxy has secured land control for over 2,200 acres with potential to scale to 3.6 GW, allowing tenant diversification.
- The pass-through cost model protects Galaxy from operating cost inflation.
- Phase II construction (260 MW) is already underway with data hall deliveries expected in early 2027.
Expert Consensus: The trend is real, but not every miner can execute. Smaller miners lack the balance sheet to finance conversions. Industry analysts at CoinMetrics suggest that only the top 10-15 publicly traded miners have the capital and expertise to pursue AI deals successfully.
Beginner’s Corner: How to Track This Trend
If you want to monitor which mining companies are pivoting to AI, here’s a simple framework:
Step 1: Check quarterly earnings reports for “AI revenue” or “HPC hosting” segments. Galaxy’s Q2 2026 report will show the first full quarter of Phase I revenue.
Step 2: Watch for partnerships. Companies announcing long-term leases with AI providers (like Coreweave, Lambda, or Vast) are serious players.
Step 3: Track power capacity. Miners with 100+ MW of available capacity near fiber networks are prime conversion candidates.
Step 4: Monitor construction progress. Phase completion announcements (like Galaxy’s July 6 milestone) indicate execution capability.
Step 5: Compare debt levels. High-quality AI conversions require significant financing—check debt-to-equity ratios.
Common mistakes to avoid:
- Don’t assume all miners can convert—many lack the electrical infrastructure for high-density AI workloads
- Don’t confuse GPU leasing (Bit Digital) with facility conversion (Galaxy)—the business models differ
- Don’t ignore regulatory risks—power grid approvals and environmental permits vary by jurisdiction
Future Outlook: What’s Next
The Helios campus is just the beginning. Galaxy has already started Phase II construction with 260 MW of additional critical IT load, expecting data hall deliveries in the first half of 2027. Phase III is targeted for 2028.
But Galaxy isn’t stopping at Helios. The company has signaled it’s weighing a second campus near Waco, Texas. Roughly 830 MW of approved capacity at Helios remains uncontracted as of early 2026, providing room for additional tenants.
The broader industry impact is clear. Sites built for hashrate are being repurposed for AI training and inference. With Coreweave committing to a 15-year lease, the economics favor miners who can execute this transition. Analysts at Bernstein predict that by 2030, up to 20% of Bitcoin mining capacity could be converted to AI workloads.
However, this shift isn’t without controversy. Some Bitcoin purists argue miners should stick to securing the network. But for publicly traded companies answerable to shareholders, the math is simple: AI tenants pay 2-3x more per megawatt than Bitcoin mining revenue, with less volatility. That’s a structural shift that’s likely to define crypto mining’s next chapter.
Key Takeaways
- Galaxy Digital’s Helios conversion from Bitcoin mining to AI infrastructure represents a structural shift in the mining industry, with 133 MW already delivered to Coreweave.
- The financial case is compelling: Phase I alone projects $300 million annual revenue with 90% EBITDA margins, compared to volatile Bitcoin mining profits.
- This trend is not universal—only well-capitalized miners with access to power, fiber, and construction expertise can successfully execute AI conversions.
- For crypto investors, monitoring which miners secure AI partnerships is becoming as important as tracking Bitcoin price and hash rate.
,
“datePublished”: “2026-07-07”,
“dateModified”: “2026-07-07”,
“mainEntity”: {
“@type”: “Thing”,
“name”: “Crypto Mining to AI Conversion”
}
}