Bitcoin Mining

Bitcoin Mining's $70 Billion AI Pivot: How Colocation Facilities Are Becoming Hybrid BTC and AI Hosting Centers

The Convergence of Bitcoin Mining and AI Infrastructure

The Bitcoin mining industry is undergoing its most significant structural transformation since the shift from GPU to ASIC mining. Companies that spent the past decade building power infrastructure, cooling systems, and data center facilities for cryptocurrency mining are now repurposing that same infrastructure to host artificial intelligence and high-performance computing workloads.

The numbers tell the story. Bitcoin mining companies have collectively secured more than $70 billion in AI and HPC hosting contracts. October 2026 marked the industry's best revenue month since January, with $1.12 billion in total mining revenue. But increasingly, the revenue growth is coming not from Bitcoin alone but from the AI hosting operations running alongside the mining rigs.

This is not a pivot away from Bitcoin. It is the emergence of a new asset class: the hybrid colocation facility that serves both BTC mining and AI compute workloads from the same power and cooling infrastructure.

Why Mining Companies Have a Structural Advantage

Building a data center from scratch to serve AI workloads takes 18 to 36 months and requires securing three things that are increasingly scarce: power capacity, land with grid access, and environmental permits. Bitcoin mining companies already have all three.

Power at Scale

The single biggest constraint on AI infrastructure buildout is power. Hyperscale AI clusters consume hundreds of megawatts. Getting that much power connected to a new site involves utility negotiations, substation construction, and transmission upgrades that can take years.

Bitcoin mining companies already operate facilities with 50 to 500+ MW of grid-connected power. That power is contracted, the substations are built, the transmission lines are in place. Converting a portion of that capacity from mining ASICs to GPU servers is an infrastructure repurposing exercise, not a greenfield buildout.

Cooling Infrastructure

Both ASIC miners and GPU servers generate substantial heat. Mining facilities have already solved the thermal management challenge at scale, often with immersion cooling, rear-door heat exchangers, or industrial evaporative systems. The same cooling infrastructure that keeps a hall of S21 miners at operating temperature can keep a rack of H100 or B200 GPUs within spec.

Operational Expertise

Running a mining facility at scale means managing power distribution, monitoring thousands of devices, handling hardware failures, maintaining uptime targets, and optimizing for cost efficiency. These are the same operational disciplines required for AI hosting, with the addition of network architecture and SLA management.

The Companies Leading the Pivot

Several publicly traded Bitcoin mining companies have made substantial commitments to hybrid operations.

Core Scientific

Core Scientific emerged from Chapter 11 restructuring with a clear dual-track strategy. The company retained significant Bitcoin mining capacity while securing multi-billion dollar contracts to host AI workloads for CoreWeave and other AI infrastructure providers. Their facilities in Texas, Georgia, and other states are being reconfigured to accommodate both ASIC miners and GPU servers, with power allocation shifting dynamically based on contract economics.

Hut 8

Hut 8 has positioned itself as a diversified digital infrastructure company, with AI compute hosting becoming an increasingly important revenue stream alongside self-mining operations. The company's acquisition strategy has focused on facilities with the power and space to serve both markets, and their managed services platform supports customers running both mining and AI workloads.

TeraWulf

TeraWulf's Lake Mariner facility in New York, powered by zero-carbon nuclear and hydroelectric energy, represents the emerging template for hybrid operations. The site's clean energy profile is attractive to AI customers with ESG commitments, and TeraWulf has dedicated significant capacity to high-performance computing alongside its mining operations.

Other Players

Iris Energy, Cipher Mining, Bitdeer, and Applied Digital have all announced or executed strategies to offer AI hosting services from their mining infrastructure. The common thread is straightforward: they own the power, they own the facilities, and AI customers need both.

The Economics of Hybrid Operations

The financial logic driving the hybrid model is compelling.

Revenue Per Megawatt

Bitcoin mining generates revenue based on hashrate, BTC price, network difficulty, and power cost. AI hosting generates revenue based on contracted rates per kilowatt, utilization, and service level. At current market conditions, AI hosting generally produces higher and more predictable revenue per megawatt than Bitcoin mining alone.

Metric	Bitcoin Mining	AI/HPC Hosting
Revenue predictability	Variable (BTC price, difficulty)	Contracted (fixed rate per kW)
Contract duration	Spot or short-term	Multi-year commitments
Capital intensity	Lower (ASICs depreciate faster)	Higher (GPU racks, networking)
Power cost sensitivity	Extremely high	High but less critical
Uptime requirements	Economic (maximize hashrate)	SLA-bound (99.99%+ typical)

Power Cost Thresholds

In the United States, competitive hybrid facilities operate at power rates between $0.06 and $0.09 per kilowatt-hour. At these rates, both Bitcoin mining with modern sub-15 J/TH ASICs and AI inference hosting generate positive margins.

Facilities with access to stranded energy, behind-the-meter power purchase agreements, or grid curtailment programs can achieve rates below $0.05/kWh, significantly improving margins for both workload types. In the Middle East and Central Asia, energy costs can be even lower, making these regions particularly attractive for hybrid deployments.

Dynamic Power Allocation

The most sophisticated hybrid operators are implementing dynamic power allocation systems. During periods of high Bitcoin mining profitability, such as after a price surge or difficulty adjustment, more power flows to ASICs. When mining margins compress or AI demand spikes, power shifts to GPU servers. This flexibility maximizes revenue per megawatt across market cycles.

Technical Challenges of Hybrid Operations

Converting a pure mining facility to hybrid operations is not as simple as swapping ASICs for GPUs. Several technical challenges must be addressed.

Network Infrastructure

Bitcoin mining requires minimal network bandwidth. Each ASIC communicates with a mining pool using kilobits per second. AI workloads, particularly distributed training, require high-bandwidth, low-latency networking with 100 Gbps or 400 Gbps connections between GPU nodes. Adding this networking infrastructure to a facility designed for mining is a significant capital investment.

Power Density

Mining facilities typically distribute power at 5 to 15 kW per rack equivalent. Modern AI GPU servers can exceed 40 to 80 kW per rack. The electrical distribution, bus bars, PDUs, and cabling in a mining facility may need substantial upgrades to support AI-density power delivery.

Cooling Precision

ASICs are relatively tolerant of temperature variation. They run hot and their performance degrades gracefully. GPU servers are more sensitive. AI workloads require tighter temperature control, lower humidity, and often require liquid cooling solutions for the highest-density configurations. Upgrading cooling infrastructure from mining-grade to AI-grade adds cost and complexity.

Physical Security and Compliance

AI hosting customers, particularly enterprise and government clients, require SOC 2 compliance, physical security controls, access management, and audit trails that many mining facilities were not designed to provide. Meeting these requirements involves both facility upgrades and operational process changes.

The ASIC Side of the Equation

The pivot to AI does not mean the end of Bitcoin mining. Quite the opposite. Mining hardware continues to improve, with the latest generation of ASICs achieving sub-15 joules per terahash efficiency. At these efficiency levels, mining remains profitable at power costs up to $0.07 to $0.08/kWh depending on BTC price.

The hashrate reflects this continued investment. The Bitcoin network is approaching 1 exahash per second, with October's average around 983 EH/s. Miners are deploying more efficient hardware while also diversifying into AI hosting, rather than choosing one over the other.

For hybrid operators, the ASIC fleet serves as a flexible baseload. Mining generates revenue 24/7 with minimal operational overhead. AI contracts provide higher margins but require more sophisticated operations. The combination provides revenue stability that neither workload offers alone.

Market Implications for Colocation Providers

The hybrid mining-AI trend has significant implications for the broader colocation and hosting market.

Competition for Power

Mining companies converting to hybrid operations are competing with traditional data center operators for the same AI hosting contracts. Their advantage is speed: they have power now, while new data center builds take years. Their disadvantage is the operational maturity gap: enterprise AI customers expect a level of service sophistication that mining operators must develop.

Valuation Rerating

Public mining companies that successfully pivot to hybrid operations are being revalued by the market. A company valued on a mining multiple, which typically reflects BTC price and hash cost, gets rerated to a data center multiple when it secures long-term AI hosting contracts. Data center multiples are significantly higher, which is why the pivot has attracted so much capital.

Geographic Expansion

The hybrid model is not limited to the United States. Mining operations in Canada, Northern Europe, the Middle East, and Central Asia are all evaluating or executing similar pivots. Regions with abundant low-cost energy and favorable regulatory environments for both mining and AI are particularly well positioned.

What This Means for Rax Customers

For organizations looking to deploy either Bitcoin mining or AI compute workloads, or both, the convergence creates new options.

• Flexible contracts: Hybrid facilities can offer hosting agreements that allow customers to shift between mining and AI workloads as their needs evolve
• Power optimization: Shared power infrastructure means better utilization rates and potentially lower per-kW costs
• Risk diversification: Operators with revenue from both mining and AI hosting are more financially stable partners
• Global options: The hybrid model works across geographies, giving customers access to the best power rates and regulatory environments for their specific workload mix

Frequently Asked Questions

Why are Bitcoin mining companies pivoting to AI hosting?

Bitcoin mining companies are pivoting because they already own the critical infrastructure: large-scale power capacity, industrial cooling, and data center facilities. AI contracts offer higher and more predictable margins per megawatt. Companies like Core Scientific, Hut 8, and TeraWulf have collectively secured over $70 billion in AI and HPC hosting contracts.

What is a hybrid BTC and AI hosting facility?

A hybrid facility runs both Bitcoin mining ASICs and AI/HPC GPU servers within the same campus or power portfolio. Power allocates dynamically between workloads based on market conditions, providing revenue diversification and better infrastructure utilization.

How much revenue are Bitcoin mining companies generating from AI hosting?

Mining companies have secured over $70 billion in cumulative AI and HPC contracts. October 2026 marked the best month since January at $1.12 billion in total mining revenue, with AI hosting growing as a percentage for companies that have made the pivot.

What power rates do hybrid facilities need to be profitable?

In the US, competitive hybrid facilities operate at $0.06 to $0.09 per kWh. Both mining with sub-15 J/TH ASICs and AI inference hosting generate positive margins at these rates. Facilities with stranded energy or PPAs below $0.05/kWh see significantly improved economics for both workload types.