Bitcoin miners spent years securing one scarce resource:
electricity.
Now that power may be worth more to AI companies than to bitcoin mining itself.
Several bitcoin mining stocks are shifting toward AI data centers by leasing sites with large grid connections. Reuters Breakingviews estimates crypto miners control roughly 14 gigawatts of operational and planned U.S. power capacity, while expected data-center demand could exceed available supply by tens of gigawatts through 2028.
That changes the investment thesis.
The valuable asset may not be the mining machines.
It may be the power connection underneath them.
Why Bitcoin Miners Have Something AI Needs
Bitcoin mining requires:
- cheap electricity
- large sites
- grid connections
- cooling
- industrial infrastructure
AI data centers need many of the same things.
The difference is that securing new electricity capacity can take years.
So a miner that already controls hundreds of megawatts may own something an AI company cannot easily reproduce.
The chain is simple:
Power rights → data-center capacity → long-term lease income
That can turn a volatile crypto business into something closer to infrastructure real estate.
What Is a Powered-Shell Model?
Some miners are becoming powered-shell providers.
They provide:
- land
- buildings
- grid connection
- electricity capacity
The AI tenant supplies much of the expensive computing equipment.
Companies including Cipher Digital, Hut 8 and TeraWulf have pursued this approach, signing leases that can extend for as long as 15 years.
Reuters cites Jefferies estimates suggesting powered-shell contracts can generate up to roughly $2 million of annual revenue per megawatt.
That creates relatively predictable contracted revenue.
Why Not Own the AI Chips Too?
Another model is to provide the entire computing service.
That can generate much more revenue.
Reuters reports that neocloud-style contracts may generate closer to $10 million per megawatt annually.
But the company must then spend heavily on:
- GPUs
- servers
- networking
- cooling
- data-center equipment
IREN, for example, has a $9.7 billion Microsoft agreement but expects to spend billions on AI chips and other infrastructure to support it.
Higher revenue therefore comes with much higher capital risk.
The Hidden Risk: Chip Depreciation
AI hardware becomes obsolete quickly.
A building and electricity connection may remain useful for decades.
A GPU may lose much of its value within a few years as faster chips arrive.
That creates an important difference:
Power infrastructure = long-lived asset
AI chips = rapidly depreciating asset
Reuters Breakingviews estimated that the return on one IREN project could fall sharply if the residual value of its chips and data-center infrastructure ends up lower than expected.
That is why investors may sometimes value the landlord model more highly than the full-compute model.
Why Long-Term Leases Matter
A 10- or 15-year data-center lease can make cash flows more predictable.
Instead of depending on:
Bitcoin price + mining difficulty + electricity cost
the company may increasingly depend on:
contracted rent + power availability
That can reduce exposure to cryptocurrency volatility.
But it introduces a different risk: tenant concentration.
If one large AI company accounts for most of the lease revenue, problems at that tenant can become problems for the landlord.
Expected Return vs Risk
The two strategies have very different economics.
| Model | Potential Return | Main Risk |
|---|---|---|
| Bitcoin mining | High volatility | Bitcoin price |
| Powered-shell leasing | More stable | Tenant/default risk |
| Full AI compute | Higher potential revenue | Capex + chip depreciation |
| Power-site ownership | Long-lived asset | Grid/regulatory risk |
For investors, the key question is:
Who is taking the technology risk?
The landlord earns less revenue but avoids owning rapidly depreciating GPUs.
The full-service provider earns more but must justify billions of capital spending.
Why Electricity Could Be the Real Scarcity
AI companies can order more chips.
Building new power infrastructure is harder.
Reuters cites Morgan Stanley estimates of roughly 68 GW of U.S. data-center power demand between 2026 and 2028, around 38 GW above projected supply.
That shortage helps explain why existing powered sites are valuable.
The competitive advantage may therefore be:
not who owns the best computer
but
who can actually plug it in.
The Bottom Line
Bitcoin miners may be evolving into AI infrastructure landlords.
Their most valuable assets can include:
land + grid connection + power capacity + existing facilities
rather than mining rigs themselves.
For bitcoin mining stocks, that creates a new investment framework.
The question is no longer only:
“How much bitcoin can this company mine?”
It is increasingly:
“What is its power capacity worth to AI customers, and how much capital must it spend to monetize it?”
For more trend analysis, risk research and model-driven market tools, sign up to TradingSimuLab and explore the Trend Detector and Risk Simulation alongside the wider five-model research framework.
SEO Title: Bitcoin Mining Stocks: Why AI May Make Their Power More Valuable
Slug: bitcoin-mining-stocks-ai-data-centers-power
Meta Description: Bitcoin mining stocks are shifting toward AI data centers. Learn why power capacity, long-term leases and chip depreciation may matter more than mining.
Primary Keyphrase: bitcoin mining stocks
Secondary Keyphrases: AI data centers, bitcoin miners AI, data center power, power capacity, AI infrastructure, data center leases, bitcoin mining companies, AI electricity demand