This report tracks listed companies that can turn power, data center capacity, or GPUs into AI hosting and AI cloud revenue.
The key distinction is contract proof: APLD, WULF, CIFR and HUT have large long-term AI/HPC leases; IREN has the fastest AI Cloud operating ramp; BTDR and HIVE have early GPU revenue and power options; RIOT, BITF, CLSK and MARA still need larger customer contracts.
Who has really turned power into AI contracts?
The miner-to-AI trade is not one story. It has four stages: controlled power, deliverable data-center shell, signed customer contract, and recurring revenue. The strongest signal is not gross GW; it is how many MW have been taken by creditworthy customers.
Bottom line: proof-of-contract ranking
A names already have large, long-duration AI/HPC leases with named customers or strong credit support. B names have operating AI Cloud or medium-size contracts. C names own large power positions but still need a hyperscaler-style customer catalyst. D names remain earlier-stage power options.
APLD / WULF / CIFR / HUT / CORZ
Hundreds of MW have been converted into long-term AI/HPC leases, making contracted revenue or NOI valuation possible.
The debate shifts from 'can they find customers?' to delivery and financing execution.
IREN / CRWV / NBIS
More AI Cloud / neocloud exposure. They have customers and GPU revenue, but valuation depends on capex, GPU turns, and customer concentration.
IREN is one of the fastest miner-to-AI Cloud operators.
BTDR / RIOT / BITF / HIVE
They have power, sites, and early AI activity. A large customer contract still determines the repricing slope.
BTDR is a high-beta option on the next major contract.
CLSK / MARA
The power portfolio and AI intent are visible, but the current setup is closer to an unmonetized power option.
Watch for customer names, delivery dates, and financing structure.
At a glance: power base vs signed AI/HPC vs expansion
Blue is controlled power or gross capacity, green is signed AI/HPC critical IT load, and gold is disclosed construction or expansion. Each column is normalized to its own maximum: power base max=IREN 5GW, signed max=APLD 1.4GW, expansion max≈CIFR 3.3GW pipeline. AI Cloud ARR, GPU count, and advanced negotiations are not treated as signed MW.
Data audit
The chart separates signed lease MW, disclosed power/capacity MW, and non-MW AI Cloud metrics such as ARR or GPU fleet size. BTDR, HIVE, BITF, and CLSK get zero in signed-MW bars unless a hyperscaler-style long lease is disclosed. HUT is treated as 597MW of signed AI data-center capacity, not only the River Bend project. APLD's expansion bar uses planned critical IT load rather than gross power.
Real map: where the sites are and which regions look friendlier
This section now uses a real OpenStreetMap basemap. Markers are approximate to publicly disclosed campus, city, county, or state-level locations, not street-address coordinates. Green means signed AI/HPC lease, blue means self-operated AI Cloud, gold means pending/in negotiation, and red means earlier-stage or no large disclosed tenant. Larger circles represent larger disclosed MW buckets.
Method: locations use publicly disclosed site/city/state-level references; MW is an investment-analysis reference marker and does not treat gross power as revenue capacity.
Low power costs, ERCOT/PPA flexibility, and mature data-center sales-tax exemptions; regulatory scrutiny is rising.
Power and land are friendly, with a clear data-center tax framework.
Act 730 supports data-center equipment and software exemptions.
WULF is making Kentucky a second growth pole; interconnect execution matters.
Hydro/renewable power helps AI Cloud and ESG positioning.
Lake Mariner has nuclear/hydro positioning, but energy-intensive workloads face more policy friction.
How to trade the map
Most contract-proven
- APLD: largest signed critical IT load base.
- WULF: Anthropic/Fluidstack proof plus Kentucky option.
- CIFR/HUT: strong lease validation, but execution timing matters.
Highest option value
- BTDR: very large power base, still waiting for a headline tenant.
- RIOT/BITF/HIVE/CLSK: power optionality, less contract proof.
Main risk
- Confusing gross GW with deliverable critical IT load.
- Financing and grid interconnect timing.
- Customer concentration and lease economics.
Next evidence to watch
- Named customers and lease MW.
- Capex funding without excessive dilution.
- Construction milestones and energization dates.