Report Board
Bitcoin Miners' Power-to-AI Pivot: BTDR vs HUT vs IREN vs CIFR
A power-first comparison of BTDR, HUT, IREN and CIFR: which miners merely own megawatts, and which have already converted power into AI leases, AI Cloud ARR and financeable cash flow.
A Reviewable Logic Chain
Each card stays open and maps one transmission node without collapsible controls or pseudo-precise scores.
AI wants power before buildings AI 先缺电，再缺楼
IEA expects data-center electricity consumption to more than double by 2030, while grid queues and transformers can delay new capacity.
IEA 预计数据中心用电到 2030 年翻倍以上；并网、变压器和电网排队决定谁能更快交付 AI 负载。
MW is not enough 有 MW 不等于有 AI 价值
The valuable asset is permitted, energized, fiber-connected utility capacity that can become contracted IT load at the right PUE and financing cost.
真正值钱的是可并网、可建设、可接光纤、可按合理 PUE 转成 IT load 并被租户和债权人认可的容量。
Contracts beat narrative 合同强于叙事
HUT and CIFR have already turned capacity into long-duration AI leases; IREN has AI Cloud ARR contracts; BTDR still needs Tydal and U.S. site proof.
HUT/CIFR 已经有长约，IREN 已经有 AI Cloud ARR 合同，BTDR 还需要 Tydal 和美国站点把期权变成合同。
Power conversion is capital intensive 电力升级极度烧钱
AI campuses require debt, converts, project finance, GPU finance and tenant backstops; upside belongs to companies that finance without excessive dilution.
AI 园区需要债、可转债、项目融资、GPU financing 和租户 backstop；能低稀释融资的一方会获得更高估值。
Rank by conversion, not capacity 按转化率排序，不按容量排序
The best setup combines speed-to-power, creditworthy tenants, construction execution, low-cost capital and a clear AI operating layer.
最好的标的不是 MW 最大，而是速度、租户、建设、资金成本和 AI 运营层同时最强。
How Four Miners Are Converting Power Into AI Infrastructure
This is a static research snapshot as of 2026-05-22, not investment advice. It prioritizes company IR, SEC filings, company presentations and the IEA report. Market prices and market caps are used only as context, not as real-time trading inputs.
| Type | Reviewable asset / contract | Ticker | Current read | Potential slot | Next check |
|---|---|---|---|---|---|
| Raw power option | 3.0035GW global electrical capacity; $69M AI Cloud ARR; Tydal 225MW targeted Q4 2026 | BTDR | AI contract proof is still early | Largest relative re-rating option | Tydal is the key catalyst; Clarington 570MW has legal/timing uncertainty and should not yet be treated as near-term AI cash flow. |
| Contracted landlord | 597MW IT capacity; $16.8B base-term leases; 9,085MW development platform | HUT | Strongest contract proof | Highest conversion certainty | River Bend and Beacon Point prove the power-first financing model; now watch Q2/Q3 2027 delivery milestones. |
| AI Cloud operator | >4.5GW secured power; $3.1B ARR under contract; Microsoft and NVIDIA contracts | IREN | Fastest operating conversion | One of the highest ceilings | The main risks are GPU capex, converts/financing, customer ramp, data-center delivery and AI Cloud gross margin. |
| HPC developer leverage | 907MW operating/contracted; 700MW contracted gross HPC; $11.4B contracted revenue | CIFR | Strong contracts, heavy leverage | High beta | Barber Lake / Black Pearl milestones and debt service determine equity upside; ERCOT pipeline remains uncertain. |
| Sector filter | IEA: data-center electricity demand more than doubles by 2030 | Theme | The power bottleneck is real | Dispersion ahead | Power is the entry ticket; tenants, financing and delivery decide valuation dispersion. |
BTDR · HUT · IREN · CIFR · Power-first AI infrastructure snapshot · 2026-05-22
Earnings releases, announcements, filings, estimate tables, and reviewable sources.
- Core signal
- BTDR has 3.0035GW of global electrical capacity and about $69M of AI Cloud ARR; HUT has $16.8B of base-term contracted leases across 597MW of IT capacity; IREN has more than 4.5GW of secured power, $3.1B of ARR under contract and targets $3.7B by year-end 2026; CIFR has 907MW of operating and contracted capacity, 700MW of contracted gross HPC capacity and about $11.4B of contracted revenue.
- Current read
- This is a conversion-stage ranking, not a simple buy/sell call. HUT is the strongest power-first developer by contract proof; IREN is the fastest AI Cloud operator but carries heavy financing and execution pressure; CIFR is a contracted HPC development platform with meaningful leverage; BTDR is the strongest raw-power and mining-hardware option but still needs AI tenant proof.
- Next question
- When AI data centers are bottlenecked by available power, should these miners' megawatts still be valued as mining assets or as AI data-center options?
The miners' AI pivot should be judged by power-to-AI conversion, not hash rate.
HUT has the strongest contract proof: 597MW of IT capacity, $16.8B of base leases and a clearer project-finance path.
IREN has one of the highest AI Cloud ceilings: more than 4.5GW of secured power and Microsoft/NVIDIA validation, but also heavy financing and execution risk.
CIFR is a high-leverage HPC developer: $11.4B of contracted revenue, but debt, construction and ERCOT pipeline amplify risk.
BTDR is the relative option: 3.0GW of power plus Tydal plus AI Cloud ARR, but it needs credit tenant contracts to earn a peer-level AI premium.
Does BTDR sign a Tydal Norway colocation tenant? What are the term, MW, tenant credit, rent/NOI and financing structure?
Which BTDR sites among Rockdale, Knoxville, Wenatchee, Niles and Clarington can realistically convert into AI capacity during 2026-2028?
Does HUT deliver River Bend around Q2 2027 and Beacon Point initial delivery around Q3 2027? Do costs stay near the stated $9-11M/MW range?
Does IREN deliver its 480MW 2026 AI Cloud capacity and 1,210MW 2027 build? Do Microsoft/NVIDIA ARR figures convert into recognized revenue?
Do CIFR's Barber Lake, Black Pearl and Stingray hit milestones? Does the $5B+ debt stack constrain common-equity upside?
If AI data-center power demand slows, or GPU/tenant capex weakens, which company falls back fastest from AI premium to miner valuation?