A bitcoin mining facility today can sell for 10x the same number depending on how it's positioned, packaged, and brought to market. The difference between $400K per MW (mining buyer) and $4M per MW (AI conversion buyer) is rarely about the physical asset. It's about what the buyer believes they're getting. This is the playbook we use with operators in our network preparing to sell.
Start with the buyer mix you're targeting
Before any document gets written or any data room gets populated, decide who the realistic buyer set is. A site with poor fiber, no water, and a queue-limited interconnection is going to clear with mining buyers regardless of the marketing pitch — and trying to sell it as AI-ready wastes 6-9 months and damages credibility. Conversely, a site with strong fundamentals undersold to mining buyers leaves real money on the table.
The honest screening criteria for the AI-buyer bid: at least 25-50MW of contiguous capacity, executed or near-term path to fiber, water access or a credible closed-loop path, interconnection capacity that doesn't require a new queue submission, and a jurisdiction with predictable permitting.
The data room — minimum viable version
Buyers will ask for everything; you need a defensible core. We see deals stall when basic operational data isn't available or contradicts itself. Build the data room around these items first:
- 01Site survey, parcel boundaries, and easement documentation. Title commitment current within 6 months.
- 02Substation single-line diagram, transformer nameplate data, and a written description of redundancy.
- 03Power contract (PPA or tariff schedule) with all amendments, and 24 months of metered consumption and demand data.
- 04Interconnection agreement (LGIA/SGIA) and any amendments showing approved capacity.
- 05Environmental: Phase I (and Phase II if applicable) current within 18 months. Stormwater permits, air permits, any noise variance.
- 06Operations: uptime data by month for the past 12 months, incident reports, planned maintenance log, current staffing roster.
- 07Financials: site-level P&L, capex history with vendor invoices, current insurance certificates.
- 08Regulatory: zoning verification, building permits, fire marshal sign-offs, any open code items.
What AI buyers ask for that mining buyers don't
If you're targeting AI buyers, you also need to address materially more on cooling, fiber, and water. This is where deals get re-priced down if the materials aren't ready.
- Cooling design study or engineering memo showing path from current air cooling to liquid cooling (direct-to-chip or immersion), including floor loading calculations.
- Fiber audit: which carriers terminate or pass near the property, with letter-confirmed availability. Latency tests to the nearest IXP.
- Water rights documentation if applicable, or a closed-loop cooling design and the chiller economics.
- Floor plans showing where rack density can scale to 30-100 kW and where it can't.
- Any prior conversations with hyperscaler or neocloud counterparties about the site (with permission to share).
How to position the asset
Positioning matters as much as the data. The best results we see come from a two-track CIM and process: one positioning sheet for mining buyers (focused on operational excellence, power economics, and immediate cash flow) and one for AI buyers (focused on capacity scalability, conversion path, and 10-year value).
Run the process in parallel, not in series. Sequential processes — 'we'll try AI buyers first, then fall back to mining' — tend to extend timelines and damage information advantage. The parallel approach lets you discover the true bid stack within 60-90 days.
Pricing — what to expect
From our recent transaction data, here's the rough bracket by buyer type for an operational, fully-permitted facility with average power costs:
- Mining buyer, operational acquisition: $300K-$500K per MW. Upper end requires sub-4¢ power, recent capex, and a transferring operations team.
- Mining buyer, distressed or off-the-grid scenario: $150K-$300K per MW. Site quality matters but the discount is structural.
- AI conversion buyer, all fundamentals strong: $3M-$8M per MW. Range driven by interconnection clarity, fiber, and operator credit.
- Long-dated lease structure with IG counterparty: $1.0M-$1.3M per MW per year, on terms similar to the recent comparable transactions we track.
Process timeline
Realistic timelines vary by structure. Mining-to-mining transactions can clear in 60-90 days from CIM release to signed PSA when the data room is clean. Mining-to-AI conversions are typically 120-180 days because the diligence is heavier and the buyer's internal approval process takes longer. Add another 90-180 days from signed PSA to close, depending on regulatory approvals.
The biggest delays we see: power contract assignments that require utility approval, environmental Phase II remediation that surfaces during diligence, and interconnection capacity questions that send the parties back to the ISO for clarification.
Common mistakes
Three patterns we see repeatedly that cost sellers real money.
- 01Underselling the AI thesis. Operators who haven't internalized that their site might be worth multiples more to a different buyer category sometimes accept the first mining bid and leave the AI premium on the table.
- 02Overselling the AI thesis. Sites that genuinely can't be converted shouldn't be marketed as if they can. The buyer set figures it out within weeks and the seller loses credibility.
- 03Ignoring operational quality. Even AI buyers care about operational track record. A site running at 85% uptime over the past 12 months prices below a site running at 99%, even if the AI buyer plans to install their own ops team.