Why a Facility Tour Is Not Enough

Every colocation provider will give you a polished tour. They will show you neat rows of machines, blinking LEDs, and a control room with dashboards. What they will not volunteer is the age of their transformers, the actual measured PUE under summer load, whether their backup generator has been load-tested in the past six months, or what happens to your machines if they lose their lease.

Signing a 12- to 36-month mining hosting contract without rigorous due diligence is committing hundreds of thousands of dollars based on a sales presentation. The inspection process outlined here is designed for miners evaluating colocation facilities before signing — whether you are placing 50 units or 5,000. Every item on this checklist exists because a real operator lost real money by not checking it.

Electrical Infrastructure Inspection

Electrical infrastructure is the single largest determinant of facility reliability. A facility can have beautiful aesthetics and terrible power delivery. Start here.

Utility Feed and Transformer Capacity

• Request the utility interconnection agreement: This document shows the facility’s contracted power capacity from the utility. Compare this number to the facility’s advertised capacity. If they advertise 20 MW but the interconnection agreement shows 15 MW, they are either oversubscribed or planning an upgrade that may not happen.
• Inspect transformer nameplates: Walk to every transformer on site. Read the nameplate for kVA rating, manufacture date, and impedance. Transformers older than 20 years have higher failure rates and longer replacement lead times (currently 16-52 weeks for large power transformers). A 2006-vintage transformer serving your section is a risk you need to price into your decision.
• Check transformer loading: Ask for current load readings on each transformer. A transformer running above 80% of nameplate capacity has no headroom for your additional load and will run hotter, accelerating insulation degradation. Ideal loading is 60-75%.
• Verify voltage quality: Request 30 days of power quality monitoring data. Look for voltage sag events (drops below 90% of nominal), harmonic distortion above 5% THD, and frequency of utility outage events. ASIC power supplies are sensitive to voltage sags — repeated events cause PSU failures and hash board damage.

Power Distribution and Metering

• Sub-metering accuracy: How does the facility measure your power consumption? Individual CT (current transformer) metering per rack or per row is the gold standard. Facility-wide metering divided by customer count is unacceptable — you will pay for other customers’ inefficiencies. Ask to see the meter hardware and verify calibration dates.
• PDU configuration: Check the power distribution units serving your allocated space. Verify amperage ratings, breaker sizing, and the number of available circuits. Running a 3,250W Antminer S21 on a circuit with insufficient breaker capacity causes nuisance trips and downtime.
• Redundancy topology: Is the power path single-feed or dual-feed? A single utility feed with no redundancy means every utility outage takes your machines offline. Dual utility feeds, automatic transfer switches (ATS), or on-site generation provide progressively better protection. Know exactly which topology covers your section, not just the facility’s best section.

Cooling and Environmental Systems

Cooling failures kill ASICs. A facility that cannot reject heat fast enough will either throttle your machines (reducing revenue) or damage them (destroying capital).

Cooling Capacity Verification

• Design cooling capacity vs actual load: Request the facility’s total cooling capacity in kilowatts or BTU/hour and compare it to the current IT load. A facility with 10 MW of cooling capacity and 9.5 MW of IT load has almost no margin for your additional machines. Target facilities with at least 20% cooling headroom above current load.
• Measure inlet temperatures yourself: Bring a digital thermometer. Measure air temperature at the intake side of machines in multiple locations across the facility. ASHRAE recommends 64-80F (18-27C) inlet temperature for compute equipment. If any location reads above 85F during your visit, that location will be worse during summer peak. Antminer S21 units begin throttling at ambient temperatures above 40C (104F).
• Hot aisle containment: Is the facility using hot aisle / cold aisle separation with physical containment (curtains, panels, or ducting)? Without containment, hot exhaust air recirculates into intake, raising effective inlet temperature by 5-15F. This is not a cosmetic issue — it directly reduces your hashrate.

Environmental Controls

• Humidity monitoring: Ask for humidity data from the past 90 days. ASHRAE guidelines recommend 20-80% relative humidity for IT equipment. Below 20% increases electrostatic discharge risk. Above 80% risks condensation on circuit boards. Mining facilities in humid climates (Gulf Coast, Southeast) need active dehumidification.
• Dust and air filtration: What MERV rating are the intake filters? MERV 8 is minimum for mining facilities. MERV 11 or higher is preferred in dusty environments (rural sites, desert locations, agricultural areas). Ask when filters were last replaced. Clogged filters restrict airflow, raising temperatures and increasing fan power consumption.
• Water exposure risk: Is your allocated space below any water pipes, HVAC condensate lines, or roof penetrations? Water damage to ASICs is catastrophic and rarely covered by standard property insurance. Inspect the ceiling above your future rack locations.

Backup Power and Resilience

Backup power is where marketing claims and operational reality diverge most sharply.

Generator Inspection

• Generator capacity vs protected load: What percentage of facility load is generator-backed? Many mining facilities only provide generator backup for networking equipment and control systems, not for the mining floor. If your ASICs are not on generator-backed circuits, you have zero protection against utility outages. Get this in writing.
• Fuel supply and runtime: For diesel generators, verify on-site fuel storage capacity and calculate runtime at full load. A 500-gallon tank on a 2 MW generator runs approximately 3.5 hours. Ask about fuel delivery contracts and priority during regional emergencies (hurricanes, ice storms) when fuel supply chains break down. For natural gas generators, verify the gas supply contract and pressure requirements.
• Load test records: Request generator load test records for the past 12 months. Generators should be load-tested monthly (minimum) at 75%+ of rated capacity. A generator that has not been load-tested will surprise you when it matters most. NFPA 110 requires monthly testing for Level 1 systems.
• Transfer time: How long does it take for the generator to assume load after a utility outage? ATS (automatic transfer switch) systems typically transfer in 10-30 seconds. Manual transfer can take 5-15 minutes. Your ASICs will reboot after any interruption, and pool shares are lost during the restart window (typically 3-8 minutes for modern ASICs).

UPS and Ride-Through

• UPS coverage: Does the facility provide UPS (uninterruptible power supply) protection for the mining floor? Most mining colos do not, because the cost of UPS at mining power densities is prohibitive. This is normal — but know what you are getting. UPS for network switches and control systems is standard and expected.
• Clean shutdown capability: In a power loss scenario, can the facility execute a clean shutdown of your machines, or do they hard-crash? Repeated hard shutdowns accelerate NAND wear on control boards and can corrupt firmware configurations.

Network and Connectivity

Network issues in mining are less visible than power issues but equally costly. A 1% packet loss rate to your mining pool translates directly to approximately 1% revenue loss from stale shares.

• Internet redundancy: How many upstream ISPs serve the facility? A single ISP connection is a single point of failure. Dual ISP with automatic failover via BGP is the standard for professional facilities. Ask for the ASN (Autonomous System Number) if they claim to run BGP — it is trivially verifiable.
• Latency to major pools: Request a traceroute or ping test from the facility network to your preferred mining pool. Latency under 50ms is acceptable for most pool protocols. Latency above 100ms increases stale share rates, especially on pools using Stratum V2.
• Network monitoring: Does the facility monitor network uptime and provide incident reports? Request the past 90 days of network uptime data. Target 99.9%+ network availability.
• IP allocation and firewall policy: Will your machines get their own VLAN or subnet? Shared flat networks with other customers create security exposure — a compromised machine on the same broadcast domain can ARP-spoof your mining traffic. VLAN isolation per customer is the minimum acceptable standard.

Physical Security and Access

• Access control system: How do you physically access your machines for maintenance? Key card, biometric, or escorted access? How are access logs maintained and how far back can you request them? If a machine goes missing, the access log is your only evidence.
• Camera coverage: Are there security cameras covering the racks where your machines are located? What is the recording retention period? 30 days minimum is standard. Less than 14 days is a red flag.
• Visitor and contractor policy: Can the facility staff or other customers physically touch your machines without your authorization? Written policies should prohibit unauthorized access to customer equipment. Some facilities allow technicians to reboot or relocate machines without customer approval — this should be explicitly addressed in your contract.
• Perimeter security: Fencing, lighting, and staffing at the facility perimeter. ASICs are high-value, portable targets. A facility in a remote area with no perimeter fencing and no on-site staff at night is an insurance claim waiting to happen.

Financial and Legal Due Diligence

The physical facility can be perfect and still cost you everything if the operating entity is financially unstable or legally exposed.

• Lease vs ownership: Does the facility operator own the building and land, or lease it? If leased, when does the lease expire? A facility operator whose lease expires in 18 months may not invest in maintenance and may leave you scrambling for a new home for your hardware.
• Insurance coverage: Request a certificate of insurance. Verify that the facility carries commercial property insurance, general liability, and — critically — coverage for customer equipment. Many facility policies explicitly exclude customer property. If your machines are not covered by their policy, you need your own inland marine or equipment floater policy.
• Financial stability: For large deployments (100+ units), request basic financial references or evidence of capitalization. A facility operator who cannot pay their own utility bill will not be keeping your machines powered. This is uncomfortable to ask but essential for six- and seven-figure deployments.
• Customer concentration: Is the facility dominated by a single large customer? If one customer represents 70% of the facility’s revenue and that customer leaves, the facility may become financially nonviable. Diversified customer bases are more stable.
• Contract termination terms: What happens to your machines if the facility closes, defaults on its utility bill, or enters bankruptcy? Your hosting agreement should specify your right to retrieve equipment within a defined window. Without this clause, your ASICs could be tied up in a landlord lien or bankruptcy estate for months. Review hosting agreement structures carefully before committing.

The Pre-Signing Inspection Checklist

Use this condensed checklist during your site visit. Every item should have a documented answer before you sign:

Electrical (8 items)

• Utility interconnection agreement reviewed and capacity confirmed
• Transformer age, rating, and current loading documented
• 30-day power quality data reviewed (voltage sags, THD, outages)
• Sub-metering method verified (per-rack CT metering preferred)
• PDU amperage and circuit capacity confirmed for your allocation
• Power redundancy topology documented (single-feed, dual-feed, ATS)
• Generator capacity, fuel supply, and load test records reviewed
• Transfer time (ATS) confirmed

Cooling (6 items)

• Total cooling capacity vs current IT load documented (target 20%+ headroom)
• Inlet temperatures measured at your allocated location
• Hot aisle / cold aisle containment verified
• Humidity monitoring data reviewed (90-day minimum)
• Air filtration MERV rating confirmed and last replacement date recorded
• Water exposure risk assessed for your allocated racks

Network (4 items)

• Dual ISP with automatic failover confirmed
• Latency to target mining pool tested (<50ms acceptable)
• 90-day network uptime data reviewed (target 99.9%+)
• VLAN isolation per customer confirmed

Security (4 items)

• Access control system and logging confirmed
• Camera coverage of your rack location verified (30-day retention minimum)
• Unauthorized access policy in writing
• Perimeter security assessed (fencing, lighting, staffing)

Financial/Legal (5 items)

• Facility lease term or ownership confirmed
• Certificate of insurance reviewed (customer equipment coverage checked)
• Financial stability assessed for large deployments
• Customer concentration risk evaluated
• Contract termination and equipment retrieval terms in writing

Frequently Asked Questions

What is the single most important thing to check during a colocation facility inspection?

Transformer loading and age. Electrical infrastructure failures cause the longest and most expensive downtime events in mining. A 25-year-old transformer running at 85% capacity is a ticking clock. Replacement lead times for large power transformers currently range from 16 to 52 weeks, meaning a failure could idle your operation for months. Everything else — cooling, network, security — can be patched or worked around in days or weeks. A dead transformer cannot.

Should I hire a third-party engineer for the facility inspection?

For deployments above 1 MW or contract values exceeding $500,000, yes. A licensed electrical engineer and a mechanical engineer (for cooling systems) can identify issues that are invisible to non-specialists. The cost of a professional site assessment ($5,000 to $15,000) is trivial compared to the cost of a bad facility choice. For smaller deployments, this checklist plus your own careful observation is sufficient, but do not skip the inspection entirely regardless of scale.

How do I verify a facility’s uptime claims?

Ask for documentation, not promises. Request third-party uptime monitoring reports (such as from Uptime Institute or similar), maintenance logs showing planned vs unplanned outages, and references from current customers with deployments similar to yours in size. A facility claiming 99.9% uptime should be able to produce 12 months of timestamped monitoring data. If they cannot, the number is marketing, not engineering.

Make an Informed Hosting Decision

Due diligence is the work that happens before the contract is signed and after the sales tour ends. The operators who perform thorough inspections negotiate better terms, avoid catastrophic facility failures, and build mining businesses that compound over multiple market cycles instead of collapsing after the first unexpected outage.

Rax Mining operates purpose-built mining facilities with transparent infrastructure specifications and open-door inspection policies. We welcome prospective customers to conduct the full inspection outlined above at any of our facility locations. Explore hosted colocation starting at $0.075/kWh, review available ASIC hardware, or schedule a facility tour to see our infrastructure firsthand. For off-grid deployments, our natural gas MDU solutions provide sub-$0.04/kWh power with the same operational standards. Consulting services are available for operators evaluating multiple facility options.