ASIC Hosting Reliability: Uptime, Repairs and Contracts

ASIC hosting reliability cannot be proved by a percentage on a sales page. A useful assessment defines what counts as available hashrate, how the measurement is made, which outages are excluded, how quickly faults are identified and repaired, and what remedy applies when the service misses its commitment. The contract, pool records, facility evidence and removal process must describe the same service.

Define what hosting reliability actually means

Reassess ASIC hosting reliability whenever network conditions, firmware, tariffs or official guidance changes.

A facility can have electrical supply while a particular miner is offline, hashing below specification or submitting rejected work. Conversely, a miner can be healthy while a pool endpoint is unavailable. The service description should state whether availability is measured at the incoming supply, rack, miner, local controller or pool.

For a customer, accepted hashrate is usually the most useful commercial outcome. It still needs a measurement window and source. A ten minute pool reading is too volatile for a monthly service assessment, while a facility average can conceal one failed customer machine. Agree the worker identifiers and reporting source in advance.

Reliability also includes detection and recovery. Ask how the provider learns that a fan, power supply, hashboard, switch, pump or pool route has failed, who receives the alert and what happens outside normal office hours.

Calculate uptime from a written measurement boundary

When reviewing ASIC hosting reliability, separate measured facts from forecasts so the result can be reproduced.

Choose a monthly measurement period and define scheduled operating minutes. Deduct only the exclusions stated in the contract. Measured uptime is the eligible minutes in which the worker met the agreed condition divided by eligible scheduled minutes, multiplied by 100.

Suppose a month contains 43,200 minutes. If 240 minutes of agreed planned maintenance are excluded, the eligible period is 42,960 minutes. A further 430 minutes below the agreed operating threshold gives 98.999 per cent uptime, before rounding. The contract should say how partial performance and rounding are treated.

Read every exclusion before relying on an SLA

No conclusion about ASIC hosting reliability should rely on a single revenue snapshot or an undated specification.

Planned maintenance can be reasonable, but the notice period, permitted hours and annual limit should be clear. Emergency electrical or cooling work may need a different rule. A provider should not be able to label an ordinary fault as maintenance after the event merely to protect the reported figure.

Grid instruction, network interruption, pool failure, manufacturer fault, customer wallet error and force majeure are different causes. Decide who controls each cause and who carries its cost. If a facility participates in demand response or curtailment, the contract should explain notice, billing and any hashrate make-good arrangement.

A service credit is not the same as compensation for theoretical mining revenue. Credits are often capped and may be the agreed remedy. Check whether they arise automatically, require a claim within a short period or disappear when an exclusion applies.

Test monitoring and evidence before delivery

The practical value of ASIC hosting reliability comes from testing the claim against current data and full operating costs.

Request a demonstration of the customer dashboard using sample data. It should identify the customer’s workers and show historical hashrate, state, power where metered, alerts and maintenance events without exposing another customer’s information. Braiins Manager, for example, documents provider and hosting-customer roles with real-time and historical fleet metrics, but the provider’s actual configuration still matters.

Keep independent pool access. Compare the facility dashboard with accepted worker hashrate and payout reports. A monitoring view should never require the provider to control the customer’s wallet or reveal a payout credential.

Agree data retention and export. If a dispute arises after several months, a current green status icon is not enough. Monthly exports, event tickets and serial-number records make service and tax reconciliation possible.

Examine repair capacity and decision authority

The repair process starts with triage. Record who may reboot, change pools, install firmware, alter a power profile, remove a machine from service or open it. An automatic reboot can restore ordinary faults, but repeated restarts can hide overheating, PSU instability or a failed board.

Ask which parts and skills are available on site, which work is sent elsewhere and how a quotation is approved. Measure response time, diagnosis time and return-to-service time separately. A fast acknowledgement does not guarantee a fast repair.

Manufacturer routes have their own ticket, shipping, warranty and location rules. Bitmain’s published material distinguishes warranty and paid work, and its repair guidance identifies faults that require board or control-board repair. Hosting terms should say who manages that process, who pays carriage and whether substitute capacity is available.

Check power, cooling, network and operational resilience

Ask for the electrical single-line responsibility boundary, protection and isolation method, cooling design basis, alarm path, maintenance plan and restart procedure. Liquid-cooled installations also need leak detection, water quality or coolant control, pumps, heat rejection and a safe response to lost flow.

Internet bandwidth is rarely the largest requirement for mining, but stable low-latency routing affects stale and rejected shares. Confirm diverse routes where promised, pool failovers, network segregation and secure remote access. Miner administration should not be exposed directly to the public internet.

Resilience claims must be specific. A generator does not prove uninterrupted mining if it cannot carry the cooling system, has limited fuel, needs manual transfer or is not permitted to operate continuously. Ask what is actually supported and tested.

Review insurance, ownership and exit rights

The contract should identify each miner by serial number, confirm that ownership remains with the customer and explain liens or rights arising from unpaid charges. Verify the declared insurance scope, valuation basis, excess, exclusions and claim process rather than assuming the word insured guarantees full replacement.

Set out the notice, account settlement, shutdown, preparation, collection, packaging, carriage and storage process. A theoretical termination right has little value if the provider can withhold equipment under unclear charges or no removal timetable exists.

Consider provider failure, facility closure and relocation. Ask who pays relocation, whether the customer can collect through an authorised carrier and how configuration and wallet information are removed before the machine leaves.

Decide whether the hosting contract is dependable

When the service makes sense

The service makes sense when the facility can show a consistent measurement boundary, customer-level evidence, competent electrical and cooling operation, defined repair authority, documented insurance and a workable removal route.

Commercial terms should remain understandable under an outage, miner fault, curtailment event, late payment and contract end, not only during normal operation.

When to pause

Pause when uptime has no definition, exclusions are unlimited, the provider needs custody of payout credentials, serial ownership is unclear or access to historical data depends entirely on goodwill.

Do not send equipment until the final site, tariff, siting charge, minimum term, repair process and exit cost are written into the agreement.

Common hosting reliability mistakes

• Comparing providers by an uptime percentage with different measurement boundaries.
• Treating building power availability as proof that every worker submitted accepted shares.
• Ignoring planned maintenance, curtailment and customer-fault exclusions.
• Giving a provider unnecessary control of the payout wallet.
• Assuming a repair acknowledgement is a return-to-service commitment.
• Accepting insurance without checking value, excess and exclusions.
• Signing a minimum term without a priced and practical removal process.

Build a one-page service schedule with the exact model, quantity, power profile, location, price, measurement rule, exclusions, repair authority, credit and exit route. If the provider cannot complete it, the headline offer is not ready for comparison.

Frequently asked questions

What is a good ASIC hosting uptime percentage?

A percentage is meaningful only with its measurement point, threshold, exclusions and period. Compare the underlying definition and evidence before the number.

Should uptime be measured at the pool?

Pool accepted hashrate is commercially useful, but it can be affected by pool and network events. Use it alongside facility events and worker data.

Does a hosting provider have to compensate lost mining income?

Only the agreed contract remedy applies, subject to law and the wording. Many contracts use service credits or defined make-good capacity rather than speculative revenue.

Who should control the mining wallet?

The customer should normally retain wallet control. Hosting monitoring and pool configuration do not require the provider to hold the customer’s private keys.

What happens when a hosted ASIC fails?

The contract should define diagnosis, approval, warranty handling, carriage, repair cost, substitute capacity and the point at which billing changes.

Can I remove my ASICs at the end of hosting?

A credible agreement explains notice, settlement, preparation, collection, storage and any lawful right to retain equipment for unpaid charges.

Conclusion

ASIC hosting reliability is a chain of measurable outcomes: suitable power and cooling, worker-level performance, rapid fault detection, controlled repairs, useful records and a workable exit. Translate every headline promise into a definition, evidence source, exclusion and remedy. The strongest provider is not the one with the largest uptime number, but the one whose contract and operating records make the number reproducible.

Next steps

Review The Mining Shop UK’s hosting service and hosting terms, then ask for a site-specific schedule showing tariff, siting charges, uptime measurement, repair authority and removal terms before committing equipment.