Where a host's liability ends and your own policy begins, which events fall between the two, and what to check in both documents before signing.
A fleet owner asks this question, in one form or another, only after something has already gone wrong. A breaker trips somewhere upstream and takes a row of machines with it, or a unit shows up at the dock with a bent chassis, and the first call is to the host, and the second call is the one where you find out whose problem it is. The question deserves an answer before that point, because the answer is rarely as simple as either side assumes.
What a host's liability usually covers, and where it stops
Most hosting agreements state, somewhere in the liability section, what the operator is responsible for and what it excludes. The common pattern is that the host is liable for its own negligence in operating the site, meaning failures in the systems it controls directly: power distribution, cooling, physical security. It is a narrower promise than it sounds, and it is usually capped, sometimes at a multiple of fees paid, sometimes at a fixed figure written into the schedule.
What a host's liability does not typically cover is the value of your hardware itself. In most agreements the operator is not the insurer of your machines. That is a defensible position as long as it is written down rather than assumed, and it means the physical asset risk (fire, theft, water damage, a cooling failure severe enough to cause heat damage) sits with whoever owns the hardware unless the contract says otherwise. We went through this same clause, and where it usually sits in the document, in our hosting agreement guide.
Your own policy, and what it needs to match
If the host is not insuring the hardware, the assumption is that you are. That means a property or equipment policy in your own name, covering the machines at the specific location they are deployed, not just in transit or in storage. A policy written for a different address, or for equipment in general rather than mining hardware specifically, is not necessarily going to respond the way you expect it to when a claim happens.
The detail that gets missed most often is the insured value. A used ASIC's market value moves independently of what it cost when new, and a policy that still lists an outdated purchase-price figure may under-insure the fleet without anyone noticing until a claim is settled at that lower figure. Keeping the schedule of insured equipment current, with serial numbers and a defensible current valuation, is the kind of task that costs an afternoon and saves a much longer argument later.
The gap between the two policies
Liability and property insurance are written to cover specific, enumerated events, and the two documents are drafted by different people for different purposes. That leaves a gap, and it helps to name what usually falls into it rather than assume it does not exist.
Transit is one. Once a unit leaves the manufacturer or a previous site and before it is formally received at the host, whose insurance is on risk depends on the shipping terms and on who arranged the carrier, and it is not automatically the host's. Our piece on hardware intake covers the receiving process that determines when custody, and with it risk, transfers.
Interim storage is a related gap. A pallet sitting in a staging area before racking, or held after a decommission while you arrange transport out, is not always covered by either the host's site liability or your own active-deployment policy, depending on how each is worded.
Consequential loss is the largest and least discussed gap of the three. Property insurance typically pays to repair or replace the damaged unit. It does not typically pay for the hashrate you lost while that unit was offline, or the difference in mining economics between the day it failed and the day it was replaced. Whether that loss is covered at all, by anyone, is a question to ask directly rather than assuming the answer is yes because the hardware itself is insured.
Proving value, and reporting a claim in the right order
A claim on a new unit is straightforward: the invoice states the price. A claim on a used unit, months or years into its service life, is not, and this is where claims most often stall. Keep records that let you establish a current value independent of the original purchase price: the price recent comparable units have traded at, the unit's efficiency and remaining useful life relative to current-generation hardware, and its maintenance and uptime history if that is relevant to value. An insurer working from nothing but a serial number and a purchase date that is long out of date has little to work with and every reason to offer the lowest defensible figure.
When something does go wrong, the sequence matters as much as the speed. Report to the host first, since site-level causes (a power event or a cooling failure) need to be established and documented on their side before a property claim can be built around it. Get the host's incident report, including what it says about cause, before filing with your own insurer. Photograph the affected units and the surrounding area if physical damage is visible, and preserve any packaging or components rather than discarding them before the claim is resolved. Log the affected serial numbers against the incident so the claim is tied to specific, identifiable machines rather than a general description of the row or the batch.
What to ask before you sign anything
A short set of questions, put to the host and to your own insurer before a claim ever happens, closes most of the gap described above:
- What does the host's liability cover, and what is the cap?
- Does the host carry any insurance on the physical hardware itself, and if so, for what?
- Is your own policy specific to this location and to mining hardware, or is it a general equipment policy?
- Does your policy cover transit and interim storage, or only active deployment?
- Is consequential loss (lost hashrate or lost revenue) covered by anyone, and under what conditions?
- What does each side need from you to process a claim quickly?
You do not need an insurance background for this. What it takes is reading both documents side by side and writing down, plainly, where the coverage meets in the middle and where it does not.
If your own hosting agreement is silent on any of the questions above, our hosting agreement guide walks through where that language usually sits and what it should say. Our operations page describes the access control, CCTV and on-site team at our facilities, which helps when your insurer asks how the hardware is protected.
