Starting a deployment through your business.
Entity, capital, hardware, facility, deployment, records. The order matters more than any single step, and two of these are difficult to correct once you have moved past them.
3 min readGet the entity in place before you buy anything.
The equipment should be purchased by, invoiced to and paid for by the business that will own it. Buying personally and sorting it out later is avoidable work and it weakens the record at exactly the point the record matters most.
If you already have an operating company, adding an asset line to it is usually simpler than forming something new, and it keeps the activity inside a business you are already carrying on. If you form a new entity, get the bank account open and the bookkeeping started before the first invoice, not after.
One thing worth settling early with your accountant: how the activity will be treated, and what your involvement needs to look like to support that treatment. It is much easier to arrange your time deliberately from the start than to argue about it retrospectively.
The sequence.
Entity and banking
Formed, funded, with a dedicated account and a bookkeeping system that will hold the records.
Model it before committing
Work out what the deployment needs from the Bitcoin price to make sense to you, and decide in advance what you would do if it does not. This is the step most often skipped.
Decide the approach
Whether you are sourcing and monitoring yourself or bringing in support. This changes the fee structure and it may change your tax position.
Hardware
Model, quantity and where it is coming from. Efficiency in joules per terahash is the number that matters, not headline hash rate.
Facility and contract
Capacity has to exist before machines can go anywhere. Read the agreement properly, particularly the term, the rate change mechanism and the security interest.
Deployment
Shipping, racking, configuration, and confirming each machine is producing to specification. Record the energisation date per machine.
First production statement
The point at which you can see what it is actually doing, and the first document your accountant will want.
What drives the economics.
Four inputs do most of the work. Everything else is detail around them.
Efficiency
Joules per terahash. A more efficient machine produces more for the same power draw, and this is the specification that decides how long a unit stays economic.
The power rate
I model at $0.08 per kilowatt hour, which is the market rate for hosted capacity and is verified against competitors’ published figures. Be careful with anyone quoting materially below that.
Network difficulty
Difficulty rises as hashrate comes online, so the same machine produces less over time. I model 10% growth a year and state it. Fitted trend figures from recent history produce absurd results when compounded.
Hardware life
Roughly three to five years before newer machines outcompete these, sooner at lower prices. I model a five year horizon because machines die at around sixty months and longer horizons just add dead years.
The hosting fee is quoted in dollars per kilowatt hour, but it is a bundled service charge covering rack space, power, cooling, network and security, not an electricity bill passed through at cost. There is no meter reading behind it to audit, and comparing facilities on that number alone compares bundles rather than power.
Three things I would not do.
Buy hashrate instead of hardware
Cloud mining and hashrate contracts give you a claim on output without owning a depreciable asset. The equipment case for doing this through a business rests on owning equipment. Without that, you are buying exposure, and there are cheaper ways to buy exposure.
Treat the quoted rate as a power price
Covered above. It is the single most common way two facilities get compared wrongly.
Plan around a price you have assumed
Decide what has to be true for the deployment to work, then check whether you actually believe it. If your plan only functions above a particular Bitcoin price, know that number before you buy rather than discovering it afterwards.
How long it takes.
From a signed engagement to a first production statement is roughly eight weeks in practice. Most of that is not my work. It is manufacturer lead times, facility capacity windows and shipping, and none of the three compress reliably no matter what anyone promises at the outset.
I would treat any quoted timeline substantially shorter than that with suspicion, including one from me. The steps that can be rushed are the ones you would least want rushed: the contract review and the diligence on where the machines are going.
Work out whether the order makes sense for you.
Thirty minutes on your capital, your timeline and your entity. If the numbers do not work I will say so on the call. You can also read the hardware, pricing and what is deductible.
Book a discovery call No cost. No obligation. No follow-up sequence.