Bitcoin miners are power dense Chinese manufactured computers with varying challenges to operate and often have significant failure rates. One of the hardest calculations to make when assessing any hardware is anticipating their performance in differing environments and at different scales.
The universal challenge all miners face, is that the durability and longevity of a miner is unknown until it actually ships. Miners can make educated guesses based on historical information, but that does not guarantee future reality.
There are general assumptions we can work with, but in no way do those assumptions necessarily translate into reality. Buying mining hardware is always a gamble, but you can minimize your risk by being properly educated on the topic. Unfortunately, the majority of the industry gets the necessary education through trial and error, and so I want to share some of my experience on the topic.
Longevity Due To Design:
The most famous example of machine design issues is Bitmain's S17 which had catastrophic failure rates, making it one of the most infamous miner models. Many people preordered expensive machines, expecting to have the latest cutting edge hardware, only to find out the heat sinks were falling off and the boxes would rattle when you picked them up and shook them.
The design of the miner matters, and so does the design. A well designed machine will minimize failures, operate with minimal downtime, and will be fairly easy to repair in the case it does break. A poorly designed miner will have high failure rates, could be difficult to repair, and will cause migraines.
Downtime and labor to service machines are costs that many miners do not consider when first exploring deploying equipment.
A machine may be designed very well, but may not be designed to handle certain conditions well, such as Texas dust and heat. They may operate flawlessly in the correct environments, but perform horribly in suboptimal conditions. Knowing where you are deploying your machines should have a major influence on what your purchase strategy.
Operating Conditions:
Manufacturers publish ideal operating conditions for miners, but a good portion of miners do not run them in the proper conditions. Bitmain is often criticized for being unreliable, and yet they have up to around 90% market dominance based on various estimates. Of the equipment Kaboomracks sells, roughly 90% are Bitmain machines. The market has chosen Bitmain despite their history of making poor miner models like the S17 because they are consistently the cheapest, and most efficient machines on the market.
Generally, Bitmain made equipment runs great in temperate environments with the correct operating conditions. The problem is that miners often don't have the ability to deploy in optimal environments, and need their initial capital expenditure to be as low as possible. It is not cost effective to run miners under air conditioning, and humid environments rule out evaporative cooling.
To offset this, miners have to find a way to pass as much air as possible through the mining computers to attempt to keep them cool. Dust and debris passing through the machines causes wear and tear, shortening the lifespan of the machines. Things like filters, which are used sometimes to mitigate these issues cause air restriction which makes cooling challenging.
Running machines in suboptimal conditions will significantly shorten the lifespan of mining hardware, but this is a calculation that works often for some miners. Sometimes the reduced costs in Texas, or other places can justify putting your equipment through hell, but it should also influence which equipment you choose to purchase, or even if you should deploy at all.
Used vs New Machines:
One of the benefits of buying used equipment is that there will be a proven track record of how that machine performs. It generally takes at least a year for a machine to become available on the new market. As time goes by, the market for used machines will grow larger.
It takes at least a year to get an accurate picture on the quality and durability of a machine. The S19 J Pro has proven to be the most durable and versatile S19 variant. Other s19s such as the vanilla S19 and S19 XP have shown to be a lot less reliable.
Waiting to get feedback on the quality of a particular miner after it is released gives you an advantage as far as knowing how it will operate in different conditions. New machines are attractive because they come with a manufacture warranty. The standard timeline for this warranty is 12 months starting from when they are shipped from the manufacturer.
Understanding the warranty process of each different manufacturer is important. Not knowing this could cause you to unintentionally void the warranty. This is a whole rabbit hole, I won't get into this post.
One of the challenges with used equipment is verifying the vender is representing the equipment properly. The equipment does not come with an odometer, or information on how it was ran. Equipment ran in suboptimal environments will not last as long as equipment that was ran in optimal conditions.
Knowing how long a machine has been on the market can give you an idea of what the age could be. The longer the machine has been in circulation, the more likely it is to have issues.
Yolo a Preorder:
Preordering machines is attractive because it generally allows you to purchase the equipment for less than the spot market. Sometimes the difference is significant. Preorders are not always shipped on time, so it can be a difficult process to deal with.
There is a massive market built on speculating on machine prices. Given the right conditions, yoloing into a preorder could be an incredibly lucrative decision. I believe that right now could definitely be one of those situations with the S21 Pro. Individuals have the ability to buy machines today for the fraction of the price that they will trade for when the bull market takes off.
The economics of everything change in the bull market, where the market only cares about raw capacity and hash rate instead of efficiency and durability.
Firmware, Underclocking, overclocking, and longevity:
Aftermarket firmware is a huge topic of discussion, and for many people has helped them be successful in the bear market by unlocking efficiency that stock firmware doesn't allow for. The hard reality is that using aftermarket firmware adds variables and can shorten machine longevity. Changing stock settings from the manufacture can have unintended consequences, especially when we know so little about the machines.
Underclocking means lowering the amount of electricity that is passed through the ASIC chips and power supply. Less electricity consumed means less heat, and theoretically less strain on the machine. The fans don't need to run as hard to keep it cool, meaning in the summer, this could be very helpful for machines that consistently overheat.
Underclocking machines has the potential to increase the life cycle of machines. Overclocking on the other hand has the potential do the opposite. When overclocking, there's a few pieces of information that are important. You need to know what the power supply can handle, the hashboards, and the chips themselves. Vnish firmware gives you a great visual of chip performance as it actually shows the hash rate of individual chips and lets you tune them.
With Vnish, an immersion setup, and an aftermarket firmware, Kaboomracks was able to push 10kW successfully through a S19 J Pro which was rated for around 3kW stock. Just because you can do it, doesn't mean you should do it, but it can be fun to see how far things can be pushed.
Firmware can lengthen the relevancy of older equipment, as underclocking could make older equipment more competitive with new models.
Evaluating Different Manufacturers:
This topic is a rabbit hole. Different manufacturers have different costs, different warranties, and different reputations. Just because one manufacturer has a better reputation does not guarantee you won't receive a lemmon batch, but it can lower the chances.
One guarantee is that when you deploy at scale is that you will have broken equipment no matter how great your operating conditions are. For that reason you need to understand the market for repairs and parts. If there's a new manufacturer, there likely will not be an established market for fixing the machines, which in my opinion makes gambling on purchasing the equipment unattractive.
Micro BT is consistently held up as making the most reliable equipment. They have replacement parts readily available, but those parts are often really expensive compared to Bitmain. The costs of parts is an important factor to consider.
Conclusions:
The mining industry is immature. There are not very many good sources of data to make informed decisions and most of the information that is available is anecdotal. Understanding the potential lifecycle of machines is difficult as there is a lot of nuance. There is an incredible information asymmetry between those who have operated in the industry, and those looking to enter it for the first time.
Mining is becoming more and more competitive as the fiat world now tries to enter. Mistakes and failures will only grow more costly, as margins continue to get cut. Be careful out there.