Hacker News Reader

Comment by xentronium | Hacker News Reader

Could someone please explain why Butterfly Labs don't mine the coins themselves, but sell their devices instead?

As far as I understand bitcoin (and please correct me if I'm wrong), bitcoin output of the system is fixed, so if you produce (or buy) a lot of super-duper fast ASICs, they will generate less bitcoins per unit. If it is correct, then butterfly labs success inversely correlates with specific miner success, as in the more ASIC-s they sell for a fixed dollar price, the less profit their buyers receive. This seems somewhat conflicting to me.

View on HN

Here are 4 simple reasons why BFL sells the devices. I explain this over and over to non-Bitcoiners who don't understand...

Developing an ASIC requires manufacturing a lot of wafers to recoup your NRE cost, mostly the mask set cost. In the case of BFL, they manufactured 75-100 thousand chips in the first batch, or a few dozen wafers. Each chip consumes ~15-25W (there are two of them in a 30-50W Jalapeno) and each chip is capable of 3-4 Gh/s.

Firstly, if you do simple math on power consumption, you find out that running 75-100 thousand chips would need a datacenter of 1.1 to 2.5 megawatt! It would take many more months to deploy and maintain up to 2.5 MW of hardware, causing BFL to be unable to generate mining profits quickly. It is a lot easier for an electronics manufacturer to merely ship hardware to end-users, than to deploy megawatts of hardware in datacenters.

Secondly, if BFL mined with 75-100 thousand chips, they would mine at 225-400 Th/s. This is 2x-3x higher than the current global hashrate (~150 Th/s as of June 2013). So they would instantly cause the difficulty level to increase by 3x-4x, therefore reducing their mining profits to only 1/3rd-1/4th of what one might naively conclude by using a Bitcoin profit calculator today.

Thirdly, there is intense competition from other ASIC manufacturers, who are already up and running! A chinese company known as ASICMINER currently already operates 40 Th/s of ASICs as of June 2013 (roughly 150 thousand of their own 130nm chips), and claims they will soon deploy 250 Th/s to keep up with the expected competition. Also, Avalon has currently shipped their first batch of 20 Th/s of devices to customers, with their second batch of 40 Th/s being shipped as we speak (as of June 2013), and a third batch of 40+ Th/s in the next months. All this to say that it would be very risky for BFL to just plan to mine for themselves, when they are already late compared to competition. Combined with my points #1 and #2, by the time BFL catch up with competition and mines with megawatts of hardware, the difficulty level will have increased by at least 10x. At this point, clearly it makes sense to sell the devices to instantly reinvest the sale proceeds to build more, rather than spending 3-6 months to mine to (hopefully) recoup the manufacturing costs.

Fourthly, mining is extremely dependent on the exchange rate of Bitcoin. If BFL sell/pre-sell a device for USD as they did, they lock their profits right away. Whereas if you plan to manufacture, then mine, then sell the bitcoins, anything that happens to the exchange rate during this timeframe puts your future profits at risk.

Hopes this help.

(Edited version of what I originally posted to http://arstechnica.com/civis/viewtopic.php?p=24459357#p24459...)

They used pre-order money to fab the devices. That means they already owe devices to their customers.

In a gold rush, it's often better to be the one selling shovels.

People keep repeating this meme. But in this case, setting up a mining business is trivial. If these devices will have a payback period of ~10 days in the near future, it does not make sense at all to just sell them to end-users.

What gives?

My completely uninformed guess is that they are. If I were the one selling these things, you'd better believe that I'd have a room full of them mining bitcoins, and my room would take priority over sales. But I'd also sell them, because why not?

Diminishing returns: The amount of mine-able bitcoins decreases overtime, as well as the amount of reward decreasing as the hashing power of the network increases. Assuming the price of Bitcoins does not increase accordingly, the amount of money gained will go down.

Bitcoin Collapse: If Bitcoins collapse, you now have a bunch of worthless miners. Compare this to a quick sell it and done.

God forbid you gain a significant amount of network control, because you have a problem. Miners will probably begin to reject your transactions for the sanity of the network, because you could double-spend, and virtually create infinite bitcoins.

Also, how is setting up a business like this trivial? I imagine that a huge part of it (initial investment, buying power, and maintaining all these devices) could be problematic, although I could be wrong.

I fully expect that they will start mining themselves as soon as they fulfill all the existing orders.

A peculiar feature of the ASIC market is that unless you are making really a lot of chips, it's entirely reasonable to assume that the marginal cost of creating one more chip is zero. If your production volume is below tens of thousands, almost all of your costs are going to be masks and design. In that way, selling an ASIC is like selling software -- at first you need pay huge sums for development, and then you can duplicate your product at (almost) no cost.

So what probably happened is that Butterfly labs realized a market opportunity, but they didn't have the money to exploit it. So, the presold some millions of dollars worth of hardware, used that money to design it and make the masks, and then delivered the first few batches to fulfill those orders. And at the end of the day, they don't just have whatever profit they made on the miners, they still have the masks, and can now make more miners for peanuts.

> But in this case, setting up a mining business is trivial.

Not exactly. I'm not exactly sure on the up front cost to develop ASICs, but it's definitely well into 6 figures. BFL sold millions of dollars in pre-orders 12 months ago to fund the design and development of the devices. Everyone in the bitcoin world knew where it was going technology-wise last year, but only 2-3 companies have successfully developed ASIC miners 12 months later.

There are many people who pre-ordered devices from BFL over a year ago and still haven't received them. It seems like BFL would be guaranteed a lawsuit if they started mining on their own before fulfilling those year old orders.

If you look at the market depth (the V-shaped plot at coinlab.com and elsewhere), you can see that ~$1M is sufficient to shift the price of BTC, which has a ~$1B capitalization by ~10%. Thus, if you wanted to make more than ~$1M from BTC, it would have to happen by doing something other than selling BTC itself, lest you crash the market.

The price is supported by speculators (like me; I've bought a little to play with) who think that the value of BTC at some time in the future is at least $100. If it were priced at its utility, at least in my simpleminded picture of things, a shift of 10% in the price should come only with a sale of ~10% of its capitalization.

The more centralized mining is, the less bitcoins will be worth. They could have accidentally killed it, if they acquired too much hashing power. A mystery why the payback period is set so generously though.

It's not exactly a shovel, more of a Bagger 288 [1] when everyone else only has that shovel.

[1]: http://en.wikipedia.org/wiki/Bagger_288

However, those that get the richest are in fact always the mining companies.

The top miners of bitcoins have gotten radically richer than those selling mining tools ever will. In this case the mining will run out, the bitcoins will remain and grow more valuable due to supply / demand (assuming bitcoin succeeds longer term).

There are far more $5+ billion mining companies, than there are $5+ billion mining tools companies.

Go down the list of the top 20 mining or natural resource extraction companies. The tools business is always radically smaller by comparison. eg: BHP Billiton, $156b market cap; Rio Tinto, $76b market cap; Vale $68b market cap. And of course the oil industry is far more dramatic.

Selling tools like the bitcoin mining boxes is foolish, unless you're selling lemon tools that won't recoup their investment (people will stop buying them quickly), or unless you're betting bitcoin won't thrive long term. If you think bitcoin will thrive, it's far better to mine each coin yourself and own it long term, and ride that extreme appreciation. The return would be far beyond any margin you could earn on boxes.

Hindsight is 20/20: It also could have happened that bitcoin prices are down to $1 right now and then BFL and other shovel sellers would vastly outperform dedicated miners.

Another possible reason is that if a small number of companies start using ASICs, and they're the only ones that use ASICs without selling them to "regular" users too, they'll quickly gain a large percentage of the hashing power. If a company like ButterflyLabs controls a significant chunk of the total hashing power, it'll enable them to do 51% attack [1] on the network, which'll lower the trust users have in the currency, leading to decreased value.

Its a bit paradoxical - people will want as much hashing power as they can get, yet if they get too much of it they risk destroying the currency value, leaving them with nothing.

[1] https://en.bitcoin.it/wiki/Weaknesses#Attacker_has_a_lot_of_...

from the article

"the second reason [they aren't just mining themselves] is far more practical: bitcoins, for all their current value, are still speculative. A large amount of US dollars is a large amount of US dollars no matter which way you cut it. It can be readily exchanged for goods and services. Bitcoins themselves don't necessarily hold value, and it's difficult to exchange them in large quantities for an equivalent amount of dollars. "

BFL can't afford the NRE to develop an ASIC alone for one. Another is that if any one miner gets so big that they could 51% the network, the other network participants would likely shun them, reject their work. Another is that it isn't usually great business to compete with your customers.

Yeah... I'm very surprised that any of these have shipped at all; it seems like any MBA in charge would want to keep the devices on hand for a while longer for 'extended burn-in' You know, quality control; we can't send devices out to customers if they are going to conk out in a few months.

That's what they did. They kept them for about a year, to give them a "burning test". People who ordered last year about this time of the year, just now are receiving what the product. People who ordered on say, September, are still waiting. Nobody saw a ROI in Bitcoin with these mining boxes (though any investment related to Bitcoin will probably see a nice ROI in dollars). People would have made more profits if they invested directly in Bitcoin, instead of falling for this scam.

To be clear... this is my 'mostly joking' paranoia. I've got no evidence. The situation does seem like buyers are placing a whole hell of a lot of trust in the manufacturers.

Reply on news.ycombinator.com