Comment by mrb | Hacker News Reader

Comment by mrb | Hacker News Reader

Actually Bitcoin needs to drop 94% for mining to be unprofitable. The break even is 1 BTC = $922. Yes, mining is insanely profitable right now. Miners are swimming in cash.

This is assuming $0.05 per kWh (which is a rate available to many miners outside of China, eg. Gigawatt Mining in Washington State pays $0.028 per kWh,) and a miner efficiency of 0.1 joule per gigahash (such as the popular Bitmain Antminer S9). The break even point is calculated with:

j_over_gh × cost_per_kwh / 3.6e6 (joules per kWh) / reward × (2^32 × diff) / 1e9

The parameter are: j_over_gh = 0.1, cost_per_kwh = 0.05, reward = 12.5 BTC per block, diff = 1931136454487.72. (This ignores transaction fees which have pushed the average block reward from 12.5 to around 17 BTC over the last few weeks, so the true break even point is even lower, around 1 BTC = $680).

I am occasionally interviewed by journalists who write about Bitcoin mining and they often get many details wrong. This one is no exception—Dan Murtaugh should have contacted me :)

I know hardware is a sunk cost, but you should probably put some amortization in there.

You are right. However even sunk costs have been recovered very quickly lately. To give an idea: an S9 deployed in early November has already fully recouped its cost ($1500 in 40-60 days).

Honest question (I have no actual ASIC experience, I'm just curious)

Where are you getting the USD1500 price tag, I've just done a quick search and I only found a bunch of S9 Miners (Bitmain Antminer S9) advertised for over USD6000 on Amazon.

The batches sold and shipped in Nov were priced $1500. That was the price when buying directly from bitmain.com. Since then Bitmain hiked the price: the latest batch released yesterday (now sold out) was priced $2300. But yeah, since Bitmain's batches sell out in minutes, S9s often reach ~$5k in unofficial resale channels.

In that case, wouldn't it be more profitable for Bitmain to keep what they make and mine coins themselves?

It is if they want to bet on the value of bitcoin. Why use some hardware for 40-60 days to mine some coin which may or may not be worth $1500 when you could sell immediately for that amount? Sure you can mine for longer, but I'd bet they have more stable returns by investing profit from sales back into the business for fast/cheaper manufacturing, expanding their area of operations, etc.

"During the gold rush its a good time to be in the pick and shovel business" - Mark Twain

the same analogy applies here, BitCoin mining is quite profitable, however hardware business is more reliable and in demand.

Pretty sure they do.

https://www.reddit.com/r/BitcoinMining/comments/7jg14f/one_o...

Yes, and they operate a decent amount of mining farms. However mining comes with a certain risk (eg. Bitcoin could crash) which is eliminated when selling the hardware right away. They can also manufacture hardware a lot faster than they can deploy data centers.

This means that miners in these areas operate with a 1500% profit after buying hardware equipment (1500% is computed with current price of 1btc=$16000 and the break-even point at 1btc=$1k). So are they scaling up and buying more hardware as fast as they can?

Yep, the best ASIC Bitcoin mining hardware is almost always sold out... at least until a new better ASIC comes out.

and the grapevine says that they are mostly refurbs from actual mining operations that already have a new unpublicized version running.

Right...why would this middleman sell them? Use them for your own operations!

They've earned the investment back many times over and the newer devices are 2-4x as efficient, probably?

There have been several instances where the manufacturer mines with each manufactured device for a few months "for quality assurance purposes" (enriching themselves at lower difficulty levels, and then shipping after they've pushed difficulty up).

The large mining operations all own factories that build the ASICs.

To put it mildly, that seems pretty unlikely given the ridiculous amount of know-how and capital investment required for chip manufacturing.

Might not be that would-be miners are acquiring that know-how, might be that people with the know-how tend to get into mining.

He's probably talking about the ASIC design, not the manufacturing.

So what ASIC boxes should I overnight to my house?

Antminer S9 is what you want. Good luck finding one. They should be $2500, but people are gouging up to 3x that for them.

How many machines would you need to be sure that you actually mine a block in a given year? AFAIK its a stochastic process right? I think I calculated once that , given the current hash power, you'd need about 20 miners in order to have P(you mine a block within a year) close to 1. Can you confirm/discard my conclusion?

I'd post my calculations but I don't have it here at work!

Don't all fleets mine as pools these days? You put your hash capacity in a pool, and if any machine in the pool mines successfully the BTC is shared according to contributed capacity.

That's right, at current diff (1931e9) you need about 19 S9s (14 TH/s each) to mine a block in a year:

19 × 14e12 × 3600 × 24 × 365 / (2^32 × 1931e9) = 1.01

Mine pools.

Does the supply of miners not affect the reward? What is the average ROI for these miners? Do you know?

It does. The bitcoin protocol adjusts its difficulty every 2 weeks by analysing the number of miners in the network. The adjustment is such that a block is discovered every ~10 minutes.

Not the number of miners but the average time it took to mine a block.

Reply on news.ycombinator.com