Thursday, March 27, 2014

ASICs vs. GPUs, March 2014 Update

Last week I found what I felt was a reasonable deal on a Bitcoin ASIC, so I decided to give it a try and bought a couple Bitmain AntMiner S1 ASICs. Each is good for 180GH/s SHA256, or when overclocked around 400GH/s combined. At a price of about 1BTC each, I thought it was worth a shot. Now I'm thinking I may be lucky to break even. Let me run some numbers for you.

Lately, like since about March of 2013, Bitcoin has been pretty consistent about increasing in difficulty by 20% or so every 12 days. In fact, I can do one better and say that the average increase every 2016 blocks since last March is... wait for it... 23.73%. That's pretty massive when you think about it, and it's the reason why difficulty has gone from 6,695,826 on March 24, 2014 to a staggering 5,006,860,589 today. Put another way, the difficulty -- and thus the returns for mining -- have changed by a factor of nearly 750. And what's worse: there's no sign of slowing down just yet.

So let's take a closer look at the AntMiner S1. At a price of $600 (around 1 BTC), assuming you have spare power supplies and everything else you need, we can get a pretty good idea of your profit estimates using any number of online calculators. Personally, I think the calculator over at BitcoinWisdom is one of the best, and it's what I used. Since I now have my AntMiner S1 ASICs up and running, I know power, performance, etc. With the current difficulty and running at 400GH (overclocked) with a power draw of 850W, my two AntMiners are looking at the following scenarios:

1) 15% difficulty increase every 2016 blocks = 233 days before the power costs outweigh the income. I can make around $454 in that time off my ~$1200 investment. This is basically the "best case" scenario.
2) 20% difficulty increase every 2016 blocks = 177 days before it's no longer worth mining. I'll only earn about $70 off my investment.
3) 23% difficulty increase every 2016 blocks = 163 days before I should stop mining; I'll lose $82 based on current prices.

Now, obviously the price of BTC is the big wild card. If we see $1000+ again, and I think we will some time in the next 9 months, I easily earn back far more than the initial investment. If we take a nose dive on the other hand and don't recover, I stand to lose even more of my investment. So it's a risk, but a calculated one. And in another 10 days or so when the difficulty jumps up 20%, things start looking even worse. At $600 for an S1, you'd stand to lose $70 in the "reasonable" 20% increase scenario -- a swing of $140 in just 10 days (and the reason ASIC prices are tumbling).

If you're thinking, "Okay, but the AntMiner is a 55nm ASIC -- what you need to look at is a good 28nm ASIC," well, you're only partially correct. If you took a chance and pre-ordered a Terraminer IV a while back, you're rolling in the BTC right now. If you order one right now from the June batch however, let's be optimistic and say you get it in 60 days (before the end of May). With difficulty jumps of 20%, you'd end up losing half your investment! Even smaller 15% increases would still result in a loss of over $1000. That's a pretty sobering thought -- one of the fastest, most efficient BTC ASICs right now is looking like a terrible risk for future buyers; either price needs to come down or difficulty needs to stabilize. BTC prices basically need to double for you just to break even, and the rate of difficulty increases will likely slow down before then, but be careful.

What about the scrypt ASICs? Well, Alpha Technologies is now promising 90MH instead of 25MH, which is pretty great to hear. Depending on when the hardware arrives and the difficulty at that time, however, it's not looking so hot. Let's say it ships in just 60 days (which is very aggressive). At a moderate 5% increase in difficulty every LTC cycle that would mean mining LTC would never recover the initial investment cost. Even if we assume you can beat LTC returns by 50%, you'd still only make about $1850 on the $10,000 investment. I suspect those who pre-ordered the first batch will do pretty well, but the future (most likely) second and third batch orders are far more suspect.

The big problem with ASICs of course is that once it becomes unprofitable to mine with them, they're essentially useless. With a PC, you can at least use it to run Windows (or Linux or whatever if that's your thing), play games, do real work, etc. You might not need dozens of GPUs, but you can at least recover some of the initial investment. Perhaps you'll find some people willing to buy your outdated BTC ASICs as well, but I'd expect very low prices at best.

In short, ASICs are mostly going to make a lot of money for the people selling the ASICs, not for those buying the ASICs. It's the Gold Rush all over again, where the real winners are the people supplying the miners. It will certainly be interesting to see where pricing and difficulty stabilize over the coming months, but don't be surprised if a lot of the Scrypt ASICs turn out to be losing propositions. Unless of course we see a repeat of last November where prices shot up 20X on Litecoin in only a week or so, which is what we're all probably hoping to see. :-)

There's more available in the newsletter, as usual. Just let me know you'd like to start your subscription with the 3/26 email and I'll make sure you get it!


  1. Hi Jarred, I've been mining for a while and would like to at another miner, would it be still more profitable ans safer to invest in GPUs than Script Asics? Please give me your views.

    1. I cover this a bit in my newsletter; short summary: unless prices go up quite a bit, I wouldn't be looking to invest heavily right now. But then, a lot of folks figure on BTC hitting upwards of $2000 this year, and $10000 within the next couple of years. If they're right (and they might be!), then pretty much any investment into cryptocurrency mining would pay off handsomely within the next couple of years.

  2. Wait a can overclock an Antminer S1 up to 400 EACH?! ...ok, so I have 3 Antminers doing the standard 180 g/hs. You said that you were pulling 850w from the wall, but does that mean that you have an 850w PSU for each Antminer? What can I achieve overclocking with what I have...which is two Corsair 500w and 1 Enermax 600w? Thanks!

    1. LOL, no -- sorry that wasn't clear. It's around 180GH for one, and I bought two, so combined with an overclock around 400GH. I've updated the text slightly to make this clear. Each draws about 450W overclocked (with an 80 Plus Gold PSU), and I'm running them off 650W PSUs.

  3. Hi Jarred, I've already done the same math as yours. But don't you think is it better buying bitcoin (or litecoin) directly in an exchange such as btc-e?

    1. It can be, sure, but then you're basically just playing the role of speculator. With hardware, you have something more tangible, and you might be able to earn a profit without prices going up. The problem with speculating for me is figuring out when to buy and when to sell, and I often seem to guess wrong. LOL