Chip designers see dollar signs in Bitcoin miners | Hacker News Reader

Chip designers see dollar signs in Bitcoin miners

reuters.com

41 pointsdiycoin39 comments

If this site have it right, is not profitable at all

http://mining.thegenesisblock.com/a/b0c353b122

However, is it? Is possible to invest in this and get some dollars per month as passive income? (Or maybe with https://cex.io/promo/ or https://products.butterflylabs.com/homepage-new-products/1-g... ?)

First of all, it's based on the current price of bitcoin.

Second, if you have access to free electricity (e.g. solar), it is profitable at almost any price.

> First of all, it's based on the current price of bitcoin.

Can you describe a realistic scenario where the price of Bitcoin rises and owning a miner produces more profit than simply just investing that starting cash in Bitcoin itself?

> Second, if you have access to free electricity (e.g. solar), it is profitable at almost any price.

This is false. Diminishing returns from difficulty increase makes it entirely plausible that you will never realise the initial starting capital you invested, and even if it does there is no situation I can think of where investing that starting capital in mining HW beats the profit on just buying Bitcoin in the first place.

I calculated this out a couple days ago. The hash rate has been rising pretty steadily exponentially with all the new ASICs coming online, and can be expected to double again within 25-30 days. The current break-even price for 1 GH/s is about 0.03-0.04 bitcoins -- that is, if you bring 1GH/s online right now, over the next 6 months you will mine a total about 0.035 bitcoins and at that time you will be minting at less than 1% of the rate you started with.

You can buy 1GH/s for about 0.12 bitcoins on cex.io right now, if you are a sucker.

If the price rises then you'd have been better off buying and holding BTC.

Even with free electricity, a box that costs $20K and makes $1 per month isn't what I would call profitable.

Maybe I'm wrong, but unless you are ready to bring a chip to market, it seems to me this ship has sailed.

That said, I'm a chip designer with ASIC experience. I think it would be pretty fun to work with a couple software oriented folks to design and build something here.

ASIC design software tools would probably require funding, but FPGA prototyping can be done at very low cost.

Yeah, the ship has pretty much sailed. Products are either being delivered or are very close to shipping.

Some of the smaller operations worked much as you imagine, though. They did FPGA prototypes and then raised funds to move to ASICs.

They often raised funds on the bitcoin-denominated stock exchanges. A handful of guys could raise money via an ipo, get the chips produced, open an operation, and then share the revenue with shareholders via dividends.

Some operations were just a couple of project manager outsourcing all the engineering and manufacturing. In at least once case, the ASIC design was done via ODesk (!)

It was a risky business, on both ends. The investors rarely knew much about the people they were giving money to, and the entrepreneurs were taking money from people they didn't know much about. With no guarantees that anyone had any idea what they were doing.

Those companies are reaching their endgames now, having either worked out ok or failed. Millions of dollars have been made, and much money has been lost.

It's been fascinating to watch. It's a crazy pace and intensity even for a bunch of startups.

There's a nice book in telling their stories, if anyone could actually find any of the people involved. Ha.

Worth looking at for their stories- ASICMIner, the first wildly successful ASIC IPO.

Active Mining, a rare US based company that IPOed.

LabCoin- A company that may or may not have been a scam, but is in the process of failing.

Like you said , this ship has sailed.

Maybe there's an open market in memchached acceleration(basically computing hashes). Convey does this with FPGA's[1], and offers 3x performance/$.

Or maybe other cloud compute acceleration stuff.

I wonder though: if we're talking about using someone like easic.com( so scaling a business would be less risky and more investor friendly, since they can do low and mid volumes) ,How would the investment costs would look like(for 45nm/28nm) ?

[1]http://www.enterprisetech.com/2013/09/24/convey-revs-memcach...

eASIC is who the mining company Active Mining I mentioned below partnered with.

http://thegenesisblock.com/easic-announces-24-ths-hash-fast-...

I wonder if anyone have designed chips to crack DES to actually sell. I am thinking of MS-CHAPv2 in particular.

Cloudcrack has a 40-core FPGA solution that allegedly can break MS-CHAPv2 in a day[1]. Looks like they charge $100/lookup[2].

[1] https://www.cloudcracker.com/blog/2012/07/29/cracking-ms-cha...

[2] https://www.cloudcracker.com/dictionaries.html (in the mouseover text in the graph for brute-forcing DES keys).

Not sure about MS-CHAPv2, but you've seen this?

http://www.copacobana.org/

For crypto applications there isn't a strong enough financial incentive to design an ASIC specifically to do that. Maybe a GPU-based solution like oclHashcat[1] can be extended to do that.

[1] http://hashcat.net/

There might be a financial incentive to make hardware to crack crypto to sell to governments, which could use the hardware to spy on people. In other contexts government contractors do make surveillance hardware, like IMSI catchers. Because the crypto involved is so easy to attack, it's possible to imagine someone building a passive cell phone monitoring box with ASICs in it.

I haven't looked through the relevant stuff in the Spy Files (vendor presentations from the ISS World surveillance conference), but that might be one place to look to find people making crypto-cracking ASICs for a profit.

http://wikileaks.org/the-spyfiles.html

They only have documents through 2011, although the conference is still happening frequently.

How do you know government intelligence agencies don't have that sort of hardware already?

I myself have been thinking of starting a business selling pickaxes, gold panning sluices, and tulip bulb display cases. Am I too late?

Yes. But you're not too late to make strawman comparisons to perishable commodities with low inherent value. That market is still quite strong.

Bitcoin right now as it stands is a cash cow for everyone but the miners.

The only miners making a profit either already have theirs in hand or will have the next-gen's by end of this next week. Difficulty has risen to 4PT in a span of three weeks when it took years just to reach 1. HashFast's "Miner Protection Program" is the only offering at least promising returns. The rest of the miners were rushed to market and skipped a number of standard steps in their design, testing, and manufacturing processes. I'd buy straight bitcoin if I were anyone seriously considering an investment.

I don't really understand how bitcoin works.. however is there a point where generating the new coins or doing the computations required is no longer profitable. At that point, how would the bitcoin system survive if no one is willing to provide the necessary computational power?

There is a fixed curve that determines the rate that new Bitcoins come into the Bitcoin economy. Right now, that adds 25 BTC per block (or about 3600 BTC per day, or $720,000 USD/day). The number of coins added per block halves about every 4 years, and has halved once already. This is the primary method of introducing a sort of 'inflation' to the economy to help bootstrap the Bitcoin economy and the miner network. [1]

The other method of rewarding miners to perform block hashing proof-of-work computation is through transaction fees; even now some (most?) miners do not accept transactions without a small fee attached, currently an equivalent of a few USD cents per transaction.

The idea is that at some point, transaction fees will have to take over from the block reward as motivation for mining.

Regardless of the costs involved, since the work factor corrects for the amount of mining activity, even a single interested person mining is enough to keep the Bitcoin network going (though that invites some bad scenarios - a group of people is much better...) As things are going and with ASIC mining becoming commonplace there will definitely be enough interest to keep things going on the technical side for the far future. [2]

[1] https://en.bitcoin.it/wiki/Controlled_Currency_Supply [2] https://en.bitcoin.it/wiki/Mining

also, the FAQ:

https://en.bitcoin.it/wiki/FAQ#How_does_the_halving_work_whe...

Most bitcoin transactions include what's called a "miners' fee", an arbitrarily small amount put there as an incentive for miners to include your transaction in the blockchain, therefore speeding up the verification of said transaction. These fees will be there even after the last bitcoin has been mined, so there will always be an incentive to mine (provided that bitcoin is still in use by then).

https://en.bitcoin.it/wiki/Transaction_fees

Transaction fees, which are specified by the sender of a transaction at the time that it's sent and are collected by the miner of the block that incorporates it.

I just steer well clear of mining hardware companies. Business model makes absolutely no sense and still haven't seen a good explanation as to how it ever could.

If you're selling the promise to make $x and selling it at <$x someone at some point in the chain is a mug.

It is not if you get the cash upfront and can reinvest to produce more machines.

Good point. Skeptical if that sole reason represents all of their business models though?

Stories like this should remind BitCoin advocates to look into the history of the gold rushes and who got rich. Hint: not the gold miners.

Well, early adoption also really paid off (whether you bought or mined), unlike the gold rush.

I wonder if it's now worth it to quickly adopt all new crypto currencies for a disposable amount, even if 90% of them fail.

> I wonder if it's now worth it to quickly adopt all new crypto currencies for a disposable amount, even if 90% of them fail.

Nope. They're all get-rich schemes by the creators, and nothing more than just Bitcoin with the name and some variables changed. Ask yourself, why would HackerNewsCoin be any better than Bitcoin?

First to market advantage is huge. Maybe Bitcoin wasn't first to market in the strictest sense but they were first to gain notable traction which is the same thing in essence.

I would never invest in one of these companies. It seems like the ultimate bad business -- the difficulty will keep increasing while the payout will keep decreasing until it all ends some time in the non-too-distant future.

On top of that, this seems a little bit too abstract. If bitcoin flops, all of this effort will have been expended for no net benefit to society.

Unless this hardware can be repurposed.

That's not a fair criticism. Do failing startups offer no net benefit to society? If so, who cares? Let's try new things.

If you think it's unfair, I'd appreciate you saying why.

> If bitcoin flops, all of this effort will have been expended for no net benefit to society.

That's not fair, Bitcoin has the potential to offer vast benefits to society on a world wide level. Why are we worried about repurposing hardware in this context if it fails? At the moment it hangs in the balance, if it does fail it's not been all for nothing, it's got people thinking talking and acting on big issues. If the cost of this is some usless computers then so be it, a very small cost to pay.

It would suck if it this stuff can't be repurposed but in the wider context it's totally irrelevant. Failure and no benefit to society is not a strong relationship.

"Silicon is Perishable": With all of the computing power perishing, I wonder if outdated hardware could be bought on the cheap and used for some non-financial projects. Could this become a boon for malaria research, SETI, or some of the other non-profit research programs out there?

Bitcoin ASICs aren't reprogrammable to perform other computations -- the hardware (at least the really fast part of it) isn't Turing-complete.

People have speculated about whether other kinds of computational problems can be represented as Bitcoin hash searches but I think the answer is pretty clearly no, because the computations involved are so specialized.

OK, so pre-existing general computational problems likely cannot benefit. But if someone were to design a new system with carte-blanche, what constraints on that system's design would still allow them to exploit all this mining hardware? Put another way, what designs are to be avoided, lest an army of re-purposed hardware be launched against them? For instance, is SHA256 slightly weakened because brute-force is orders of magnitude less difficult now, than it would have been sans specialised BTC hardware?

If I lease a miner from a third party http://www.hashtrade.com/ am I engaging in a currency transaction?

Reply on news.ycombinator.com