Wouldn't it be profitable to make an ASIC capable of calculating private keys?

Just looking at vanitygen-plus (currently supporting almost one-hundred crypto-currencies) do its stuff, and it is capable of over 548,990 keys/sec on my CPU alone.

Well developed ASIC is several orders of magnitude faster than my CPU, wouldn't it be profitable for a company with the resources to do so to develop such a key miner, even if it is a one-off?

The reward (actually, theft) is access to the private keys of any address. Surely the development costs are not so prohibitive for ASIC as they are for general-purpose CPU's?

Future developments in this area seem like the most realistic threat to crypto-currency security so far.

Willtech

Posted 2018-03-04T10:26:08.650

Reputation: 2 657

Multiple organisations have tried that. I think, there is still 1 active "brute-forcing-pool (forgot the name)", but mining transactions is always worth it (over brute force) unless you win the lottery and find one of Satoshi's keys. I think, the brute forcing pool got 7 private keys of used addresses with 3 of them having Bitcoins on it (but less than 1 Bitcoin all in all). (but they wasted a lot of energy to do that)... – ndsvw – 2018-03-04T12:18:59.687

https://lbc.cryptoguru.org/about - there are positive and negative comments on the project. On the onehand side it proofs, that the math is incredibly “big”, and a good use for the bitcoin setup. On the otherhand side there are comments, that the found addresses are not really “found”, but merely “known” to be found. There was a longer thread in bitcointalk forums last year... Hint: ASICs do sha256, whereas keys is ECDSA logic. Haven’t seen ASICs yet to do ECDSA math. – pebwindkraft – 2018-03-04T20:48:01.160

@pebwindkraft True, neither have I. I am aware that vanitygen-plus uses Open-SSL for the EC math and figured that any process that does 'do this, this, this and then this' every time might be ASIC-able. ASIC performs any one function it is designed to do, it is only the Bitcoin mining ASIC that do sha256 but there are I, I make the presumption, plenty of varieties in industry purposes. – Willtech – 2018-03-05T09:07:52.247

If you could steal everyone's bitcoins, no one would want bitcoins anymore; the market and the value of the coins would fall to zero. You would have successfully stolen coins that no longer have any value. (Alternatively, you could try to steal a few coins at a time in a sneaky, undetectable way. But then your payoff becomes much slower, and also not worth the effort) – abelenky – 2018-03-08T19:56:10.803

Answers

You could in theory build an ASIC to brute force private keys of any cryptographic scheme, like EC or RSA.

But ASICs only give linear improvements. That means that they 'accelerate' some search only by a constant factor, like 2, 3 or 100. Even if your ASIC does something a million times faster than a normal CPU, it's still a linear improvement.

If guessing a private key takes the current age of the universe in a CPU, an ASIC a million times faster would still take 13 thousand years.

Cryptography relies on problems being exponential on the key size. Exponential growth beats linear growth. If some day a powerful ASIC or CPU could guess a ECDSA private key, users of ECDSA would only have to increase their key sizes.

To yield a better than linear improvement, you would need a better algorithm. Implementing current day algorithms in faster hardware only gives linear improvements. It's widely believed, even if we still don't have formal proof, that it's not possible to get better algorithms for prime factorization and elliptic-curve factorization. (At least in non-quantum computers.) In other words, they are considered intractable problems.

Osias Jota

Posted 2018-03-04T10:26:08.650

Reputation: 769

ASIC does SHA256 hashes continuously. Getting Private key from Public key requires Elliptic Curve Multiplication in reverse. Elliptic Curve Multiplication is assumed to be very hard problem. i.e. By having point on the elliptic curve , You can add or multiply it as many times as you want but by having a point which has been multilpicated, you won't know its generator point. That is the hardness You have to solve in order to get that point. So no machine in this world can efficiently compute Private key by having Public Key.

DOLLY PATWA

Posted 2018-03-04T10:26:08.650

Reputation: 505

Doesn't an ASIC do whatever it is designed to do very fast? sha256 ASIC does sha256 very fast, but what about other ASIC? ASIC is not limited to only one task but a specific ASIC can only ever do one task. – Willtech – 2018-03-05T09:19:56.493

Solving a problem faster doesn't mean magically faster, see my answer above – Osias Jota – 2018-03-08T15:54:23.433