Well, the history of actual 50% attacks on cryptocurrencies in the past few years seems to suggest that effect of a 50% attack on the value of the cryptocurrency is... pretty close to nil, actually. Admittedly, a successful attack on a major cryptocurrency like Bitcoin or Etherium is likely to have far different impact on cryptocurrency society than one on a minor cryptocurrency, but the ability of the community to twist every bad news into "this is good for Bitcoin" should not be underestimated.

A 51% attack can only reverse a transaction anyway, it can’t arbitrarily take money from someone you haven’t transacted with. But they could get short exposure to btc/eth in the market, profit when it went down from the ensuing panic, and then sell the graphics cards to retail (there are other complications that make this unrealistic, of course.)

I suppose any of the Faang companies could sacrifice a data center for a day or two to do the same. Just crash the currency and get out, to eliminate proof of work currencies.

Again, why would they? They only stand to lose on it, so why do it in the first place?

NVIDIA would reduce demand, but for the FAANG companies it's a neutral proposition, aside from the cost of doing the work to crash the currency.

And there's plenty of reasons to want to do so: Proof-of-work currencies are what make ransomware payments possible, and there's a large daily CO2 cost for proof of work currencies. The question is simply when is the cost of the externalities greater than the cost of acting?

> One of the first ransomware attacks ever documented was the AIDS trojan (PC Cyborg Virus) that was released via floppy disk in 1989. Victims needed to send $189 to a P.O. box in Panama to restore access to their systems[0]...

[0]: https://www.crowdstrike.com/cybersecurity-101/ransomware/his...

Am I missing something? Why would NVIDIA want to reduce demand for their graphic cards?

There is actually a good case to make here. Nvidia currently is in the tough spot of a) not having the capacity to saturate the market and b) not being able to expand the production to match. B is because they don’t have enough confidence that the crypto people will be there as costumers in 5 years. And A is a problem, because their traditional user base (ie gamers) won’t match the spending, but will still get their fix (just not from Nvidia). And that breaks the strangle hold that they had a couple of years ago on that market.

So short term: having the high prices paid by crypto people is nice. But it destroys their relationship to their old customer base. Depending on where the crypto market will be in 5, 10, 15 years, this can be the 5x of their market cap or the 0.2x

Relying on single customer can have quite severe negative consequences (walmart crushing whole businesses is a good example). GPU shortage is destroying PC position as a viable gaming platform. Give it couple more years of GPU shortages and Nvidia will irrevocably lose whole market segment, leaving it with compute and shaky scam riddled crypto.

Yes, nvidia and AMD should collude to destroy crypto from a market risk reduction perspective.

They have done things like come up with a mining only SKU, apparently partially because they are annoyed that people aren't using their cards for their intended purpose. Probably, while a sale is a sale, they see the crypto market as flakier and would rather keep their gaming reputation intact.

Can such an attack be done without detection?

No, the whole point of using a blockchain is to have a publicly verifiable ledger of all transactions. If anything goes wrong, there will be so many alarm bells ringing across the world that probably it'll be a hearable event for most people living in places that resembles modern civilization.

You can sell your cryptocurrency for real dollars before openly performing your 51% attack.

The idea that you could buy a real-world item worth half as much as the amount of ETH/BTC you would mine in an hour or two or 51% is simply unrealistic.

ETH mining at 500TH (a solid chunk over 51% just for ease) would net you 300 ETH an hour...that's 1.1 million USD, which means you would have to be able to purchase millions of dollars worth of real world items that would be released to you within the time-frame of your attack (in this case an hour) to make this even approaching "worth it" from a risk/reward viewpoint and even that calculation is ignoring just how profitable they'd be just mining normally for the foreseeable future.

No one is waiting for just an hour worth of confirmations for millions of dollars of real world goods exchanged for ETH.

Wouldnt 51% attack allow you to rewrite contracts?

no, contracts are immutable, I'm unaware of any base-layer chain attacks that can affect contracts negatively (outside of contract level logic that may be vulnerable itself)