Analyst: buying a high-end graphics card for mining would be "very foolish"

midian182

Posts: 9,745   +121
Staff member
A hot potato: We know that one reason it’s almost impossible to find a new graphics card right now is miners grabbing them to take advantage of high crypto prices. According to Jon Peddie Research (JPR), however, buying a high-end card for mining would be “very foolish.”

In JPR’s latest report on the PC market, it notes that “the pandemic has distorted all models and predictions as has the gold-rush in Ethereum.”

The report reveals that total GPU shipments were up more than 20 percent in the fourth quarter of 2020, jumping 12.4 percent year-over-year.

It’s important to note that the figures include integrated graphics, which is why Intel takes a 69% share of the market. AMD’s QoQ shipments increased 6.4%, giving it a 17% share, while Nvidia’s decreased 7.3%, dropping its share to 15%.

It’s a different story when looking solely at dedicated graphics cards. The market decreased 3.9% from the last quarter, with Nvidia taking an 82% share, dwarfing AMD’s 18% share. JPR says the fourth quarter is typically flat compared to Q3, so this is normal.

The pandemic is highlighted as a factor behind the demand for AIB cards across the last two quarters, but crypto mining also played a part. The market research firm has a dire warning for miners eying a powerful graphics card.

“The power consumption of AIBs greatly diminishes the payoff for crypto-mining. Ethereum, the best-suited coin for GPUs, will fork into version 2.0 very soon, making GPUs obsolete. A person would be very foolish to invest in a high-end, power-consuming AIB for crypto-mining today,” JPR writes.

The report predicts GPU shipments will grow 21 percent across the next five years, while semiconductor vendors are guiding up for the next quarter. Let's hope availability starts improving soon.

Permalink to story.

 
Disclaimer: I know very little about this whole Ethereum deal about this "fork to version 2.0" stuff but I've been seeing it for a while now. Not sure what "very soon" means: could be weeks, could be end of the year, could just not happen.

So it seems to me that the ROI for a mining GPU to be still "worth it" to miners depends entirely on EXACTLY WHEN this changes happen to measure that up against power usage costs which we know also varies a lot.

So saying is a "bad idea" is based on...What exactly? The "very soon" change that hasn't been clearly defined? The power costs that also are so variable as to render the entire statement meaningless if one can save enough on power?
 
I also like that last graphic of AMD's dedicated GPU market share from Q42019 vs Q42020: It immediately makes me remember how AMD when out of their way to mock Nvidia for their "paper launch" yet even with a product line that for all intends and purposes, finally competes with Nvidia on the high end directly, its still steadily losing market share meaning that even if Ethereum and the pandemic weren't concerns we know people want their cards and can't get em: they botched their launch far, far worst than Nvidia and the numbers show this: they're losing ground at the exact point that they should be gaining ground in market share.
 
A friend of mine built a mining rig in 2017 and continued to mine even after the value of crypto fell away at the time. He continued building up a balance and just last month he sold up and has enough money from it to be able to put a deposit on a house. So whilst it may not be immediately profitable to mine at today’s rates, you just need to wait for the value of the crypto you are holding goes back up again.

However if Eth becomes obsolete on GPUs then yes buying now to mine wouldn’t be a very good idea. Unless there’s another coin, I don’t know enough.
 
I also like that last graphic of AMD's dedicated GPU market share from Q42019 vs Q42020: It immediately makes me remember how AMD when out of their way to mock Nvidia for their "paper launch" yet even with a product line that for all intends and purposes, finally competes with Nvidia on the high end directly, its still steadily losing market share meaning that even if Ethereum and the pandemic weren't concerns we know people want their cards and can't get em: they botched their launch far, far worst than Nvidia and the numbers show this: they're losing ground at the exact point that they should be gaining ground in market share.

Quick reminder when next gen cards were launched:
RTX 3080: September 17th
RTX 3090: September 24th
RTX 3070: October 29th
RTX 3060Ti: December 2nd

RX 6800 (XT): November 18th, AIB models: November 25th
RX 6900XT: December 8th.

I think it‘s pretty reasonable to assume that due to the lower number of models and a late (in the quarter) launch AMD would lose market share. Particularly since they only launched the high(er) end cards.
 
Last edited:
I also like that last graphic of AMD's dedicated GPU market share from Q42019 vs Q42020: It immediately makes me remember how AMD when out of their way to mock Nvidia for their "paper launch" yet even with a product line that for all intends and purposes, finally competes with Nvidia on the high end directly, its still steadily losing market share meaning that even if Ethereum and the pandemic weren't concerns we know people want their cards and can't get em: they botched their launch far, far worst than Nvidia and the numbers show this: they're losing ground at the exact point that they should be gaining ground in market share.
True, but unavoidable sadly. The demand is so high now and the allocation from TSMC is limited, so even if they had the money to purchase x3 times the wafer allocation from TSMC, they simply cannot.

Both nvidia and AMD are making and selling more now than they ever did, in any new GPU generation. So they do make much more profit each, it's just that nvidia choosing Samsung foundries was smarter, because they can make much more GPUs there than anywhere else.

The only theoretical way AMD would be able to make x3 or more GPUs would be to forfeit the contract with Sony and MS for the consoles and make GPUs on those wafers, something that will never happen. Consoles and CPUs have much higher priority for AMD than their GPUs. That's the (sad) reality.
 
Looking forward 4 the crypto crash and all that tasty tears
Same. They're great opportunities to buy and offload worthless fiat. Buying has a better ROI than mining to boot, which is why I only mine casually with the GPU I already have... and buy shares in Nvidia and AMD, the folks selling the shovels.

Ethereum, the best-suited coin for GPUs, will fork into version 2.0 very soon, making GPUs obsolete.
So the news is finally catching up to what I've been saying for months. Great job

To also offer some technical corrections: this is less a fork to Eth 2.0 as the merging of an experimental fork (the proof of stake Beacon chain) into the main proof-of-work Ethereum chain. There will be other mineable coins out there but none of them are anywhere near Eth right now.
 
Last edited:
True, but unavoidable sadly. The demand is so high now and the allocation from TSMC is limited, so even if they had the money to purchase x3 times the wafer allocation from TSMC, they simply cannot.

Both nvidia and AMD are making and selling more now than they ever did, in any new GPU generation. So they do make much more profit each, it's just that nvidia choosing Samsung foundries was smarter, because they can make much more GPUs there than anywhere else.

The only theoretical way AMD would be able to make x3 or more GPUs would be to forfeit the contract with Sony and MS for the consoles and make GPUs on those wafers, something that will never happen. Consoles and CPUs have much higher priority for AMD than their GPUs. That's the (sad) reality.

True: Let's not forget that even Nvidia switched to Samsung for fabs at least on the initial runs, only thing AMD and Nvidia could have done was predict this would happen when Apple decided to leave intel behind but even then I am almost sure they did not have enough notice to suddenly ramp up their space on TSMC or seek alternatives, Nvidia did react but barely and there were not without some issues by going with Samsung.

What's really laughably bad is just the PR people from AMD, as usual decided to start throwing rocks through their glass house. The only other lesson for AMD is more engineers or at least more of their input when it comes to marketing and PR stunts.
 
AMD PC GPUs = uber marginal niche. they are 100% in the console business now

I am not sure they'll sustain much of that side of the business for anything other than supporting the tech that makes it to consoles.

However I think AMD would like to get into the compute side of things outside of mining: That is lucrative business and does has a decent future.

Mind you I am not saying that they will, or that even if they try that they will succeed in taking that business away from Nvidia, but merely that trying is not a terrible idea: even intel wants in on that angle after all.
 
Sales dropped in a year of the most hyped upgrades? Isn't it time to stop blaming "mining" and consider that the problem is purely manufacturing issues that Nvidia and AMD are glossing over with twitter pics of random mining farms?
(edit for grammar)
 
Last edited:
LOL Ethereum is just one coin of the many that are profitably... As for the 2.0 version of it might take about a year or so.... Mining will still be around for a bit longer boys. Happy with my 6 GPU 3070 mining rig bringing in $1k-1200k a month and only costing me $40 in power.
 
Last edited:
Isn't it time to stop blaming "mining" and consider that the problem is purely manufacturing issues that Nvidia and AMD are glossing over with twitter pics of random mining farms?
(edit for grammar)

How can a company who makes video cards for gamers and creatives, which is normally one card per person, keep up with demand for a third market, miners, that buy between 2-20 cards per person?

Manufacturing capacity cost money and there was no reason for NVIDIA or AMD to buy more capacity than they did at the time they signed the contracts for it. Also there is no way to know if the capacity they bought is what they are getting during a pandemic. I don't see how you can purposefully ignore mining and put all of the blame on the manufacturers especially during a pandemic that brought more people inside and online.
 
How can a company who makes video cards for gamers and creatives, which is normally one card per person, keep up with demand for a third market, miners, that buy between 2-20 cards per person?

Manufacturing capacity cost money and there was no reason for NVIDIA or AMD to buy more capacity than they did at the time they signed the contracts for it. Also there is no way to know if the capacity they bought is what they are getting during a pandemic. I don't see how you can purposefully ignore mining and put all of the blame on the manufacturers especially during a pandemic that brought more people inside and online.
You're ignoring point I made to deny what looks like the biggest lie in tech in a long time. Sales have dropped, not sky rocketed.
 
Back