Nvidia RTX 5080 16GB Specs Leak | 5090 & 5060 Price Analysis | Nintendo Switch 2 Update

[@MooresLawIsDead]

[SPON: Support MLID by checking out KTC’s OLED 240Hz Monitor: bit.ly/3P16MEB ] #KTC #OLEDmonitor #G27P6S #gamingmonitor

[@InternetListener]

10:08 those numbers are wrong or your explanation of 60% of margin is incorrect. $2175 if 4090 cost was $640 hence 60% of margin. 1/0.4 not equal to x1.6 Or margin is 1-1/1.6=37.5%

[@MrNPC]

$2k or less is tempting but anything above makes me want to wait for the 6000 series when AMD might compete again.

[@jlrockafella]

Do you expect the 8800xt to have similar or more performance than 7900xt ?

[@TheWildBois]

@@jlrockafella Yes he does and he believes 4070 Ti RT performance

[@freddylippstick674]

I disagree with your predictions about the performance gains for the new 50xx series ! If you look up the generational lifst from nvidia cards starting from the rtx series, i have noticed every generation has like an 10 - 15% peformance uplift AT BEST 🤨 .. THATS ☝😐a lightyear away from the 30 - 50% gainz you are predicting !!

[@ktvx.94]

We will charge $2500 and you will love it because the more you buy the more you save

[@quintonNofziger]

Bruh they could make it 5 grand and idiots like you would still pay it.

[@RussianAverageGamer]

Honestly, i bought 4090 for about 1,2k and sold few weeks ago for almost 2k, so it kinda makes sense, 4080s and below are all same value or cheaper

[@RickTrajan]

The more you pay the more we save

[@davidlandrum]

@@ktvx.94 Nvidia isn’t wrong. You will have top tier gaming with no compromises for the next two generations at least.

[@Zemla]

What a argument :O i need 2 cards now!!! minimum..

[@vgamedude9811]

remember when tech got faster AND cheaper, and "its 20 percent faster than our old card at 15 percent higher price" wasnt a selling point?

[@gucky4717]

If it is 50% faster but only 20% more expensive, it is faster and cheaper $/perf, there is the catch...

[@DrewDragoon]

Mongoloids will complain in millions of KZbin comments and on Reddit and 4chan and wherever else, then they'll go out and buy the dang thing anyway. That's how we got to this point. Nobody cares about the words people write online, only actions matter and people can't stop themselves from consooming.

[@XiaOmegaX]

@@gucky4717 Yeah but we used to get like, 80% faster *and* 20% cheaper.

[@e2rqey]

It still does. Its called buying used or previous gen technology. Or making sure your income keeps pace with value or your currency. It used to be far easier and vastly less expensive to do the process node shrinks that enabled that kind of scaling in performance at a consistent rate. While we can still get close to the performance gains, the financial cost of doing it has increased massively. So of course the price of those products will increase, and that's before factoring in any wider macroeconomic trends

[@defeqel6537]

We are hitting the physical limits of silicon, so expect prices to go up in the future too. 2nm is the point where progress will essentially halt.

[@Violet-ui]

Imagine the 5060 having less VRAM than the 3060 Makes no sense. THe 4060 already made no sense, but 5060 would be laughable

[@adriankoch964]

Wait for them trying to sell it with some 5000 series driver locked neural texture compression/enhancement that let's it save 50% VRAM in about as many games as dlss 1 & raytracing in 2000 series' first year on market.

[@De2t3ny]

there is new ram on that thing so why doesnt it makes sense? also this might be shocking to you but.... just dont buy it? why are you idiots all so caught up about this lmao

[@godoflight558]

3060 was supposed to have 6gb ram, but nvidia last minute changed it to 12gb, which is why it had more ram than the 3070's 8gb

[@N4CR]

5080 has insufficient VRAM for 4k already, 22gb in modded titles, RIP gullible novideo owners

[@HanSolo__]

"We will NEVER make "3060 12GB" mistake again" -Nvidia

[@zrpggamer]

5070 with the performance of the 4070 TI for the same price, ridiculous.

[@pedro.alcatra]

Don't worry. Someone will buy it

[@christophermullins7163]

​@@pedro.alcatra**everyone

[@JBrinx18]

Hanging on to my 6950xt for another 3 years

[@ws1-p9u]

​@@pedro.alcatraMany more than [some]one.

[@endlesslogins156]

No you misunderstand, Jensen says its very generous. ha. jk

[@blackknight50277621]

80 class being cut HALF of 90 class is GREED manifests

[@haukionkannel]

No problem! 5090 will be double the price this time! 😂

[@Real_MisterSir]

making room for the Ti (or super) version with 24gb down the line after 3gb gddr7 is here

[@Frozoken]

I remember when the 03 die was first made with ampere to actually give proportionally MORE cores and and bus width. Now it's somehow become much less than the old 80 class which was literally using 04 dies shared with the 70 class BG203 should have 120sm and a 320bit bus with 20gb vram

[@StreetComp]

Yeah only 10700 Cuda cores for 5080 is crazy. The 3080 has approx 8900, 4080 has 9700, and now 5080 will have 10700 - meanwhile the xx90 has gone from 10700 to 16700 to 21700! Seems so wrong

[@hochhaul]

​@@StreetCompIt will only continue to get worse in the years to come.

[@AndreB23]

$1500 for a card that has 16gb is insane.

[@50H3i1]

3090 and 7900XTX can do much better in professional work like editing videos because the Vram

[@mikelay5360]

Who said 1500😮? I guess we'll find out next year.

[@atomic3325]

@@50H3i1 High end professionals would not use any of these cards. They would have a Quadro which is purpose built for professionals with even more VRAM. Its only gamers moonlighting as professionals that would get a 90 series. Of course those cards are even more eye watering in price, but for professionals its a business expense. Possibly also pros trying to save money I suppose.

[@SolarianStrike]

I mean the 4080 was $1200, so $1500 is about what you expect from nVidia. They know it will sell anyway.

[@mikelay5360]

@@SolarianStrike we have less than 2 weeks. We'll know.

[@jkteddy77]

"Yea, 5070 is $599, which is $100 more than the 8800XT and has 4 less gb, BUT it only loses by 5% in raster, is more efficient, and wins in Raytracing" I can see it now.... the Green tax.

[@Healthy_Toki]

8800XT is probably not going to be $500.

[@fluffyRabbit01]

​@@Healthy_Tokiya, i simply don't see it. even if tarrifs don't happen, it'll still be much more than $500 if the leaks about the performance are true

[@flimermithrandir]

Exactly what i said. The 8800XT cant be 600. It would be dead at that Price if the RTX5070 launches and sells for the same Price. RX 8800 XT needs to be 500 and even then it will hardly sell anyways.

[@haukionkannel]

5070 will sell tens of times more than 8800XT with those specks as allways!

[@haukionkannel]

@@Healthy_Toki $600 to $700 is more likely.

[@alilokhd4638]

Nvidia 2000$ gpu Apple 1600$ phone/ 2000$ mac Sony 750$ console Microsoft 1000$ surface pro Google pixel 1000$ Hp, dell ....... Basically, everyone is raising the prices.

[@zerocal76]

Yup. I say this All the time. I do not buy new tech anymore. So expensive it has to be open-box or used!

[@takerulol]

And then they say "the price increase is due to inflation". Inflation is literally the increase in price and therefore devaluation of the currency/income. Since they fuel on the gigantic technology market, they play a big part in the reason FOR the inflations, not the other way around.

[@10percent4DaBigGuy]

nobody is raising prices the value of paper currency is only declining....

[@rareapuenjoyer]

That's what happens when we print money for decades and demand also increases across the hole world and some of these companies have near monopolies in their respective categories

[@4evahodlingdoge226]

To be fair there are really good deals to be had currently, they were selling 4070 prebuilds for $1k during black friday and a lot of 4000 series laptops are on good dales currently and are better to buy than waiting for next gen unless you only are concerned with having the best performance no matter the value.

[@adityapatel1839]

NVIDIA seems to be getting cocky, but we all saw what happened to Intel when they got overconfident. Hopefully, Intel and AMD can start making big moves soon.

[@cajampa]

They have good reasons because they actually provide massive performance advantages every generation to their enterprise clients. They are planning to build new nuclear powerplants to run future data centers. And the new OpenAI o3 model cost $4k!! To run per task.

[@adityapatel1839]

@@cajampa To be fair though if AI fails on return on investment, Nvidia will sure stop selling shovels. Even intel were/are dominant in server and enterprise but slowly they are losing. Hill looks quite steep but last I heard, Apple, Arm, Microsoft, Google and Tesla and bunch others are joining rivalry against Nvidia for accelerated gpu. Sooner then later the shovels wont be as shiny.

[@confusedcounsel5285]

@@adityapatel1839 AMD and Sony joined forces in Project Amethyst to confront Nvidia. Tesla has tons of research in AI so does Microsoft. Google... they get easily distracted and pivot.

[@damianich4824]

​@@adityapatel1839 intel are losing because they fell asleep at the steering wheel, their designs suck, and their board has been mismanaging the company for 20 years.

[@Anankin12]

I don't think so. Nvidia are getting complacent in their handling of consumer products, not in their R&D nor in their architecture design. If tomorrow AMD or Intel come out with something competitive, Nvidia can just lower prices and/or provide more memory and they're done. Intel had basically atopped proper development. Which is why they still had high costs and no performance gain to show, so when Ryzen showed up they were backed into a corner. The situation is entirely different.

[@ScooterBean]

Just remember. You, set the price. They can set the msrp, list it at whatever they want but if YOU don’t buy it. And it doesn’t fly off the shelfs. Then we might have a chance at getting these consumer gpu prices down. I know this won’t happen. But it’s a good dream.

[@slothywun]

If people only understood that they could lower the prices by exercising some self control and not buying. But the masses are just too stupid and impulsive for that to ever happen.

[@zerocal76]

It has already happened friend, just not much. Remember the original 4080 did not sell well at 1,200 and a 4080 S was released for 200 less. I think what's stopping prices from truly falling is PC gaming has become a lot more popular. For ex, you'd never see pre-built rigs at Best Buy before. Last month I saw like 5 on display. So newcomers think these expensive AF cards are pretty normal 🤦🏻‍♂️

[@zerocal76]

@@slothywun U cant say new PC hobbyist are stupid. They'd have to research how much value the 1080Ti or even the 1060 for ex provided. Ignorant for sure but Nvidia is taking of advantage of ppl wanting the newest tech. Those older cards are fine for anything but the latest AAA!

[@slothywun]

@zerocal76 New PC hobbyists don’t make up that large of a percentage of early Nvidia adopters. Its the diehards that cream themselves every time there’s a new launch and would rather starve for a few months than risk their friends having a newer GPU than them.

[@ScooterBean]

@ that and it seems with any hobby now a days, people jump in neck deep right away. Someone wants to get into wood working? They think they need to go out and buy all top of the line tools next weekend. Want to get into drones, proceeds to spend thousands on a fpv headset gear, etc. And that’s what has really messed up the pc market. All people want is the best of the best, no matter the debt.

[@alangeorge5592]

Dont ask questions, just consoom nvidia products!

[@iulian2548]

That just soomarizes it perfectly!😊

[@alangeorge5592]

@iulian2548 I couldn't have poot it any better!

[@zaxmaxlax]

the more you buy the more you save

[@Keyaerts]

Lets be honest here, the bulk of gamers will spend no more then $500-600 on a graphics card... For me that is high end (RX 7800XT owner). If Nvidia prices the 5070 above $550 with 12GB of ram I hope lots of gamers will say no thanks.

[@andersjjensen]

But you know they won't. Because Nvidia is going to have this shiny new proprietary software feature. So every generation you have to buy a graphics card AND a software license. And since software can't be endlessly duplicated at no cost, that's gonna add up you know....

[@xanderx51]

The market over the last ten years disagrees with you.

[@bigpoppa1234]

Most new card sales are 4060 level. Even 4070's are premium in the eyes of the market.

[@xanderx51]

@@bigpoppa1234 and how much are those cards? How much would they have been ten years ago for the same specs? Prices have gone up in general, trust me I'm not happy about it. I got a 4090, sadly I waited figuring the market would do what it had done in the past and at least slightly react to the newer 50s coming out, paid about at minimum $200 more than what would of been normal. But Nvidia seems to know folks will buy plus they getting that AI money. 30s and 40s should be at least 10 to 15% cheaper and now we got 50s with worse specs for more money. And lower amounts of 40s floating around in general, less stock drives prices even higher.

[@DonJohn-h8l]

I think the stars will align and the 50 series will flop so hard Jenson will wear jean jackets from then on.

[@bjones1794]

Haven't even watched the video yet, but Nvidia still giving the 5xxx series 80 class only 16gb of VRAM when even AMD's 7900xt had 20gb VRAM is hilarious. It will obviously sell insanely well no matter what it's price is, but I still find it stupid to give that much performance while limiting potential longevity with only 16gb VRAM.

[@The_Word_Is_The_Way]

Limiting longevity is the point. They want to prep you as a customer to come back and buy a new card in a couple years, and in their eyes, much sooner rather than later.

[@davidsmith4186]

Someone who's serious about hardware wouldn't even consider that card. 5090 or nothing. Nvidia is done with brokies. 🤷‍♂️

[@kapler8550]

Tbh if the 5080 launches at the leaked prices, i see another 4080 repeat. Where it sells poorly and gets a super refresh later. However i also think that its 100% intentional. Make the 80 unattractive and “force” an upsell to the 90. I can also see nvidia doing this with the 5060/ti. Give the 60 8gb and the 60ti 16gb to upsell. And then later release a 12gb 5060 super.

[@HeliumFreak]

Planned obsolescence is literally the corner stone of most manufacturers. If you make something that lasts for ages then you don't sell as many

[@liberteus]

​@@The_Word_Is_The_Way exactly. I have a 3080 10gb and it's been showing its limits for a while already, and when it's not vram starved it's still performing well, what brings it down is maxing vram then every goes to sh*t. My card can only play 1440 and 1080p when I wanted to game ultrawide. Next time I'll get the 90 class... only ones that will stand on their own for 5years and more. a 3090 with 24gbs is better than a 4070S with 12, even though it may be a bit slower.

[@TheRisingMiles]

Thanks for all the news this year!

[@godnamedtay]

🥴

[@muiwols6709]

Ngreedia made b580 look even better

[@endlesslogins156]

Jensen's Generosity: The story of the 50 series

[@lolmao500]

I secretly hope that some Luigi copycat pays a visit to Jensen

[@seacowDUGONG]

Maaaariooo 😅😮​@@lolmao500

[@Moonbogg]

The Arc B580 is more exciting than all those $1000+ cards.

[@EmuQuest]

RX580 spiritual successor, except B580 is even cheaper than RX580 at launch considering inflation

[@haukionkannel]

B580 is sold lost or no profit. NVIDIA at least 60% profit margins… so…

[@SweatyFeetGirl]

@@haukionkannel 4090 costs about 500$ to make, sells for 2k, thats way more than 60%

[@EmuQuest]

@@haukionkannel is there any evidence to what you're suggesting?

[@haukionkannel]

@@EmuQuest The Gamer Nexus Intel presentative interview! He did say that they are not making money with Battle mage, but hope that they will do some with celestia or other future GPU…

[@adeptuspotatocus6451]

Anyone else remember when DRAM dies doubled in capacity almost as fast as logic node shrinks? Well, that's gone out the window and we're pushing 5 years from the time 2GB/16Gb chips came out til 3GB/24Gb chips will be available. It's not just Nvidia being greedy, even though they certainly are.

[@gorky_vk]

Memory controller area of die is still shrinking just fine. So no, it' Nvidia being greedy.

[@Anankin12]

An old A100 is half as fast as a 3060. It has 32 GB. 3060 is 12 GB A 12 GB 2060 exists. People have made 16 GB 3070s and 3080s. 4060 16GB exists. 7600XT, 7800XT, 7900GRE, 7900XT, 7900XTX exist. I really don't see how it's not Nvidia's greed here.

[@adeptuspotatocus6451]

@@gorky_vk They certainly are greedy. I've said it before and I'll say it again. Nvidia shifted their tiers. AD103 replaced GA104, AD104 replaced GA106, and AD106 replaced GA107.

[@richardyao9012]

@@gorky_vk The memory controller area of the die does not actually shrink anymore as IO logic scaling died a while ago (maybe around 14nm?). This is what motivated AMD to adopt chiplets.

[@gorky_vk]

​@@richardyao9012 Except you can't see that on any of the latest generation, die area reserved for memory channels are still almost the same for the last 3-4 generation. But you're free to believe that corporation are your friends.

[@marcusaureliusanonymous]

What I have heard from hardware unbox and others is that rtx 5080 is gonna be cut down so much that it will be closer to a 5070 series in terms of performance than rtx 5090 series card. With that in perspective, prices are truly insane for something like a 16gbvram card.

[@marcusaureliusanonymous]

@@Daniel_Martin777 not sure about raw hardware performance gain as lately Nvidia's marketing team is counting more on performance gains on software side to justify these insane prices... Let's hope that Intel and AMD can make a dent in Nvidia's software moat in coming years... For AMD, RDNA4 seems to be a missed opportunity...I really hope AMD get its act together in UDNA. For Intel seems to have everything to challenge Nvidia and Intel did release good cards but I think with Pat Gelsinger gone... shareholders will again install a finance guy as CEO to lead Intel and destroy it from within....

[@marcusaureliusanonymous]

@@Daniel_Martin777 i'm also waiting for a good 16 gb card for AI workloads not for gaming... Currently i'm on ( 2060 super 8 GB, ryzen 7 2700x, Asrock fatal1ty b450 gaming k4).. i'm gonna keep my PC but could go for an upgrade to 5070 ti or 5060 ti... depending on whichever give the most value....

[@Anankin12]

​​@@marcusaureliusanonymousjust get a 3060 12 GB ma, or if you can make it work get a B580

[@marcusaureliusanonymous]

@@Anankin12 not sure 3060 12 GB is the upgrade from 2060 super. i do have an option 4070 super 12 gb 675 dollars... but i'm looking for a 16gb card at the very least for the AI workloads

[@haukionkannel]

Well 5090 will be double times more expensive than 5080 so NVIDIA will price 5090 right this time!

[@e2rqey]

Say it with me!!! THE MORE YOU *BUY* THE MORE YOU _SAVE_

[@MooresLawIsDead]

Lol

[@timg2655]

The more you buy 🤖 the more you save 🧟‍♂️

[@richardyao9012]

The more you buy. The more you save. :P

[@asdfjklo234]

The more you buy, the more you save! Jensen is our savior and my wallet's menace! Take my money, Jensen, who needs money anyway!!

[@kozad86]

"Value increase" is a line of crap. Historically, prices don't increase with each new CPU/GPU simply because it's faster than what it's replacing, this is exclusively an Ngreedia move because they are *desperate* to maintain wild COVID-19 era margins. If prices did increase to match performance increases, an Intel desktop CPU would be tens-of-thousands of dollars.

[@anitaremenarova6662]

This guy drank the koolaid, what insane bs to justify 400-500 dollar ENTRY LEVEL graphics card. Whatever, Intel has us covered with 250 bucks and 12GB.

[@justhomas83]

Forcing gamers toward RTX 5090 is terrible madness. Not everyone wants to spend 2400 plus for a graphic card

[@danh2716]

They don't have to. There's plenty of lessor options, including not buying anything at all. Nobody is being "forced" to do anything. If a guy decides to spend $2k+ for a 5090, then he is doing it because he feels the 5090 is worth more than that amount of money.

[@defeqel6537]

there is always the option of not buying

[@Verociity]

used 40xx are suddenly very cheap which are still great cards, now the prices are actually reasonable, unlike the 50xx.

[@justhomas83]

I understand people have options my point is Nvidia does not change it's marketing strategy. By pulling the other tiers away by a high graphics systems gap obviously the Intent is to sway more people towards the RTX 5090. Performance is a chase got it doesn't mean it has to cost the price of a used Honda Civic.

[@haukionkannel]

Buy 5060 instead! Problem solved! 😂😂😂

[@pdxholmes]

At some point 60% profit margins are going to bite Nvidia. We're entering an economy reality where Wall Street might be making money hand over fist, but Main Street has less buying power than it's had in 40 years.

[@davidlandrum]

This is why Nvidia sells to corporations instead.

[@haukionkannel]

So far that has not bite! 4000 series did see Nvidias market share to go up from 86% to 88%. I expect that now they go from 88% to 90%!

[@flimermithrandir]

Nvidia doesnt make only 60% anymore either i think. That was 10 Years ago probably lol. The RTX 4090 sells for 1600 Dollars, the RTX 4080 for 1200 (MSRP at first). The RTX 4090 has an Die Size of 609mm², the RTX 4080 379mm² (38% Cutdown). If you do the Math and you take 1600-38% you get 990 Dollars. And that was the RTX 4080 vs the 4090 which needed less Components to make to Begin with anyways. Dont expect that those Cards lower than that do any worse for the Company. They do even better in Favor for Nvidia. They probably make more like 60% for the RTX 4090 and more for the other Cards. The RXT 4080 Super used again less Components and yet it sold for 1000 Dollars... all of a sudden. (Thats just me but i will also assume they will die sooner but who knows) No. Sorry, i cant see them only making 60% in 2024. They did not "give away" RTX 4080 Super all of a Sudden. They still made a huge amount of Money for sure with those. Yes i have only my Feeling here but just plainly looking at the FACTS we have it just doesnt add up to make Sense.

[@pctechloonie]

@@flimermithrandir that's not how the gross margin works. oh it's easily +60% GM across nearly the entire stack. You would be surprised at how small the costs are.

[@patrickptmonk8673]

You could buy a rtx 5090... or a used Honda Civic. Damn that's good "value"😂😂

[@kapler8550]

Im taking the civic

[@rkwjunior2298]

​@@kapler8550a used Civic with no engine...derp

[@OrlandoCaba]

lmao

[@kapler8550]

@@rkwjunior2298 maybe where you’re from :p

[@davidlandrum]

A Civic for $2K is not value. 🤣

[@jkteddy77]

Radeon VII launched with 16gb of very spendy HBM2 memory for $699. ... over 5 years ago... Now 16gb for $1200+? Nvidia is valuing memory like Apple

[@lharsay]

it was vaporware not a real product

[@haukionkannel]

And being very succesfull just like Apple!

[@jkteddy77]

@@haukionkannel sure, no doubt, but consumers raked over the coals for enough memory, they're witholding it on purpose so buyers move up their chain

[@richardyao9012]

Being new on the market, GDRR7 likely costs significantly more than GDDR6. It is so new that there is no pricing at dramexchange yet. Also, the HBM2 memory for the Radeon VII likely cost even more than GDDR7, which really highlights the price disparity. Presumably, the higher cost of the 5090 is due to the giant die, the high cost process node and Nvidia’s huge profit margins. Being an Nvidia shareholder takes the sting out of the high prices somewhat.

[@uss-dh7909]

And just like with Apple products, you /will/ buy them for an absurd amount, and you /will/ be happy.

[@mrdarkguy2]

Just FYI margin and markup are not the same thing. 60% margin does not mean 1.6 which is a markup. The markup for 60% margin is 2.5.

[@WaterZer0]

40*1.5=60 40+60=100 Isn't it 150% increase?

[@kevinerbs2778]

@@WaterZer0 when you added 100% it's already double price so 2x is correct, the 0.5 is added to that.

[@mrdarkguy2]

@@WaterZer0 Margin is in relation to the price and markup is in relation to the cost. So 60% margin means that 40% of the price is the costs and 60% is the gross profit. The formula you use to convert the margin to markup is 1/(1-x) where x is the margin. So 1/0.4 is 2.5 or 150% markup ((2.5-1)*100).

[@astrea555]

Why the hell are those practices allowed? There's still 3060s on the market for a good freaking reason. Who seriously buy those at this point? I hope they actually rot unsold forever, same for the 4060s.

[@darkkingastos4369]

Internet gaming cafes in Korea will buy about 400k of them

[@ronchum5178]

Exactly. They're too arrogant to just drop the prices and let people with anemic cards upgrade. Although I think AMD has been doing the same thing to a lesser extent with the 6600 and 7600. The 6600 has been hovering around $190 for a year and a half. They could drop it to $150-$160 to let it sell through, then let the 7600 replace it's spot at around $200.

[@zuckdaddy1596]

@@ronchum5178the arrogance lies with gamers thinking that Nvidia needs their money. they're rolling in money from AI, datacenter, and professional. gaming is a drop in the bucket

[@kenshirogenjuro873]

It’s disgusting but 4060s continue to fly off the shelves. It occurs to me we’ve hardly had any revisits for GPU reviews on most channels most of this gen and I think the general market has simply forgotten how awful the 4060 actually is.

[@kevinerbs2778]

Unfortunately, the RTX 4060 is one the biggest volumes of RTX cards on the steam survey that has increased recently. Consumers are dumb.

[@lindsaybruce1396]

Ten years ago nVidia's net profit margin was 12.8% and their market cap was $10.1bn. Today their net profit margin is 56% - FIFTY SIX! - and their market cap is $3.3 TRILLION. Do you really think they are giving consumers value for money?

[@askalds]

Im in scandinavia and to be honest I wouldn't be surprised to see the 5090 selling for close to $3k and the 5080 around 2k

[@zenon3021]

Americans don't realize how good they have it. Anywhere outside the US the card will be an extra +$1,000

[@KamWysz]

@@zenon3021 UK is very good prices too unlike the rest of europe

[@JakeySurani]

@@KamWysz in the UK 4090s new are £2400 which is €2892. Explain to me how €3k for a 2+ year old card is good

[@themarketgardener]

@JakeySuraniCurrent pricing isn’t accurate because stock is too low. Thank Nvidia for stopping 4090 production

[@KamWysz]

@JakeySurani they clearly aren't selling for £2400 in the UK, on pcpartpicker the lowest is £1850 for a msi triple fan. Thye are overpriced cause Nvidia is not making them anymore.

[@406Steven]

I'm glad GPU prices haven't scaled according to performance historically. The 4090 would be over $1.4 million compared to the Voodoo 1, adjusted for inflation.

[@InternetListener]

@@406Steven PSUs were free, came with the case. $thousands or $20 is infinite increase... Mechanical keyboard without ghosting keys from free to some tenths... GPUs are not the problem as they deliver (as Apple products)... Lack of software optimization or support and cost of features or upgrades as subscriptions are truly expensive... Lack of improvement on CPUs etc in last 20 years or other components is appalling...

[@406Steven]

@@InternetListener Gotta miss Antec, you could get a nice case with a quality PSU for $80. Subscriptions are a toxin on every market they're introduced into but they make a ton more money than just buying something and owning it and people fall for it because of the "I can make payments on it" mentality. Before you know it actually owning anything will be a thing of the past and you'll be forking over every last cent you make to make payments on everyday items.

[@pieflies]

Pretty much nothing scales like that over time. It’s a nonsense calculation.

[@406Steven]

@pieflies People who defend the increase in GPU prices always say the extra performance means you're still getting more for your dollar. I'm pointing out how flawed that argument is.

[@damianich4824]

​@@406Steven the missing half of the argument is that Moore's law was making the historical trend possible in the past. In recent years, well, refer to the name of this channel

[@xpeterson]

The real reason that the 5090 will be $2k+: People couldn’t wait to gobble up the price jump of the 4090. Like, what motivation is there for them to try a reasonably priced flagship GPU when people have proven they will follow whenever the line is pushed?

[@DonJohn-h8l]

At the time, 4090 was the only "good" GPU for 4k high frames. Improvements in drivers and especially in FSR and XESS has changed that. Hopefully the 5090 flops hard.

[@Anankin12]

​@@DonJohn-h8lthey won't, it is still the cheapest way of getting 24 or 32 GB of memory in an Nvidia card.

[@davidlt]

@@DonJohn-h8l Unlikely. That's because 80-90 class gaming GPUs overlap with professional offerings these days. Folks working on HPC, ML/AL, number crunching workloads, etc. will get those. Buying 80-90 class cards for professional work is way cheaper if memory is not a problem for your workload.

[@arclyte1859]

The industry is coming up to an inflection point where mid or low tier GPUs are going to be powerful enough to get the most out of modern games and NVIDIA seems to be purposefully gimping those models to delay the inevitable. If they released a 5060 with 16 GB it would probably be the most popular card for years. Similarly, right now, anyone who owns a 4090 has all the horsepower they need - even at 4K. If I need to drop one or two settings from Ultra to high to save $2K+, that’s a trade off I can live with. There is no reason for the 5090 to have 32 GB for gaming - it just allows them to claim it’s a prosumer card. They should have just called it a Titian and been done with it.

[@demischi3018]

4090 is the Price/Performance King for LLM/GenAI and the 5090 will be the new king. Because of this, NVIDIA knows it can get away with any price increase that is still below their Workstation cards.

[@Neonvarun]

What about AMD and Arc B770🥲My only hope left.

[@blueeyednick]

there is no hope

[@moldyshishkabob]

Your only hope is to be content with what you have and play older games. The backlog calls

[@richardyao9012]

I would not expect Tom to publish any positive news he hears about Intel discrete graphics given that he has a vendetta against it.

[@Neonvarun]

@@moldyshishkabob Grinded War thunder 1k hour's on Iris.

[@0perativeX]

$1800 for a 5090 is understandable. It's not great but i can see why they'd charge that. But what grinds my gears is when scummy AIB's add an extra $500 to MSRP just for a different cooler that doesn't even perform much better than a Founders Edition, that's when people start hissing and booing.

[@cajampa]

With the AIB $500 extra it is going to be $3k

[@rkwjunior2298]

​@@cajampaFalse. Unless it's a limited edition.

[@0-B1]

I see AIB being around $2000ish I don't think they believe they can make much more than that honestly

[@nipa5961]

5090 8GB will start at $1999.

[@dgillies5420]

$2299, according to GamerMeld, dude. VLSI is not free.

[@Paelmoon]

5080 is DOA at those prices then.

[@cajampa]

Cope

[@davidlandrum]

Are you sure? The 4080 is pretty much sold out.

[@PCgamingbenchmarktime]

@@davidlandrum lmao the 4080 sold poorly dude, how do you not know this ? its common for things to sell at launch, that means nothing in the grand scheme of things, it sold so poorly that they had to release the 4080 super to price correct, you'd have to a moron to have bought that card at those prices lol. no one wants a heavily cut down card for over $1,000. its just not going to sell well if they try to pull that off again lol. i dont see why they would repeat the same mistake twice in a row

[@PCgamingbenchmarktime]

@@davidlandrum oh you're saying they're pretty much sold out now ? well maybe where you're from but theres been plenty in stock here in australia and they're not even selling at sale prices. Nvidia also stopped production of the 4080 not that long ago, sooooo its not because they're selling in high demand dude, they're just running out stock because they're no longer being made lol

[@davidlandrum]

@@PCgamingbenchmarktime I hear it’s summertime in Australia. You should go outside and enjoy the weather.

[@Gluosnis9]

So 5090 here in Europe will start at 3000€. Great.

[@eggchucker1]

Nvidia's greed is sickening

[@TracksideViews]

Buy shares and be greedy with them. It’s not rocket science.

[@terpfen]

The 4000 series was priced well above MSRP and sold like hotcakes. They're responding to what the market is telling them. Not their fault that people seem allergic to buying AMD cards.

[@Dr.WhetFarts]

@@terpfen People don't buy AMD cards because they are meh, AMD is making good CPUs, not good GPUs. It is a CPU company first and foremost. I use 9800X3D + 4080 and I won't buy any GPU from AMD before they deliver something decent.

[@fffrrraannkk]

Remember to add $300 to $500 and maybe more to all those prices because the Nvidia cards will always be out of stock, and most people will have to buy partner cards.

[@Мирич-з4е]

*Title:* Specs Leak *Video:* Rumours

[@Gr1mPenguin]

So uhm, any explanation as to why Battlemage wasn't cancelled?

[@WujoFefer]

RTX 5060 with 8GB should be called 5050 or 5040… GPU for old games or 3E/EEE „Excel Experience Edition”.

[@libaf5471]

So. A question I haven't seen anyone in tech ask. When will nvidia put dlss behind a paywall?

[@PookaBot]

Wait what is DLSS4 compared to 2 or 3? This crap is so confusing, frame generation should have had a completely separate name.

[@haukionkannel]

More fake frames or only fake frames so every frame is AI generated!

[@AugmentedGravity]

If the 5090 costs over 1999 imma cry.

[@JakeySurani]

Aib models are around 2500

[@jasonhurdlow6607]

Grab a tissue!

[@gucky4717]

It is interesting, you showed the leaked pricing of some Acer prebuild PCs in your thumbnail, but didn't talk about it (or i missed it). The leaked pricing is about a 3499€ prebuild PC with a 5080, 265k, 32GB RAM and 1TB SSD. But there is another PC with the same spec, except a 5090 for 4699€. Both PCs are sold in germany, meaning 19% tax is included. If you convert the difference of 1200€, it is 1050$ between those PCs. So it is very likely that there are 1000$ or less between the 5080 and 5090. A 5080 being 40% faster is quite optimistic, seeing there is no increase in shaders. I personally doubt that it is faster as a 4090, at least without DLSS4. So 20% more performance/value is probably still 999$. That pricing above means the 5080 is 999$ or 1199$ at most and the 5090 is 1999$ or 2199$ at most.

[@ParacelsusCaspari]

i expect 16gb vram on the 5080 is nvidia's concession for keeping the price the same as the 4080

[@gucky4717]

@@ParacelsusCaspari Exactly, same as the similar die size. But if the die and VRAM gets more expensive, it might be a bit more then 999$.

[@ParacelsusCaspari]

@@gucky4717 seems nvidia think gamers don't need more performance lmao. be happy with 5070 ti = 4080 as the "high end" card and leave the rest for professionals

[@joost1183]

Dont use 'faster with dlss' as a metric please

[@ParacelsusCaspari]

@@joost1183 everyone says then enables dlss anyway cuz it looks better than taa rofl (tho i get why not for benchmarks ofc)

[@simongentry]

so why are 4090’s still selling for $2500?

[@Thex-W.I.T.C.H.-xMaster]

Low stock and the 50 series is about to drop.

[@simongentry]

@@Thex-W.I.T.C.H.-xMaster ahh

[@RafitoOoO]

I hope all the people who only want AMD to be competitive so Nvidia don't go crazy enjoy Jensen's better value.

[@TJGibson0]

Is the 5080 still only getting 16GB VRAM? Or will it be 20 or 24?

[@Elvewizzy.]

Id imagine a 5080ti down the road will be 20GB.

[@nipa5961]

16GB at best.

[@zerocal76]

16GB my boy. Zotac accidentally put some specs up on their site. Ngreedia execs laughing at gamers 😑

[@MechAdv]

The 3 GB GDDR7 chips aren’t available yet. Once those get here next year, you’ll see a 5080 super with 24 GB of memory on the same 256 bit bus. Until then the 5080 will be 16 gigs.

[@zerocal76]

@@MechAdv 24GBs on a 5080 but at what cost? $1,300 oem, 1,500 3rd party? 😵‍💫

[@rkwjunior2298]

Nvidia buys ram in huge bulk. They arent paying retail price for parts. 5090 will be 2k$ or less msrp. Everyone wants to exxagerate.

[@Mattribute]

The cost of making it has no effect whatsoever on price whenever it’s a high margin product with no competitor. They will find the highest price that still permits them to clear their supply.

[@damianich4824]

What does bulk vs retail price even mean for gddr? Who buys gddr other than graphics card manufacturers?

[@flimermithrandir]

@@Mattribute For sure. Agreed. Nvidia doesnt care what you (we) want. It will be at least 2000 Dollars. GDDR 7 is new. Nvidia will make sure ppl know it is expensive. Even if Nvidia buys it for 1 Cent for 1 Modul. THEY WILL TAKE YOUR MONEY. Its that simple. (Of course its not gonna cost them only 1 Cent. Thats just to make a Point because i dont know the Price. I only know it will be more expensvie for GDDR 7 compared to GDDR 6X. At first for sure. Nvidia might even pay extra to get it sooner as they have done even... i think to remember hearing something like that at least.) Anyways. No Way it will be only 1800 Dollars.

[@Mattribute]

@@flimermithrandir On the other hand, there are "mid-range" cards that can let you do 4k above 60hz, and 4k 120+ if you ignore silly settings like ultra, that are way cheaper. A few months ago I got two prebuilts with 7800X3ds and 4070Supers and was amazed at how low (relatively) the price was. Perfect for high quality high framerate 1440p gaming. Including the 1440p/144hz monitor, each was under $2k. The amount of fun my niece and nephew can have gaming on those machines is like 99% of what is possible if you double the price and get a xx90 card. 4070S, 4070Ti, 4070TiS and every AMD GPU priced between $350 and $800 since they became relevant again are all okay values in the "not enough supply high margin" era.

[@rkwjunior2298]

@Mattribute It absolutely has an effect. No competition? What about AMD? You can buy an AMD. Or keep your old super-fast GPU that u said was such a bargain for its performance...lol

[@golozhopique]

Why is it so difficult to give couple extra vram chips to laptop version of gpu? to save 3-5 extra watts?

[@1SaG]

5070 *just* beating the 4070 Ti would be a huge bummer - I really would expect more of a generational uplift considering that the 4070 Super is already very close to the OG 4070 Ti's performance. Nvidia seem to have forgotten that the 4070 Super and Ti Super exist, at least when it comes to performance/performance uplift. Been my worst case-scenario for a while now: Nvidia will use the OG 4070 as the 5070's benchmark when it comes to performance, but the 4070 Super when it comes to price. And a 5070 that'll be as much or slightly more than the 4070 Super but only have ~15% more performance than it is a little silly. I still could see myself getting tempted by the 5070 *if* it came with 16GB, but ...

[@Frozoken]

this is why imma go with the 5070 ti as long as it's $800. Should be like xtx raster performance. Not bad

[@nem_bangz4334]

​@@Frozoken doubtful 😂

[@Frozoken]

@@nem_bangz4334 And whys that? You think the 5080 with only +4 sm's will be able to get +30% performance to match a 4090 but the 5070ti with the same +4sm's that are proportionally more of a gain won't be able to get +25% performance to match a 7900xtx?

[@1SaG]

@@Frozoken That's the real question, isn't it? Meaning how much of an uplift will the new cores/architecture provide? Comparing the supposed number of cores on the 5090 with the 4090's number makes me think there won't be *that* much to gain here ("per core") - from the top of my head: ~16,000 on the 4090 vs ~24,000 on the 5090 - which is a huge increase. So seeing those lesser 50xx-cards and their relatively small gains in total cores makes me think anything below the 5090 won't be getting that much of an uplift over their predecessors. For example: 4070 = ~5900 cores, 4070 Super = ~7200 cores, 5070 = 6400 cores. That's 50% increase on the 5090 but only 10% on the 5070. Of course, this is just me speculating here and I could be totally off the mark.

[@Frozoken]

@@1SaG Yeah except the 90 class always scales the worst with cores seeing there's so many. Remember the 4090 has 60% more cores than the 4080 but is only 30% faster

[@allxtend4005]

the crazy thing is that the 5080 will eat up to 450 watts and the 5090 up to 600 that is stupid, we had the RTX 3090ti that was eating up to 600+ watts and now again ?

[@Ja_Schadenfreude]

My Hope is that they don't sell it all and this leads to reasonable prices. Odds are low but one can hope.

[@Elvewizzy.]

4090's hold their value extremely well, so an upgrade would make some sense for that crowd. And I think many of the 10-20 series users will be looking at the 5090 or 5080 as another 6-8 year GPU.

[@MinosML]

Even if they don't sell much of their stock, there's a huge AI bubble going around so the consumer market going poorly is not an immediate concern for them, sadly. Now, if (when) it crashes...

[@confusedcounsel5285]

@@MinosML When it crashes, Nvidia's marketing department will sell the hype of the next fad. Crypto... Umm AI... Umm New Jersey Drone Defense...

[@richardyao9012]

I guarantee that the AI crowd will buy as many of these cards as they can, just like they have been buying 3090 and 4090 cards. A guy on reddit recently posted about this 14 card 3090 setup.

[@iamdmc]

You were right about 4090's getting crazy expensive after the bottom in August 2023, and on your advice I bought my 4090 for 1350 - thanks for that! I can see the 5090 being 2000, with the bottom being 1500 (which is when I'll upgrade)

[@Arakki]

If I want more performance for more money, I go up an SKU. As a consumer I want more perf for less money each gen and a company wants the exact opposite. Trying to justify the price of a product because it has somewhat fancier parts in it while being no real increase in value is not it for consumers.

[@Litoh14]

Will this trend ever stop? Or in 10-15 years the 90's series card will release at 8.000$ MSRP?

[@mcduvall2000]

A year ago, GDDR6X cost $3.34 per GB, I doubt nvidia is paying 4 times as much buying in bulk...

[@MooresLawIsDead]

That's not true.

[@damianich4824]

Who doesn't buy in bulk? I doubt there's a consumer market for gddr

[@aladdin8623]

GDDR6X was an exclusive product by micron and did cost accordingly. GDDR7 is not.

[@richardyao9012]

​@@MooresLawIsDead What did GDDR6X cost a year ago? Unfortuantely, DRAMExchange does not track GDDR6X prices, so we cannot look at historical prices on their site. I do recall hearing that last year, there was a crash in GDDR pricing. I am not sure if it affected GDDR6X. That said, the price Intel pays per GB of GDDR6 is around $4 according to DRAMExchange's spot pricing for 16Gbit GDDR6 ICs.

[@pctechloonie]

Incorrect. That was GDDR6 8Gbit ICs widely reported in May 2023 as having cost $3.34/GB or $26/8GB not GDDR6X. Cost of GDDR6 16Gbit modules is currently ~4$/GB as per DRAMexchange, and Trendforce says that GDDR7 is 20-30% more expensive. Conclusion: GDDR7 is not $12/GB, but more likely 5$/GB.

[@loopba]

6950xt for life

[@nipa5961]

best card

[@emperorSbraz]

69 nice

[@anthcesana]

It's very much looking that way for us. I cannot find a single XTX worth buying and we are on the cusp of a new generation of cards 2 years after it was released. So, long live RDNA2

[@Dr.WhetFarts]

Radeon 6000 is shat on in pretty much all new games due to terrible RT perf and sucks in UE5, no thanks

[@nipa5961]

@@Dr.WhetFarts You mean in all 3 games a year with forced RT? XD Maybe think about the hundreds of games the 3080 is struggling in because it runs out of vram.

[@bignose1752]

Man I just want to see what they're offering for $250-$350. I think I'd buy an 8600 with 16gigs of ram if it performs like a 7700xt for $300.

[@arthurparkerhouse537]

Why is the GB of Vram not increasing appropriately on these? I'd Expect 128GB VRAM Consumer Cards by now.

[@b130610]

Nice analysis, hope you have a good holiday season!

[@MooresLawIsDead]

Thanks, looking forward to the new year!

[@flimermithrandir]

Said the same a few Weeks ago. 2000-2200 most likely. Start at 2000 for their Cards. 2200 for others also possible. I can still see 2500 though as well. Especially at the Start when its sold out. Nvidia will probably be fine with AiBs getting a little bit more out of the Cake at first.

[@ixonal8843]

Nvidia being who they are at this point, it's pretty much convinced me to use either AMD or Intel for my next card...

[@c.s127]

Thanks Ngreedia: GTX 1000 Pascal percentage difference of cuda cores from 80 to 80ti model = 40%. Rtx 2000 Turing percentage difference of cuda cores from 80 to 80ti model = 47.8% Rtx 3000 Ampere percentage difference of cuda cores from 80 to 90ti model = 23.5% Rtx 4000 Ampere percentage difference of cuda cores from 80 to 90 model = 68.4%. Rtx 5000 Blackwell percentage difference of cuda cores from 80 to 90 model = 102.2%

[@mitsuhh]

4090 is only 25-30% faster than the 4080, for double the price.

[@XashA12Musk]

6:24 Bro has been eating Good

[@connorharris1900]

cant wait for the 5090! should i get msi, asus, or jenson?

[@floobstoob3462]

Prices are irrelevant. The price is going to be higher than the MSRP when the card releases. There's low supply and high demand of cards. All the while, scalpers and bots instabuying the cards.

[@kaiserpr0728]

Where I can preorder the RTX 5090? I don’t want to get the scalper route and paying a ridiculous amount of money.

[@richardyao9012]

Lately it is not a MLID video without Tom’s trademark pot shots at the B580, which he does at 18:39. Nvidia GPU launches have limited initial supply too, yet then Nvidia starts producing them en masse and they shoot up the sales charts. Given the B580’s unmitigated success and the ability of Intel to sell these at a profit at scale based on simple napkin math that anyone can do (well, anyone but Tom apparently, as he cannot calculate a 60% margin correctly at 10:03 where he does the math for a 37.5% margin instead of the claimed 60% margin, much like how he had failed to do just about any math correctly in all of his recent videos), Intel should be churning out as many as they can to satisfy demand and reach profitability. It is business 101. The current spot pricing at dramexchange for 16Gbit GDDR6 ICs put 12GB of it at $48. With the GPU die at $57 and everything else at $80 (which is the price of an Asrock AM4 motherboard, which should be around what the rest of the card costs to make, distribute, etcetera), the B580 begins looking profitable with a breakeven point at $185, provided that they can reach the same sales volume that Asrock reaches on its $80 motherboards. Note that this is assuming 214 dies per wafer, which should be a minimum given all of the dead redundant silicon intel put into it, at a price of $12095 per wafer, which is a number Tom posted at 9:36. The truth is that the B580 is profitable to produce at scale, which is exact opposite of Tom’s claim in a prior video where that a $20 loss per card in low volumes would be a $200 million loss if they made 10 million of these. Tom keeps ignoring his severe logical mistakes by trying to point at the few things he did not get wrong, as if it justifies everything he said that was provably wrong. He talks about the truth, but if he really cared about anything other than his ego, he would retract the claims of unprofitability at scale, his baseless slander of Tom Petersen as a professional liar and his repeated cherry-picking every slightly negative thing he can find from positive reviews to try to make the B580 look like a disaster when it is not. Let’s not forget Tom’s clear rooting for discrete ARC to be cancelled in his last video where he put words in Tom Petersen’s mouth about discrete ARC’s cancellation after he declared him to be a liar about 10 minutes prior, and then cheered with a malicious smile at his false claim about discrete ARC’s cancellation. Dismissing his viewers who are objective enough to complain about his bias as corporate cheerleaders (18:53) as he does in this video is a continuation of this pattern of bad behavior. Personally, I do not even like Intel and beyond their SSDs, which they no longer make, I have not brought Intel products in the past 7 years. I dislike nonsense, which is why I feel compelled to speak against Tom’s claims.

[@pctechloonie]

How did you obtain that GDDR6 16bit figure? I can't find it on their website, only quotes GDDR5 and GDDR6 8Gbit ICs. Indeed and Intel confirmed weekly shipments for their own B580 cards. That #52 will shoot up in the sales charts. For those still doubting Newegg bot said this: "Newegg tracks bestselling GPUs based on sales data over a period of time, typically reflecting trends that can span weeks to months." Indeed GM math is incorrect, dividing additional cost by "(1-GM)" is easier, 60% GM = 2.5x multiplier. As for the rest of ther math it's BS you're right. GPU die is not $57 it's closer to 95-100$ with a 20,000$ rumoured 4/5nm 2025 wafer price. A GPU board is a lot more simple than motherboard, so costs are indeed likely the same or lower, despite the additional cooling + VRM costs (negated by lack of chipsets + IO + smaller PCB). Edit: INFO HERE RETRACTED. Forget about the RX 580 math it's based on Frankenstein card, not a new card at all, so not apples to apples. Sorry for the confusion. You can't use that math without adding the AIB, Wholesaler and retailer gross margin on top of the entire AIB cost basis. You have to subtract that from the MOBO cost and adding it to VRAM + GPU die cost and compare that figure with the AIB cost basis (MSRP x (1-AIB cut) X (1-Wholesaler cut) x (1-Retailer cut). I got B580 being sold either at cost or a slight loss in a spreadsheet, smaller than Tom's, but still a loss. This is really TSMCs fault + Intel's lack of an die area efficient architecture. I don't like the way Tom behaves either, this is no different than the past (been here since 2019). so don't think I'm coming to his rescue. I'm not, just trying to provide som nuance about the launch volume + production costs.

[@ElderGamerX]

If the 5080 is comparable to the 4090, they'll charge a similar price for it, since there's no supply of 4090s at MSRP to stop them, and no other market competitors at the high end. Which will free them up to charge $2000-2500 for a 5090. The only reason a 5070/ti will be even remotely competitively priced is because the 8800 xt will exist. And it'll still be more than people are happy with, but they'll buy it anyway.

[@kasmidjan]

Jensen huang is real life mister crab.. #changemymind

[@incognitobr]

I just want an inexpensive entry-mid level GPU. GPU prices are just insane lately and this AI bullsht trend is just gonna hurt us more.

[@thillyard1]

Look at the used market. I built a gaming PC for a friend and picked up a1 year used 6650XT for $150. Card works great for 1080p games at medium to high settings depending on the game.

[@aflyingmodem]

the rtx 5080 will probably be more like 30% faster than the 4080 for 40% more money than the rtx 4080 super so $1399 and they will say it's 4090 performance for $200 bucks less lmao

[@takerulol]

Yes and the die configuration (same as the 4080) is just roughly 50% of the top die ... which historically would be a 70 class card.

[@BenDover-ie3sy]

Nah, 1400 is too high for a 16 gig card.

[@aflyingmodem]

@@takerulol which the 70 class gpus typically would be like $450-$500 now their like $700 so a doubling in pricing for a 70 class card lol

[@defeqel6537]

Nah, it will be better value/$, but will still increase in price

[@aflyingmodem]

@@defeqel6537 I highly doubt that, they have no competition at the high end, there will still be some rtx 4080 supers out there for people to buy, and most likely once people see the price to performance they'll run and buy the rtx 4080 supers, so nvidia will see no reason to do a small price increase, maybe the founders will be $1200, but the aftermarket cards will all be $1300 and above at best, but honestly i see them setting it at $1400 msrp and after market cards being higher than that

[@pyrolight7568]

My question is... what is the point? I finally updated from a 1080TI to a 4070. Not really seeing the reason

[@johnvandeven2188]

I'm expecting the RTX 5090 to be available for purchase priced at USD$2,500. It is very expensive and high end gamers will buy these in droves, at least in those countries whose currency is similar in value to the American dollar. I have no issue with that, afterall it is their money, but in Australia that same video card will cost AUD$5,000 and then it becomes far too expensive for even the most hardened gamer. Why double, well there is the currency value difference of 40% plus there is an additional GST component of 10%, shipping costs of another 10%, handling costs of another 10% and finally the vendors want an additional markup of 30%. Go figure.

[@cajampa]

@@johnvandeven2188 Just you wait. The US will very likely get huge tarrifs f all the GPU buyers. Before tariffs it might be 2500. But after it....who knows how high they will put those on top to force the production to the US.

[@NahBNah]

I just highly doubt founders is going for 2.5k highly highly doubt it. AIB? Sure

[@johndoh5182]

Shipping costs don't make such a big deal for total costs for small items. It's not going to add $50 to the cost of each product. I know that a 5090 comes in a big box, so the issue is how many 5090s fit into a shipping container along with the pallets that carry them. We can assume that companies have gotten REALLY good at shipping the max they can in a single container. So, if the cost is $1000 more/container it's maybe $10 and even that number sounds large, maybe more like 2 - 3 USD max per GPU. But I haven't done the math to try to calculate any of this. With shipping into the US, probably the worst part of the handling of the products inside container is what happens AFTER it leaves the container and loaded onto trucks in the US and then anything past that point because the shipping inside the US is not good. So, what different companies have gone to is repackaging in the US. I don't think any computer parts are repackaged in the US but I know there are Chinese companies that repackage in the US. Yes, it will add a LITTLE cost. However the bigger cost will be tariffs added back onto electronics coming out of China, like they were in 2021 - early 2023. This will hit certain AIBs that haven't gotten their manufacturing, or assembly out of China.

[@Bassalicious]

I just bought a 6750xt on sale for 300€ (including tax and shipping) as an upgrade to my ancient EVGA GTX960 2gig. From what I'm hearing I'm pretty confident it won't fall behind as quickly as the 960 did. Time to catch up on the last 6 or so years worth of games I couldn't play properly before. KCD2 is right around the corner too, very excited for that one.

[@SniffyMcAdderall]

Nvidia is smoking something with the pricing of all cards 5080 and below.

[@Fordance100]

They will price it perfectly, leave no money on the table and still sell a ton of cards. Honestly, 5090 will be sold out for a while.

[@gucky4717]

Honestly, I hope to get one day 1...

[@briank3754]

I can’t believe I just found this channel. This is the most in depth explanation of the 50 series pricing. I’m a 4090 owner and am looking forward to the 5090. The 4090 is great, but this Pimax Crystal is hungry with all these pixels to feed. $2000 for a 5090 is a decent price if you ask me. I know people freak out about the pricing, but there is always a premium to pay for what is clearly the top performer. This isn’t unique to graphics cards. The X 90 cards are a better value than people think. I paid $1700 for my 4090 two years ago and it is worth more now than when I bought it. Even if it costs me $250 a year to own a card, it’s worth it. I don’t know how Nvidia could justify any price increases in the low to mid tiers. Especially with competition with used cards from the 40 series.

[@corey2232]

My critical thinking skills tell me that if the 5080 is launching with only 16GB of VRAM, the tier below that will launch with 12GB. That implies that a tier even lower, the 5060, will launch with less than 12GB of VRAM. And since it's Nvidia, and we all *KNOW* they love increasing prices while offering less gen over gen, the 5060 will only have 8GB of VRAM. In previous gens, the xx70 series cards would offer comparable performance to the prior gen's xx80ti or even xx90 models (if you include "Super" variants). If the 5070 only offers 4070 Ti performance, that tells me they're once again shuffling their GPUs down a tier but keeping the same name. And if that 5070 only offers 4070 Ti performance + a price increase? That's a flat out *SCAM.*

[@pieflies]

It’s not a scam if they’re not being dishonest and people are choosing to buy it knowing the specs it has. That’s just people over paying, not a scam.

[@joost1183]

Yes they're going down a tier. As soon as cuda leaks came out it was obvious

[@MooresLawIsDead]

12GB isn't enough, and they know it.

[@Frozoken]

No it won't the 5070ti will have 16gb ram which makes total sense when the 4070tiS literally exists. Don't be stupid

[@LucianPrimeOneMillion]

If i paid for my 4090 here in South Africa 2800K then I don't even want to imagine how hard they'll hit us for the 5090.

[@andrescarrasco1248]

I hope rtx 5070 goes for $699, gamers love the Monopoly so pay for it

[@lasvista2tech]

they're probably going to increase vram slightly when RTX 50 Super series comes out in 2026 knowing nvidia being nvidia they'll try to make max money from the initial hype of the card and then make even more money from the vram increased super series from those who didnt buy and complained.

[@akin242002]

$600 to $700 for the RTX 5070 $1,200 to $1,500 for the RTX 5080 $2,000 to $2,500 for the RTX 5090

[@4evahodlingdoge226]

The $1200 4080 didn't sell well, it will be $1k just like the 4080 super, they were looking for price discovery and they found it. The 5090 can be anywhere from $2-3k due to no competition and the fact it won't just be for gamers.

[@markterek6232]

@@4evahodlingdoge226 Sir.If it didnt sell well,tell me where are the cards? 🤣 They sold every last part 4080 for 1200 easily.And the 4080 super never reach that 1000$ starting price.Sorry........

[@kael13]

Can't wait to not buy. Back in the day I thought £600ish for a 980Ti was a lot and my max... $1000 is my hard limit on a GPU, I simply refuse to pay more.

[@rangersmith4652]

There is just one reason why Nvidia prices are so high -- people pay those prices. Competition from AMD won't lower Nvidia prices. We know this from comparing RDNA3 price/performance to Ada Lovelace price/performance. The only way to get Nvidia to lower prices is to stop paying the asking price. We've seen evidence of this as well in the 4080 vs. 4080 Super prices. The 4080 sold poorly, so we got the Super for $200 less.

[@haukionkannel]

AMD did outperform NVIDIA price wise for years. And customers did buy NVIDIA instead! So yeah! This is our fault!

[@N0N0111]

My volume was at 40% for all other YT video's, for some reason my ears wanted this video to be on 5%.

[@jackkunasaki5044]

tbf, some 4090s are already selling for 2k. I wouldn't be surprised if it costs around $1900

[@snakeinabox7220]

brother 1900$ is on the low end, , i was expecting min 2k . prob 2200-2.5k The same price as Former TItans .

[@christophermullins7163]

​@@snakeinabox7220If Nvidia gives a 20% uplift to price to performance of the 4090 street price.. $2500 easy. I wouldn't be at all surprised. "This is a work GPU, there is no $2500 gaming product" just to minimize the backlash

[@liberteus]

MRSP and street price are two different thing. Jensen could go "oh and it's only 1499USD" and we all know nobody will be able to buy it at such a price, because demand will be so high anyways.

[@pctechloonie]

Tom I really think you should end the ARC BOM cost debate once and for all by publishing your BOM math for that 20$ figure, because people keep complaining and whining that ARC is profitable and you're lying, which is BS. I've tried my best too estimating the costs with a spreadsheet and I can't get this to be a anything other than a slight loss or at best a breakeven, which is terrible considering the additional design and support overhead. HUB also stated that their contacts said that this is not a low volume launch + Intel stated that weekly shipments are coming. Is this wrong info or misleading? Oh and Tom Peterson kinda has already admitted that ARC is not making them any money on HUB's podcast, go to 26:35 in the video.

[@richardyao9012]

Multiple people have posted estimates showing that it is profitable. Those who do not are using ridiculous $100 or even $150 figures for the GDDR6. The actual number is $48 according to dramexchange. You can check this by making an account with them and then looking up the 16Gbit GDDR6 IC average spot price. Multiply it by 6 and you get the cost of the 12GB GDDR6 used on Intel B580 cards. Interestingly, Intel actually left some cost optimization on the table. If they had used a clamshell GDDR6 configuration using 8Gbit ICs, the memory would have only cost $28. As for the GPU silicon price, using gamers nexus’ die measurements in a calculator shows that Intel gets at least 214 dies. At $12095 per wafer that Tom reported for TSMC 5nm in this video, the B580 costs $57 per die. Given all of the dark redundant silicon in the die, we can assume that they can use close to 100% of them between the B580 and B570, so the die price is unlikely to be any higher than that. Try doing your spreadsheet again now that you have these figures. Also, it is well known that making more of a something lowers its price due to economics of scale. When building only 10,000 cards, it is quite possible that Intel will at best breakeven. When building 1 million cards, Intel is likely to turn a profit. Tom Petersen made those remarks before the B580 was a success when Intel likely expected the market not to buy many B580 cards. The profitability will change as Intel ramps production to meet demand. Finally, I would not count on Tom being able to back up any numbers he produces since he seems to be unable to do math correctly in his recent videos and likely in many prior videos that I have not seen. In this video at 10:03, he claims to calculate the cost after a 60% margin, but used the wrong number. He had to divide by 0.4 or multiply by 2.5 to calculate the cost after a 60% margin. Multiplying by 1.6 is for the cost after a 37.5% margin.

[@pctechloonie]

​@@richardyao9012 Thanks a lot for that info. Will definitely make an account then. I know. Didn't agree with Tom here either. That clamshell idea is very interesting, but fear the volume is no longer there. It's only a matter of time before those 1GB production lines are retired. Just started a thread on Reddit about that wafer price used in the video and everyone agrees that it's too low + they think 20K figure is more accurate. This is where I get that 96$ figure from or 92$ with 100% working silicon. Thats a total of 140 bucks. Then there's the AIB, retailer and wholesaler ~5% cut, leaving a cost basis of 213.48 bucks to cover everything: heatsink, fans, fan controller logic, VRM components, control ICs, caps and resistors and various other ICs on the board, IO and power outlet, PCB, transportation, backplate and shroud, packaging, testing/validation and assembly. Subtracting GPU + VRAM = 73.5 bucks. That figure will get completely eaten by those additional costs. I've tried to independently check this math with days of work trying to find the best info available online. I even used an LLM, which took hours to get proper answers based on info it could find online. It came within 5-10$ of my estimates. I can't push the numbers to more than at best +5-10 dollars on each ARC B580 sale. But that -20$ makes no sense. Can only get that figure with 8-10% cut for wholesaler, AIB, retailer. Made post to get a response from MLID, where he publishes his BOM math for B580. I hope you're right and Intel has secured and insanely good deal with TSMC + get components for less than their competitors, but I doubt it. 100% the GM math is completely wrong.

[@posmoo9790]

I always thought $2400 for the 5900 once I heard 32gb that's why I was unhappy to hear 32gb

[@gucky4717]

And (AAA) developers want even more VRAM.

[@pieflies]

Why are you unhappy it has 32gb? The more the better for lots of use cases. If you don’t need that much you get a lower model that has less for cheaper.

[@mulsimin__]

This seems to be expected from Nvidia, They see that there is no Competition on the high end so they Raise the price just as much as they raise the performance.

[@danh2716]

Would you do differently? ...ok, now be honest. Yes, you would.

[@mannyc19]

4nm from TSMC is a lot Cheaper then it was 2 years ago for the 4k series, AND GDDR7 is cheaper then GDDR6X, SP there is no need to raise the price on any 5000 series, except pure Greed.

[@darkfire3691]

GDDR7 cheaper than GDDR6x? Tell me your source.

[@WaterZer0]

Shareholder value must go up.

[@Dr.WhetFarts]

Its called no competition. Business 101. If AMD was in the lead, they would be doing the exact same thing. Look at their CPU prices now.

[@HYBECTIVE]

I dont understand consoles usually have a life of 6 years, so why do so many people feel the need to update every generation and complain?

[@XMG3]

I strongly advise not buying the 5080 16GB. Here's what I found: first, my setup is 7800X3D, TUF 4080 Super, 32GB DDR5 6000Mhz, CL30. So in Indiana Jones the great circle. I'm getting 130 FPS at 4k DLSS performance, frame gen on, ray tracing high (not ultra). As soon as I switch to DLSS balanced, I ran out of 16GB VRAM and get 3 FPS... So even before the release of 5080, 16GB is not enough to run DLSS balanced with high ray tracing! This is completely unacceptable, imagine spending thousands and get 3 FPS on DLSS balanced with ray tracing... what a joke.

[@N4CR]

LMFAO I'd heard 2 games prior to this, but that's hilarious. Patting my XTX. It's a little slower in RT but at least it'll do more than 3fps lmao. Damn. Sell it now and get an XTX before the new 5080 launches?

[@nightowl3582]

Gonna be hanging onto this RTX 4080 indefinitely. It's just gonna have to do for quite a while longer.

[@wills.5762]

PCs are getting close to pricing themselves out of practical use for entertainment. Ive always liked PC gaming because it was by far my cheapest hobby. It still is my cheapest hobby, but I cant fly or race my PC, and my new rig is worth 25% of my truck or 40% of my ultralight. The way Nvidia is driving the market, it will probably be my last PC, could get a lot more bang for my buck putting that money into other hobbies.

[@qntmsea4631]

Nvidia does not need the gaming market any longer, and yet it still dominates it without trying. Of course the prices will go up - way up! I don’t really understand how could this be in doubt. This is what technological dominance looks like.

[@bankaimaster999]

Regarding the 5090 laptop leak... can't you recheck with your source or other sources to corroborate the 24GB VRAM variant leak? I mean, if you have so many sources, it's already so close to an announcement; even companies like Asus are slipping up with leaking their own products online... so is there no way to double-check that specific leak? I mean a 24 GB RTX 5090 laptop, if corroborated from more than one source, is a major leak and a potential "buyer's advisory" for laptop consumers. Additionally, there are leaks of the Intel Core Ultra 9 processors paired with the 5090-5080 GPUs from the Asus leaks, but nothing has been seen or heard about the Fire Range X3D processors from AMD coming to ASUS laptops. Any news on those?

[@4evahodlingdoge226]

I was expecting the 5090 laptop to be 24gb of vram with a 225-250 wat tgp, it can reach 4090 performance if those are the specs, otherwise it's going to be pathetic.

[@bankaimaster999]

@@4evahodlingdoge226 I doubt they would increase the TDP to that level, but I would have expected Nvidia to delay just the 5090 series for laptops in order to put the 3GB modules from Samsung. Those should ensure the same 175W TGP for the GPUs while increasing VRAM to 24GB, boosting performance, and giving them a reason to price hike it. A 24GB 5090 laptop, at faster speeds due to the newer module, is much more compelling to buy over a 5080 one because currently there is no just reason to get a 5090 laptop when the 5080 is on the same die, has the same VRAM, is a lot cheaper, and generally the gap in performance compared to the 5090 counterpart is small (maybe 10-15% less powerful depending on the brand). The 4090 had a clear VRAM difference that I believe also contributed to its performance difference over the 4080 (which was still small)... so now all I can say is it's just a stupid decision by Nvidia to do that because there is no proper incentive.

[@brucewayne_ind]

Love your content and unbiased opinion on tech 💓