Welcome back to Hardware Unboxed. It is that time, the November Q&A. I suppose if you waited all year, it's that time, isn't it? We've had so many product releases this year. I've barely been able to keep up with what month it is, what's going on, what's happening. So, given all of that, there is no doubt going to be tons of questions about all of the things that we've seen recently, all of the things that are happening. So, I'm really excited. I don't know what else to say.
I just like the idea that someone has waited the whole year specifically for the November Q&A, not any of the other Q&As that we've made. This one specifically.
Yeah, because like, you know, around the end of the year, November, you start to get all the new products. Like, we've got those RDNA 5 GPUs that just dropped, Zen 6, Nova, like all of those things that you'd expect to happen around now that just happened. So, we want to talk about it.
So, let's do it. But before that, today's sponsor spot is brought to you by Thermal Grizzly and their Wire View Pro 2. The best way to safeguard your 12-volt high-powered connector from melting. This little device plugs in between your graphics card's power connector and the power supply's cable to measure current, voltage, temperatures, and the power of each individual pin. It has a built-in display with configurable views for per-pin current, total power, and temperature, plus audible and visual alarms if thresholds are exceeded. The CNC machined aluminium body includes a semi-passive fan for cooling, automatic data logging, and optional automatic system shutdown via the included splitter cable for extra safety. Thermal Grizzly are so confident that the Wire View Pro 2 will keep your graphics card safe that they're backing the product with an extended warranty covering the GPU connector for 2 years. That is a crazy level of support. So, for more information, please check the link in the video description. All right, Tim. First question leads with: do you think? So you've got the crystal ball there.
I know. We've actually brought it out. How good is that?
Yeah. Usually it's hidden behind us, but we've both put it front and center for you. Even polished it.
Did you? Yep. It's looking very clean, to be fair. Yeah.
Okay. Well, do you think the 5080 24GB will receive an MSRP at all given its market value is higher than the 5090 MSRP? Will Nvidia just make up a fake MSRP? I guess if they do go with an MSRP and then sell it for $3,000 US anyway, I think the most likely scenario, just going back across history of sort of looking at what Nvidia's done in the past, would be probably the RTX 3080 12GB launch. So that was sort of right in the middle of the cryptocurrency boom. The 3080 was obviously no longer $700 when the 12GB version launched. So, and I believe they just put that out onto the market with no real MSRP. They didn't say the 3080 12GB is coming and it'll be $800 or $900. It just slotted into where it sort of made sense compared to the existing parts. So, I think at the time, you know, the 3080 was well over $1,000, the 10GB model that is, and then to get the 12GB version, you had to spend a little bit more. But they didn't explicitly tell you that. It was just, this is an option for people. So, I wouldn't be surprised if they did something like that. But, you know, because anything else wouldn't really make sense, right? They would have to launch it for at least $2,000 in the current market. And then if they made that the official price, everyone's going to be, "But the 5090 is $2,000, so how do they avoid that?" The answer is not making a price, I would have thought.
Yeah. I mean, ultimately, I guess it doesn't really matter because we're not going to review it.
I think maybe people are concerned we're going to review it based on a fake MSRP and be like, "Oh, this is pretty good value." Or maybe it's worth... I don't think that would happen anyway because even a fake MSRP, like they're not going to say it's going to be, you know, $1,000 to compete with the 9070 XT, for example, which is approaching probably $900 at this point.
You know, it's not going to be $1,000, and it's not going to be $1,200. If they give it an MSRP, it's going to be well north of that. So, at that point, we would just recommend if you want high-end gaming performance, get the 9070 XTs before they become absurdly expensive. And if they give it a $2,000 US MSRP or whatever, it's like, well, I don't know who's paying $2,000, who's recommending anyone pays $2,000 US for that level of gaming performance. If you have 9070 XTs that are $1,000 or less, it's not worth paying twice as much money for a 24GB 5080. Surely.
I wouldn't have thought so. I mean, the expectation is that this product is going to be the 5080 16GB that we have right now, but with 24GB of memory. So, no other changes. It's not going to have more cores unlocked because the GB203 silicon is already fully unlocked on the RTX 5080. It's probably not going to be clocked higher. I don't think there's much headroom there for that. They're already pushing the silicon to the limits and I believe the memory is pretty much going to just be the same sort of spec as well. So, I mean, there's going to be more of it, but it's not like you're going to get more memory bandwidth because they're just shifting from the 2GB to 3GB chips. So, yeah, we already kind of know what this product is going to offer. We've already tested the 5080. What the 24GB version is going to bring is, you know, maybe in some games where you're right on the edge with 16GB, you get a better experience, but those games are hard to come by today. So, yeah, I mean, who's spending $2,000 for... I can't see this thing flying off shelves, but then you sort of look at the market and people tend to spend absurd amounts of money on things at times, and because it is the, I guess, fastest gaming GPU that you can buy, there's always going to be some audience for that. So, it's hard to say, but yeah, it's not going to be like the 5090 was where it was so much faster than all the other options that you'd sort of make some justifications for it. It's only going to be a small amount faster.
It's a budget AI GPU. Disguised as a gaming product.
Pretty much, yeah. If you look at like the 4090 used pricing, they're up around, I think, $3,000 used at the moment. Obviously, it launched at $1,600, so people have made a profit if they still have that card. And so, you know, the 5080 isn't as powerful as the 4090, but I think a lot of people are buying the 4090 for its 24GB of VRAM as opposed to its performance.
So, the 5080 is going to be like a light version of that, basically.
So yeah, I don't know. We'll have to see where they're actually going, I guess, once it's launched. Are gamers buying it? Are AI people buying it? Is it being funneled into servers in like Singapore, quote unquote, China, actually? We'll have to see. But yeah, who knows? It's supposed to come next year. We'll see.
Do you think AMD missed an opportunity by not releasing a higher performance RDNA 4 GPU? For example, a 9080 XT 24GB.
So, you know, this is the sort of question we get asked pretty much every generation when it comes to Radeon GPUs. Did they miss an opportunity? It's a crystal ball question. It's not even a crystal ball question. And it's, well, it is in part, but it's also, you know, what were the technical hurdles that they faced they needed to overcome in order to make that product? You know, how well does the RDNA 4 architecture scale up to, you know, deliver that level of performance? Is it because we've seen a part like a 9070 XT, for example, it's not particularly efficient, 9070 is much more power efficient, so, you know, that doesn't necessarily translate to a bigger RDNA 4 GPU, but it also could. So anyway, there's just unknowns there. I haven't looked into it. I haven't asked questions because it's not really a relevant product. It's not something we would cover. Not something they made. But look, if they, I guess if they could make it and it would make sense, it would be odd that they didn't make it. I guess you could attribute that to just AMD stuff because unexplainable AMD stuff often gets filed under that. But yeah, if they could have, obviously they missed an opportunity, but I like to think there's a very good technical reason for why they didn't.
Yeah, I feel like this question could have been asked. I assume it's being asked right now because the 5090 has been discontinued, right? There's sort of maybe more of an opportunity to have this high-end product. I think it almost doesn't matter whether the 5090 is available or not as to whether they missed an opportunity because maybe there's this stigma that Radeon doesn't offer high-end cards. You know, it's only a mid-range option for people, which obviously has like brand and mind share implications, even if it's not a particularly great high-end card. I think people still want AMD to be competing in those segments, sort of like what they did with the 7900 XTX, which wasn't the fastest GPU, but was more of a high-end card than the 9070 XT. I think what happened though, I don't know whether this is all just my speculation, guessing, my take on what has happened, is that with the RDNA 3 generation, they saw how the 7900 XTX performed in the market, like how excited people were for that product, how well it sold, how well the 7900 XT, the cut down version sold compared to, you know, GeForce GPUs, how the whole lineup sold really. I think they made the conclusion that it wasn't worth investing in that tier of product because it didn't go that well for them. I'd say the whole generation probably didn't go that well for them.
Yeah, I think that's probably most likely. So, if that is the case, then they did miss an opportunity because they don't have a working crystal ball. And so, if they had their time again, you would think now they would release a higher-end model.
Yeah, I think that was potentially as well a misread on the situation, like why RDNA 3 wasn't successful. I don't think it's because necessarily that everyone was always going to buy GeForce, like that all high-end buyers just automatically go to GeForce. I think there were problems like the performance of the product relative to the price. Obviously, you've talked extensively about the 7900 XT and how poorly that product was priced based on how it performed. They were clearly expecting it to perform better based on the performance they showed versus how it actually performed. So they were clearly expecting it to be a bit better there. And then the lack of investment in features, something that's really come to the forefront today with things like upscaling and all that. I think if they knew that they were going to be able to compete on, you know, have an FSR 4 for example, and be able to improve the AI performance of the cards, improve the video encoder, and do all these things, then in hindsight they would have gone, well, we've actually got more to offer now to gamers, and so maybe there would have been a bit more interest in a card that was more performant than the 9070 XT, especially as how things have gone, it has been relatively successful, I would say so from that sense, regardless of what happened to the 5090 and it being discontinued now, it would have made sense from that perspective, even if it was sort of 5080 tier or something like that. I think there would have been interest if it was sort of like a more affordable version of that product. But again, sort of AMD things, right?
Yeah. I think if I had to guess, it'd be less sort of technical reasons and more than looking at what RDNA 3 did, like you said, how the 7900 XTX and XT were received. And it would just be a situation where, well, you know, the 5080's $1,200, the 5090's what, $2,000 MSRP meant to be, right? And, you know, are they going to sell at say $1,700? And people are going to be like, "Don't bother saving the $300. If you're spending that much, you might as well get a 5090." That was almost certainly going to be the situation, especially at the initial release when they didn't have a lot of those features.
So, they probably thought there's no point making this really large, complicated, expensive die when we can just make a smaller die that, you know, ends up being a bit more popular.
Yeah. And the 5090 was like almost a 600-watt GPU. So, if they sort of saw the writing on the wall there, what are we hearing our competitors doing? And the answer is a near 600-watt GPU, and they're not able to hit those targets. So, even if it largely maybe wasn't a technical reason, they might have still been limited in what they saw they were able to achieve and were like, well, how competitive are we really going to be? And then obviously there's silicon allocation as well. Like they would know we've bought so many of these wafers. Is it better to make this huge die and make one high-end product that's not going to be competitive with the 5090, or make two 9070 XTs maybe, or two 9060 XTs with that same wafer and hopefully, you know, just through volume or something else it'll work out better for them. So I think they made those calls. But yeah, obviously right now you're sort of seeing that the 5090 doesn't really exist. There was an opportunity for a product to be the fastest gaming product from Radeon right now. But I don't think they would have foreseen what has happened because those cards were released over 18 months ago now. So things are just so different now. They couldn't have possibly predicted that. I don't think no one could have.
All right, Tim. I see you've added a spicy question here. I imagine this one was probably fairly heavily upvoted.
Yeah, I think I know what you're referring to. Yes. So, the question is, will you willingly part with your RAM sticks if Sam Altman asks you to donate it?
So, this is a bit of a funny one. I saw what he said. I saw what he said.
Yeah. Like, what he said was obviously dumb. Not reading the room. I mean, that's probably an AI bro CEO, tech billionaire out of touch sort of stuff.
Yeah. So the original conversation was specifically about local AI inference, you know, running AI models on your own computer rather than in the cloud. He wasn't proposing that, you know, consumers be prohibited from buying RAM. But, you know, that's the general context of it, right? What he actually said was from a global allocation perspective, actually it's quite terrible to have everybody have a bunch of the world's limited supply of DRAM sitting in their own computer and being used 5% of the time. Don't know why you can't with 5%, but whatever. I suppose you're not using all of your memory all of the time, so maybe that's how he derived it. Doesn't matter.
It's a made-up BS thing that he said. It's not based on anything, I don't think.
Yeah, I don't know. Yeah, just failing to read the room. I don't know what he's trying to convince people of or why that was a smart thing to say or why he thought it was. But yeah, he went on to argue that RAM would be used more efficiently in cloud data centers. So obviously that's what he's trying to promote because that's the product that, you know, OpenAI have essentially, where, you know, hardware can serve users continuously. So using all of our hardware all of the time because somebody's wanting to make an AI slop something or whatever. And then on November 4th, he jumped on X to confirm, I actually do want the world to have tons of memory and people to put as much as they want in their computers. So whether he actually wants that is unclear, but sure, he said it.
Yeah. Ideally, what he wants is everyone paying more for his service than what they're ultimately willing to pay and then use it, because then that makes them profitable and makes them money. I mean, his comment is dumb and not reading the room, but he's not wrong when he says that, you know, it is more efficient from like a hardware utilization perspective, servers more efficiently utilize hardware because they run all of the time.
I mean, yes and no. So if you dive into this, right, his entire product is based on other people who are creative or, you know, have made things or come up with things, whether it's mathematically or creativity or whatever, they've done things over time that these models have been trained on, right? So, how valuable is that, you know, that 5% if he wants to put it that way, that those people have created those things?
Yeah, I guess, I mean, that's an entirely different way to interpret it, I guess.
So, I guess I'm just putting the value and the worth on people using their own computers to do things. No, I'm not denying this, because if you use your computer for 5 seconds, but you like cure cancer or crack some sort of code or come up with something that's invaluable to the human race, but you only used a small portion of your computer to do it, then... And also, what has OpenAI, since it's come along, what really has it done?
It's consumed huge amounts of power. They've done all of this stuff. And I'm not saying it's worthless. Like, I don't actually hate AI and think it's useless. I think it's a really useful tool and can do some things really well, but it hasn't exactly come along and, like Sam Altman famously said, cured all diseases and done all of these grandiose things that they said it would do. It's like if I grabbed you, if I plucked you out of existence 5 years ago and brought you today and then you had a look around, apart from the fact that you can't buy a computer at a reasonable price and there's all this product scarcity, you'd be like, "Oh, yeah, this all looks pretty similar to..."
Oh, it's the same thing as like, you know, someone promises flying cars in like 10 years from now. Or someone... No, Elon Musk did it though. He did. He did it. I'm joking. Move on.
He did a controlled fall of a car using rocket thrusters. So, congratulations, Elon, on that great success that you'll definitely be able to buy. I mean, I sort of liken it maybe more realistically to like self-driving cars, where it's like there is technology right now that can obviously drive a car by itself, but it's not ready for mass... like we're not in mass adoption. No one is just constantly using self-driving cars. Similar to how... Well, yeah, there's all this talk, right, with AI about how like it's going to, you're going to use it for everything. Clearly, that's not true right now. People are not using AI for everything and don't have a need to use it for everything.
So, yeah, I think his comments are really tone-deaf. You know, I was more meaning it from like a pure technical perspective. Like if you have a limited amount of RAM, then a user using it in their PC 5% of the time is less efficient than a server using it 100% of the time. But whether or not that's actually good is another question. I don't think it's good to have the world's supply of a resource go towards a mega corporation that claims to utilize it 100% of the time.
To use your favorite analogy of a turd sandwich, Tim has a big thing with turd sandwiches. It's like... a big thing. What is this? A big thing. Turd sandwich, what?
You've brought it up in your content way too much.
I've brought it up a few times. Yeah.
It's like how efficiently can you make a turd sandwich? I don't know. Why did you bring the turd sandwich? Right. I just... You always crack me up with your turd sandwiches. I swear there was a video where you referenced it like six times talking about GPU prices or something. Maybe I'm making it up.
I think I do. Maybe I vaguely recall that. Anyway, it's just not reading the room. Those comments definitely shouldn't have been made, especially at a time when you already know everyone's sort of out to get you. Like people don't like Sam Altman, people don't really like AI in general, you know, they don't like OpenAI. You try to win people over or convince them that, you know, your product isn't as bad as what they think. And then to say that sort of stuff, especially when you have other CEOs like Jensen who have gone on the record saying like scarcity is good, it's a good thing for him, like it's good that people can't buy the things they need and want because it makes him richer. With all these CEOs saying so many dumb, stupid, failing to read the room things, it's just he really couldn't have picked a worse time and a worse thing to say regardless of what he actually meant. There would have been a hundred ways he could have conveyed that message better, but at the end of the day, it's still entirely self-serving. You're talking about a thing that again, you're making about people in general and then my thing, like really everyone should use my thing because that's better for me and my company and ultimately I will make more money. So I don't want you using your thing. But even then you need to use your thing to use his thing.
Yeah. I mean, that's what a lot of company CEOs are trying to say, but they're trying to say it in like a more diplomatic way. Like that's exactly what they mean. They mean, we want to make the most money that we can. But that is obviously not an appealing message to people, so they try and massage it. But these people are so... I want you to have nothing so I can make as much money as I can. The thing is though is that they're so out of touch with the everyday person that they think that when they're making these justifications that they're sounding very reasonable. Like when he says it's better to have the RAM in the servers, he thinks that is a reasonable point that an everyday person is going to relate to and be like, "Yeah, I'd love to use more AI." Surely the everyday person wants to use more AI, right? So it'd be bad to have the RAM in the servers so that the servers can do more AI. That's how he's thinking. He's not thinking how normal people think, which is... and you would have seen this, you know, not to get super political or anything, but recent polling in the US because they've just had midterm elections, one of the real deciding factors for a lot of people has been AI. They've played...
What are your key concerns at the moment? And along with all the economic factors and stuff, especially people who live near AI data centers being built, they don't want them being built, right?
That's a key deciding factor for people in these things. So, I don't know, maybe Sam Altman and his friends will get more of a message now that, you know, the polling data is sort of showing these sorts of things. It's not just a small group of people that don't want to oppose it. It's like a large group of people in existence in the general population have a very negative opinion of AI. And you can't just go around trying to sell your company being like, I'll take away what you want to benefit my company because that is not going to work if the majority of people are like, I actually don't want this.
So that's just, yeah, not reading the room and
Yeah, a really, really poor, really poor form for him. But it's like tech CEO stuff, right? It's not the first time we've heard tons of completely out of touch and deranged statements that are made by these billionaire CEOs. It just keeps happening. It just, you get into that position and you just lose your mind. That's what happens.
Yep. I was going to add to that to make it worse, but you know what? We'll just go to the next question. All right, Tim. Glad we got the crystal ball there courtesy of Balon because this question starts with: how long do you think it'll be until the release of a consumer/gaming GPU that surpasses the RTX 5090 performance? Over or under four years from now?
Well, I think they're going to try with the next generation. However realistic it will be to purchase it is another question because obviously I've sort of been claiming for a while prior to the 5090 being discontinued that while it was available, the price was too high and kind of didn't make it a gaming card. So, if we saw like an RTX 6090 come out and it's like $5,000, but it's technically like a GeForce gaming GPU, is it like, it may be faster and it may be a gaming card according to Nvidia, but is it really like a gaming GPU at that price? That is sort of maybe a more relevant question because I think like it's pretty clear that there will be a 6090 next generation. They're going to try and sell it for some absurd amount of money and it will probably be faster than what they've currently offered. But the question is whether you'll actually be able to get it for any sort of reasonable amount of money.
Yeah. Look, it's an absolutely, even without crystal ball, impossible question to answer because what's your prediction for this time next year? Because that's about, I imagine, the time frame that we're looking at the GeForce 60 series being released late next year. I would think somewhere around that, you know, give or take.
Well, I think that would probably be the earliest, right? And then the latest you would think maybe six months after that.