Hi, this is Wayne again with a topic “Intel ARC is facing even MORE issues”.
More trouble for intel arc. It feels like just a few short weeks ago that we had representatives from intel here on the show telling us that things are going swimmingly for ark, and we should be expecting a launch imminently. Wait. That is exactly what happened: yeah, that that exact thing i just said that it’s literally a fusion that wasn’t a fever dream right, nope like that was a thing: okay, cool well, according to igor’s lab, as if all the other rumors of intel’s gpu misadventures haven’t been Bad enough for team blue, it now seems according to rumors, allegedly that third parties are abandoning ark. Igor says there are lots of unsold but already produced graphics cards sitting in inventory, likely thanks to the uh fall of cryptocurrency values and the scramble to keep production high to meet that prior demand. That is a bad time.
Yeah anthony writes, which, as linus should know, is a pretty bad time when it comes to tying up your capital. As a result, intel has been focusing first on oem partners, but they face a chicken and egg scenario. In order for system integrators and oem customers to place orders, there must first be a demand for the products to begin with, but since arc is an unknown quantity i mean all we’ve seen is the small demos that intel has done since it’s unproven.
There is apparently little to no interest from prominent european distributors and dealers, and this was the case long before a380 benchmarks reared their ugly head there’s another still. There was a. There was still a weird way to launch it, considering it’s. It’S basically always been halo product and then dropping down from there, and they did like the opposite, which i think amd pulled it the other way once when they launched. I think it was polaris. I want to say okay, mind you, i don’t think they had a high-end product for that generation like at all, so maybe they were just pitching it as no no, they launched because they did 470 for no, they launched the whole stack. They just didn’t.
Have a high-end product, i remember now: nope nope, never mind yep you’re, right yeah, it’s weird! So i i i think that that set people off on a weird foot as well. You generally try to put your your best foot forward when it comes to launching a generation of products. Now, there’s another angle here: intel allegedly didn’t want to give price guarantees and the conditions around rma returns were worse than their competitors, which is a tough pill to swallow for system integrators whose margins, at least on desktop computers, are often low.
You know what i want to know. I want to know where these rma conditions for returns are coming from, because back at ncis we couldn’t return jack when we bought something it was ours. How did we get host so hard? Is it because we were canadian that sucks? That’S a pretty good joke. Okay, to be clear, we did occasionally get stock rotation and price protection, but once once you committed to having something you pretty much, you pretty much had it. If i was intel, though, i would recognize that i’m going to have to offer price protection of some sort for these arc gpus, because if i was a reseller i mean i remember going through this back when we did the original wan hoodie and our strategy for International sales was working through resellers and we basically had to bend over in order to get the thing in stock in the store because they were sitting there going well. If we don’t sell it, then we want this to be entirely your problem.
We want you to cover shipping back to you, take back the entire thing for the entire, the entire value and because we were just you, know small and had absolutely no leverage whatsoever we pretty much had to take whatever offer was given to us. So uh hilariously um we had, i think, three partners i want to say it was either two or three. I can’t remember it’s been a long time. I think we had three partners, though, and the only one that didn’t manage to sell through and that we did end up taking inventory back from was ncix yeah. I know right they weren’t even the ones that gave me a really hard time about terms. I just told them like just look, look, just don’t worry, we’ll get it taken care of, and then they just like, didn’t sell it they’d i mean that’s ncix right and just yeah.
If your product sells itself, then great it’ll be fine, and if it doesn’t, if they have to actually do any work, then it could be a challenge. Their site was so bad. It was atrocious yeah anyway, without interest from commercial customers, then board partners are kind of left out in the cold. I mean man, if you guys have been following along with the screwdriver and the backpack saga.
I mean these are relatively uncomplicated products compared to well. First, you’ve got this. This highly complex, integrated circuit right, you’ve got you’ve, got this asic right. Then you’ve got the highly complex board designs, uh, the development that goes on around those the mass production that has to happen in order to reach the kind of scale that it even makes any sense economically.
To produce these things, then there’s the distribution. You got to kind of you got to kind of bet, move all your pieces into play globally. You got to bet on where people are going to buy these things and and which ones they’re going to buy and what volumes like it’s. So it’s so complicated and all this has to take place in the in the years and then months and then weeks leading up to actually pressing go. This is available and it’s all got to be basically perfect or you end up looking like an amateur right, because you’re going up against amd nvidia who’ve been doing this for decades, yeah. So without interest from commercial customers. Board partners are left out in the cold and one has allegedly halted production altogether, and this is due to quality concerns.
Whether these rumors are true or not, they are emblematic of the challenges faced by anyone attempting to enter the gpu market in this day and age. Without a clear value, proposition or brand recognition, sis are going to have a tough time, accepting it without sis accepting it board. Partners are less likely to hop on board and without performance leadership or aggressive back-end agreements on pricing, especially price protection, not to mention rma support. Neither situation is likely to change, so the cycle continues until either intel blinks or the proverbial plug gets pulled. Now we do have some discussion questions here and i don’t like this first one is art gon na be another larabee.
I don’t. I don’t think that’s a realistic outcome at this point like i think i do think intel is a smart enough company to not go completely sunk cost fallacy. Well, we’ve already put billions in what’s another few billion, but it is very clear from a strategic standpoint that having a gpu play is what the a players do today you must have in addition to your x86 compute product. You must have a gpu compute product and it’s also clear from amd and nvidia strategy that, in order to hit the kind of volumes that you need, you want a consumer product. You don’t just want it to be a data center compute product which is ultimately where larabee ended up. If i recall correctly yeah, so i don’t see it happening. I think intel has to go down this path, even if it’s like beating their heads against a wall.
In order to get there, even if it takes another three years, i think it’s, i think it sucks. I don’t remember the exact problem there was. There was some type of problem which is causing a lot of these performance issues.
We talked about it last week. I don’t remember exactly what it was, but i think that’s like really unfortunate, especially considering that you would kind of hope that, with the pedigree that they brought on board that they would, they would have been able to avoid something like that. But it is extremely technical concepts and it’s not like they can exactly just copy paste. What these guys have done in the past yeah, i don’t think roger kaduri, like just recreated yeah. That would be a big problem, so you can’t exactly do that so you’re gon na run into some issues it’s unfortunate, but i do agree with with linus. I think they’re gon na try to make it work um.
I hope i hope they’re gon na try to make it work. I hope you’re right yeah, i don’t know we’ll see um there’s another discussion question here. Courtesy of anthony gamers probably share the concerns around pricing rma and driver support that system integrators. Do how will this impact the role of early adopters? I think early adopters are absolutely critical to arc, and you know from talking to shroud and tap about it. They seem to understand that they’re going to have to be priced very aggressively to drive some of that early adoption, but if they don’t have the buy-in from upper management, if they don’t have, you know tens hundreds of millions, a billion bucks to just kind of get Some arc gp freaking use out there into user’s hands. Do they even really want to? If it has a lot of issues with this generation, i i think they produce them allegedly yeah.
Supposedly, they just have wafers upon wafers upon wafers of rgb and they’re going to have to make them incredibly cheap. If there is you, don’t just grind them up make new gpus, although the biggest recycling effort ever yeah, exactly don’t bury it in the desert, like uh the et video game yeah, i usually grind them up and make new ones um yeah, i uh we’ll see. I thought there was a comment earlier, but like no brand recognition, i don’t think that’s fair at all, because it’s going to have intel on it and people that are building computers, know who intel and intel’s onboard graphics to their credit, have gotten a lot yeah better.
In the last five years i mean continually they’ve gotten continuously better ever since they introduced them back. Well, i shouldn’t say: introduce them, but, okay, sorry jaden just said something in floatplane chat. If they allow v, gpu they’ll, move them yep, there’s there’s absolutely 100 strategic levers. They can pull and epos vox did a video recently talking about their av-1 encoder. I haven’t watched the video yet i do need to watch it, but i did at least make it to the headline. Apparently it’s outstanding like it is.
It is a better video encoder. So as long as obs gets support for intel’s av1 encoder, i could see a bunch of streamers just throwing it in in addition to their nvidia gpu, just so that they have av1 encoding. That would actually be pretty cool, because then you can lift a lot of the.
It hasn’t really been worth it to have like a dedicated invent card, but it i mean it could be if it’s like, actually really killer yeah. If it’s legitimately better, i mean remember too we’re talking about people that will build an entire separate machine so that they can use cpu encoding for the best possible quality. If av-1 meant you just don’t have to do that, although there are other reasons to use a separate machine like if your game crashes, it means your stream doesn’t crash et cetera, but if money is no object or if maybe it’s some money as a partial object, That that tier, if just adding one of these arc cards, for you, know 130 140 bucks getting that av-1 encoding means that your stream will just immediately look sharper. Look better. I could absolutely see people doing that yeah if they do, if that’s genuinely really good and they have vgpu support, they’ll move them. 100 they’ll move them.
I there’s a comment from from anthony saying regarding brand recognition: it’s more about the intel, hd, graphics, being poor performers and people generally seeing intel graphics as low and bare bones. That’S fair. But i do also think that intel is like a household name that people trust um and like maybe, if you’re, if you’re, very technical and you’ve like messed around with intelhd graphics, a bunch and stuff like that um, then you have that opinion. But if you’re, just someone who, like you like building your own computers, you build one every like three four years and you’ve had intel chips since you were fairly young, like i think it’s going to hit for those people and then on the other end, like we’ve.
Just been saying now, there might be specific scenarios where, with the highly technical people it just it slays and and like i don’t know, i we try to tell them here and i’m gon na try to tell them again you’ll solve. In my opinion, they’ll solve your problems. If you enable vgpu one of the issues with that is it’s gon na be hard to take it back, oh yeah, with battle mage or whatever the ones that are going to be really hard, oh yeah, so you know, i think you shouldn’t to be fair. I think you should just send it out there and just leave it out there and just um, you know drop it on the table, be like hey, we’re actually doing it. We might have cards that have issues, but we have this enabled people are going to buy it and if the competition doesn’t answer to this, we’ll just take our piece of the pie and run with it.
But i don’t know: .