285. AMD RDNA 4 is "Out for Blood", Nvidia RTX 5090, Intel B580, Zen 5 | James Prior
02:08:17 Share

This podcast has brought to by C D key offer dot com, get microsoft software for twenty five percent off with offer co broken silicon, and then also use offer broken silicon to get six percent off at silver night PC and will talk about these sponsors more later. But for now.

let's just go with the show.

And what could have broken silicon a gaming in harder box? I I am your host, tom, and my guest is someone who's coming on the show, I believe, for the third time. So anyone who has been around the channel for a while now will undoubtedly recognize you.

There are a few people when I announced you are coming on again that we're incredibly excited. And I I really do think you give really insightful like advice and ideas and input on the things going on and game in hardware. But for those who haven't listened to those episodes, ET recommend all go isn't seem to be fair.

Please, who are you? What's your work history? And I do say that because this is quite extensive actually. And what do you do now? Yeah, thanks.

So now i'm just about to start a brand new gig. I'm going to work marketing at MIPS, the processor IP company. They build compute IP for customs and look at enabled in automotive data center ebel ded system.

It's a all brand name is been around long time, started off with the first round of risk and now moving to risk five. And they're really leaning into the the big ecosystem growth. See with all the augmented use cases for eye and purpose built chips that need specific performance, specific types of applications.

But i'm going to them after spending a few years at N X, P. I was part of the product management team on the advantage processor innovation, looking at applications processes that run linux, android that kind of in multiple embedded a use cases from all a motive to industrial to I O T, A lot of different things. Um before that I was with size five working on marketing and messaging and driving the media and analyst strategy, product launches and ahead of that, the reason how we came to know each other with my work at A M D was in the class computing team from uh twenty thirteen to twenty nine.

So I was part of the team that helped define and launch rising. I was product manage thread ripper. A lot of business development for this top processes helped to define and launch the A M for platform, the S T R for platform, working with the lot of those ecosystem partners, the mother able guys, the storage, memory, power supply, cooling, all of those guys to to get that going and to bring rise into the market. So that's the last decade of my life in that show.

I think I ask you this every time you come on, but I feel like could be different answer every time. That's why I do. I mean, how do you see rising now?

I mean, it's been a while now since you've worked at A M D. I ah you loft in twenty nineteen was IT or is IT. So you you must have been I mean, that does not too long.

You must have been feeling very confident about the company then. But has the confidence grown? I mean, how do you see empty now?

Yeah I view there they're executing immensely. well. I see a lot of innovations in the technology, big fan of the three D V cash and A A lot of the technologies that bringing in um you know the technology side, the the architectures and real big fan sports that need work.

Um yeah they've got a daunting task of Frank as they chase in video, but I don't think it's any more daunting from the starting play that we were trying to chase. So um whether A M D you're gna get the the we handed to them by an video fAiling like you have, I don't know. I don't think that's gonna en.

I think it's going to be A A pure race to excEllence, which is the best thing for the market. But i'm really proud of what all former colleagues have done. They've done some amazing work. They continue to do so they seem to be continue to keep the the etho s um of building great products and doing in a way that encourages everyone to want to work with them and find reasons to do things together. This is A A more competitive approach some of their competitors take.

That's an interesting thought because all the time here, unlike earnings calls or just on its online run, T, V, you know, in video, so dominant, A M, D, they should grow. There's no where they could catch them. You would compare catching in video right now in A I, you would say that really isn't any harder than catching intel and cp, do you think IT is harder? Just to clarify. Yeah.

great way to freeze that question. I think it's harder.

I think it's harder because of the lack of commonality in the ecosystem for software, right? When I was X D six, first six x six, yet back when we had effect processes and I this when I first joined, I was trying to know work with the media to product reviews on the effects lodged, that the eighty seven ea know if the intel core was worth one hundred and an AMD call was worth ten, right? So you had that huge disparity.

I don't think that AMD that far back in A I, but the software really reframes that conversation because there are, when you think about programing on x ty six, there were ninjas that can make those machines do anything they wanted, and they did. And they were in gaming and in high and applications. And that was why you saw huge dip ties in performance between intel and and d, because that was so tuned for one architecture.

This is the case today for in video architecture and AI models, because everybody dies. But in video had the vision or the strategy to say, let's see the market to make everybody use our architecture, making the easy as possible, and then really foundationless everybody builds on that. All the models are built or in video.

So it's easiest to run on in video as you continue to, is right? And that investment is what A M D has to catch up with, not just the hardware making a great product that runs in video models, but also no software packs how to go deploy um something like an elemental voice, uh, common module, something a voice intent, voice control translation, something like that, those are software. But you really look at what invidia talk about a gtc and other things the universe is, they grandly call IT.

They'd really telling you, like, here's my solution accelerator for you to lay on top with as A A software, as a service powered by a eye, go build something great is really no different than saying, I, he's my C P. U. And yeah, you get LLM another compilers and libraries to make your stuff go fast, just tune, optimize for your model.

Just now it's all fully branded and it's got the buzot of eye around IT. So I feel special and new and different. But this is the same old way we've ever seen exploration work like direct ex or vote things like that. So um yeah they've gonna a bigger task ahead because, uh, how fast the iterations of the models are involving incredibly rapidly and they're stacking on top of each other and just got to get on the treat meal and stock going right.

So it's not so much that like intel, especially like the boats, a really piled driver versus intel days didn't look as bleak or like as big of a mountain, the climate in video as just in videos, to mix metaphors, a smarter, more entrenched mountain, right? And that's what makes .

IT harder yeah they definitely took the they learned the lesson from into intellect, everybody like and how to hold your market share and build the world garden. And everyone is taking that now to the next level. So in video have definitely built the mote um and now you've got across IT with their permission or you have to uh just build a bridge and bridge the other side doesn't want you to really so you just got to you've tt a find ways to new work around that or work with that.

So it's I don't think I I think because A N D is more of collaboration company that they will find a way to get, uh, know the other end point anchored so they can build a bridge, would IT be as big and as wide as enough to give everything across. Probably not. But that that's the game they're onta play, right? They can't force IT and the video won't get them force IT. They're taking a different strategy, I think, in terms of building with customers like if they show up and say this is a mutual benefit, great example with the recent announcement, uh, microsoft with the custom epics with the in video IP and the M V I P together right there, there's your answer of how can they catch up by working together because the customer is big enough to say we're gona work together. This is how IT goes.

This question here might come a bit out of left field but there's been rumours that and actually i've heard this was a certain degree from people i've torture you as well in retail that the blackwell laptops coming out you know in a couple months um are mostly being pared with rapture lake, not a lake because o EMS don't really want to use area lake. We'll get to that i'm sure eventually um but it's interesting because you think if you are in video, why would we not pair you know in R T X fifty seventy with tricks or or I mean honestly above actually even hot point you know our fire range or all these other A M D parts.

But there's a part of me that and I think it's kind of saying the obvious well in video wants to keep into in a certain position in laptop because if AMD were to get an entrench position in laptop, if you could cut them out, do you think that's possibly true? Because I you know I mean, ice lake first is then plus pick your poison. They're both pretty good. But right now, IT feels like in videos deliberately trying to part with intel when it's getting a little long in the tooth and laptop. Do you think that's possible?

Or how do you see that? I'd flip IT, intel's pushing incredibly hard to maintain that market share. The um I know as as a metaphor i'd say we need buy by rap to late from the IT comes detective from shipping rapped in dollar bills, right?

So they are paid to play. They wants to keep that cash cow. They want to keep the perception of leadership.

They, anna, keep the perception of om wins. There's a lot of inertia there, right? All these oim know how to work with until really, really well. So they have the the tech ical contacts is really hard to take a team and just say i'd every is now working with a complete difference of the contacts in a new different way. Those platforms, they never they never come out you raising.

So the next, there was always sideways or always behind despite the underlying technology, these needles, there's things you didn't know this stuff because because what I thought you knew that to do well that we put that in the area, you didn't read that right, all of that kind of stuff that way, working together, take time to develop. And you say, well, hasn't IT been gears since I M. D.

Entered the market on this. But he really is that he been like three laptop to CPU generations, right? Was still really early.

IT takes a long time to move over. This is so much heritage. And these leaders at these companies work together to preserve and maintain their targets. That goals. They have responsibility, ie.

S and if they can maintain cost of goods, right even, you know, because when you think of if these and laptops, I are the fastest homes in the world, they're not right there, just the most available that maybe the world best supported, because you can find someone to talk to anywhere and everyone knows how to use them. They got brand recognition and they got trust there. You know, no one built enterprise small business and buy dell laptops or H.

P. Laptops or the novel laptops and gets fired for IT right there. But if you go to the local White box, so am guy, and pick, you know, two thousand of those laptops, one bad service school there, and your toast, right? So all of that heritage is all of that.

All of that momentum takes a long time to this rut. And A N D is still pushing on that. They still doing the right thing and go into them. But as long as you have cash to burn, they're gona keep that position. They are going .

to keep that sockets, which is definite for debate, and much longer, they have cash to burn. And again, so our notes, but another thing I wanted throw IT, what do you think this idea that like strict hilo, which of course is like the forty cu sixteen core mega A P O coming out in again a couple months? Um how much of that do you think is not just a cool apple you know ultra pro like a pu that's cool for the PC market to compete with apple directly in a similar product category?

Like how much we do you think also is hundreds of trojan horse to force OEM to use radios, graphics? So i've heard this idea brought up over and over and and um i've been put out a video about this yeah but i've recently heard over the past week that I really like strict hello they really like IT like. How much of like if strict halos succeeds, is that just ams to make bigger and bigger A P S? Because it's the only way to get o, ms. To use their graphics cards.

Yeah, well, yeah, this the APP. Can we talk about this? Because I remember asking these questions, different kind away, but similar to be back in the brazil days, the line of days, I was still back as a tech journalist recovering this stuff from from your side.

So yes, IT definitely helps on the attach rate for for radio on because you can't you can't not have IT right um may make for a Better integrated ate solution. The everything is easier and simply as a commonality of reduces the amount of um reduce the amount of rework for the number of combinations. You know you are tuning software, but I still at the same time, you ve still got all of the software out there that runs really well on intel GPU and NVIDIA GPU because theyve had so much entangled market share.

So I think is not really a how do we get in video out? It's so how do we take over into how do we make sure there's no way that someone says will look, all my stuff is optimize for intel. Can you just keep put into in there, please? And that coming through in R, F, Q from government agencies, from government funded places you know, you can like how ford claim they have.

The best selling f serious trucks was because every government, and that comes out, specifies you buti four, four, one fifties and two fifties. So they just, they have a natural market that happens in the, in everywhere else as well because of the heritage don't want to have, hey, I just deployed, you know three thousand new p tops into my a new site and they are all different from the other fifteen thousand we've got. And all our apps broke.

And now all my in house teams have to rewrite reports, they have to update software, they have to tweet and turn. You don't want to to do that. And the easy way to avoid that is to say that shell use intel graphics.

They will shell use intel CPU or subs to any other I P you want so that that kind of approach is what you are trying to break out by putting in the scale of saying, we're look, you know, some of you guys use in video, some of you guys use to and they are usually doing this, so want to go top to bottom. And d, you got a tiny A P, U. You got a midsize C P U.

You got a big A P, U. And a big C, P, U. And put discrete radio on graphics in IT. Now, a scalable portfolio. Top the bottom. Any particular use case, performance point, rise point, energy requirement that you want to use? Amy has the answer and that's the that's the strategy.

Be all things to these these big enterprise and government customers and the the small and medium size businesses that have these chAllenges and the big A P U or the, uh, bigger A P, U, is that coming in really allow you to keep the a discrete attach out, which means you move to discrete attached ups to the Price point and then you get into the place where like who's really using this? okay. So it's the two hundred guys that use auto cat need those laptops yeah will find supporting them with a special set of software and whatever else.

But the rest of the fifteen thousand people in the enterprise, they can just choose the same thing for the next five, ten years. That's fine. So I I think it's it's more about kicking out in tel right so and video is theyd love y'd love more market sharing graphics um in in laptops so A M D with discrete attach right now the screen attached choice is in video but that so that .

we yeah yeah but that's where .

the weak is because of the manipulations you can do is like go, hey, well, I can I can give you a rising laptop CPU plus a radio on laptop GPU for Better combo pricing if we hit X, Y volumes over time etr. All of that negotiation goes on. And yes, i'll do something for you.

I M. D, was never in the position to do that before. He used to be with you. Get the request. Will I need the software pack? I need a special custom radio and driver just for me.

Can't do IT now there everywhere, right? So once you in that position, yes, you got the domenic m and this is how they grow IT, how to keep IT. And they and and you write right, is this is how they answer the question from the global EMS. Tear one, tear two had a compete with, and for the next indication, that of a compete with A I notebooks. This is the answer.

alright. So moving into A A category where empty basically dominated out the gate, caron cor. Rison says, was hd t as separate platform, a temporary phenomenon today? I believe h dt is filled with r nines.

I nes are organic products that are on desert platforms. And a separate platform was only really need to do to intel unwillingness to launch higher core count products. And amy s an ability to do the same with their technology at the time, because the counter argument is often expanded ability. But every time I ask for specifics, IT just seems to become clear that this is needed for a work computer. In other words, we're talking about work stations at this point, not high and dust up.

Yes, that's a great way to try and approach what the markets doing. So for me, defining thread ripper IT was never just about pure CPU. There were three aspects.

There was CPU performance, there was memory channels banned with, and then there was I O connectivity. Like what how much can you attach to IT? And the three went together, right? You would scale for a all of those things in parallel.

What we're seeing is you and four and five have taken where we initially launched thread rapper, right? First report. Sixteen years is now the high and of the A M five market, which fantastic right now.

You gotta socket IT that goes from like twenty five, thirty five watts to one hundred and eighty watts, never before seeing scale, right? That's that's amazing. I really appreciate that.

But from one eighty to two eighty or three hundred, well that's where he gets in the that verified air of high end enthusiasts top but that the work station market is below that. They want that high end. They want sixty four and ninety six hundred and twenty eight years.

Very, very few people buy those. And even in the um o EMS markets and those special systems do that, right? What they want is a platform, the scales from eight to one hundred and twenty eight.

So that means you've ta have a lower and C P, U, that still have all the I O, that still have all the memory band with. So you can you can modify this. And what this is then saying is what my writing is per core instead of her I O lane or uh memory band with and all these things.

And that's really hard to reconcile because now you get the equal then rise in product and an equal thread ripper, and you can buy some side by side. The platforms don't work. The scale was just in. This is always up.

And then people say, why would I buy that? Because this ones so much you get in the cause don't need the best because only a cause and he just goes in a circle, right it's purely circular logic. So the the I I think this is there was an initial expansion of workstation and higher test top when there was rising and thread ripper as uh rising approached into that.

And D, R, five delivered the memory bang with there was needed for sixteen years, and PCA and five delivered the P, C, A band with that was needed. Now they put the extra lanes into the A, M five platform. They really capture the the heart of that low end workstation market.

And now you're just less with special tea high and it's out there. People still want that. They need IT. They want to be able to add in multiple crazy cards like I think right now, we could see an expansion of work stations like if you had a thirty two core thread ripper a for new random number thousand dollars with a thousand dollar a motherboard, and he had a ton of P C I E by eight or by sixteen s hundred and twenty eight P C I E lanes or A I accelerator cards, not G P S, not blackwells.

Three things, just think razi .

fast S S A I accelerators for what data scientists to do, their crazy Franklin stan data scientist, things that they do, building new language models, building new small products to go into the market. Then I think those would be ganging busters um but we're not here that being unable to try. Everyone's trying to either do IT on either by a rise and sixteen core test top.

Will you go in cloud computing and do IT there? Um this there's no win between and some people would say, why would you ever own the hardware? Another people would say, where why would you ever put all your stuff in the cloud? And risky right there. There's both sides of lab IT .

kind ded to me sounds like you're saying, well, look, H D T still there, still then exist, maybe won't released as often as I used to and that's because there are some people that still wanted. But like IT is just maybe a fact to that when thread ripper really had a day, IT was the perfect market for that. It's not a perfect market anymore.

That still doesn't mean IT shouldn't exist, right? It's there. It's more age. But you know it's more need for a reason that weren't reasons before. But IT is what IT is, is kind of what i'm getting from you.

Yeah definitely way that's the right way to say. I think got the we built the moment you bringing in eight coals against four from intel, bringing in sixteen verses eight from intel, the Price for court was crazy. We created and build that market and til you know their core x product line disappeared in eighteen months, caused they just couldn't they couldn't respond, they didn't have anything and they they remain out of that.

So what do you what has happened is neon oval move to is rising, is a thrip, is an epic, epic in the workstation, I think, is where all the attention is. But because it's branded epic off, everybody's change, right? No one's looking for epic work stations because you just know inherently that's an expensive box and IT doesn't have the high CPU core frequency.

Es IT has much more is more about sustained and that really is what the A I guys and other people who would really benefit from all those things are looking for plus the reliability that wrote management. And you can you start talking yourself into just buying cloud time anyway unless you're deeply worried about what you're doing on your machine being. Lost or no access, shared exception for privacy of security. So I I think .

to my memory, this was something we talked about the last time you're on as well. Like is there any argument to be had? And I and I remember making this argument because at hard wearing box, they did this testing where they were like, I think I saw this at some other youtube s channels as well, that the issue was that thread ripper was becoming so needed. The software support wasn't even there.

And so there's an argument shouting there, just be a ninety class motherboard for A M five for the extra I O and then work station just look, if you want the extra, just buy the work station, dude and will let you over clock IT or something like is there a chance that you think because I still find IT odd that they have both, that there's both work station and iron this top right now. And I think most of the people that would want hd t really want rise into cost fifty percent more or twice as much that five times more because they they don't need ninety six cores. Like I think that there is something we kind of agreed on last time.

You are all but like I just bring IT up again. I mean, there's rumors is coming out now about five threat report. Of course, that does actually go to ninety six cores and use you like the epic soccer basically. But like do you still think there's an argument that like maybe there should just be a boat to be now called A Z ninety extreme motherboard with db, the I O and then work station should just let you over clock and call today. So there's two platforms, more support yeah .

that that would be my approach. I would love to see more, more I O on these platforms. Um definitely what I would like to see happen would be along those lines, either expand a and five four could tail the epic platform to.

I don't know yes .

exactly just stay IT back. Give you you want the um an epic light call IT a thread ripper, enable the boost, give you some of the I O all dependent on what Price point you're targeting and and get to the server features. So I don't have to buy a register.

Dims eeta amuse on buffet to E, C. Dims and and C, U. Dims eeta that that i'll gave you the stuff to get like get to the south twa point there is um we found where scaling rose off right sixteen to thirty two is where the R Y um it's really nation one of times and server they don't .

want more than thirty two course to the day.

You have to really go programming and get really wide. Then you go what actually I need multiple sockets, and I talk about scaling this across rax. And if from that parallel one are using, you know, the really parallel celera ors from GPU, this IT IT really bounces around in in terms of, like, what should you be trying to do in until someone builds IT and tries IT and they, they, they don't know what IT actually is.

But if you ask AMD or intel or in video, how should I approach this parallel problem? They're not going to say by my cheap CPU, by my big expensive GPU. Yeah, right.

Because they they built them for exactly that purpose, and that's where they are. Software efforts are focused. So yeah, the ponds of issues in there, I think, is a healthy market.

And I would I don't know how to size. IT was multibillion back in the day. It's probably still multibillion. But looking at the way, uh, everything else is being prioritized, that's you to say these places .

are mobile. A hundred billion right now.

The ri just doesn't work for these .

companies, right? Well, so we touched on in a little bit, but I ve got actually get into this piece of content is brought to you by CD key offer dot com just in the past few weeks, then built a new PC with a fresh install of windows eleven on a blazing fast gen five and B M S S D. And of course, that code for windows eleven came from C D key, offer a dot com, and we've been using CD key offered that comment.

Mos laws that for very long time I use IT for my office software on both my desktop and my laptop. And that's because they reliable theyve always been there for us and they offer reasonable Prices for my microsoft products that are usually set, in my opinion, at monopolistic pricing. And you know what, they actually have a black friday sale right now.

So if you're looking to get the best pricing on microsoft software or games or other products on their website as well, please show support for a company that shown support for the mos law is dead team in community for years by using offer code, broken silicon to get twenty five percent of all microsoft software and dashing to get three percent of everything else. Again, you're supporting the channel even if you just click on the links in the description. But of course, also if you need these products getting them through C, D, key off dot com during their big black friday sale, well, that helps the channel even more to support mores laid by checking out C, D key offered dot com today.

Waiting about and five. I mean, the initial launch was undoubtedly ssi, shall we say? I would say it's I mean, while I I don't have to say and I can just look at the facts, you look at the sales and five seems to be deemed itself like that.

I mean, they fix the prize. They fix the performance issues. L said, hold my beer or make you look even Better with a like and now and five is just selling well.

But like how do you feel about like just in general, how AMD handled the and five lives? Because just kind of setting the table, I really don't get up to my memory, its arty years of blur like I believe I launched in August. why? Why not just launched in september? I wasn't ready for selling well.

Why did we just give me a month, make IT finished? Like why was IT law? I I wonder if you have any thoughts on how you would have handle that lodge and if you have changed how they segmented things and where comes a segmentation? I think they should have at least announced the hundred eight hundred x three days so that every review said, don't worry though this one isn't for gaming but anyways.

please go on well, you just answered why they didn't do that, right? So until they launch the idea.

hundred and thirty anyways. And the same thing happened was for everyone waited for extra ity. Once extra came out, not extra.

Is all Better too, because people know there. fine. Yeah.

you got, you got convinced. N, B, A, that's what's gonna happen, right? They're going to say that pretty could happen.

Specially, is there a chance that we can earn more money by not doing that? And you have to say, yes, this is a chance. And then where I always take the chance, so is the zero m your argument.

You don't you don't lose anything from IT. Personally, I like this in five line up, uh mostly because i'm going to upgrade from around two thousand and three thousand rise on and off intel at the house. Yeah rise in five is A A big shining, glorious moment for me.

IT wasn't a great bunch, but you know those hard to do these days with um the way the market is like everybody is super Price. Everyone is looking for every little thing to criticize because of the high Prices, because of the availability. Um that's that's that's the nature of the market now.

So um do you know the G P U cycles are soaking up all the C P U. And mother board money? So trying to increase A S P and trying to maintain margins is really tough. Um but no you you don't know that until you do IT really reminds me of um I was when we were in santa go and they were doing this this hotel, we had this a big ballroom that we brought all the media and tech press into to do the presentations of pitch on this stuff.

And the day of IT, it's seven thirty the morning walk into the elevated to go down to the the ballroom and least do is I have to be on the same floor. SHE walks around the corner and she's to say, how i'm gray, how are you and SHE looks like he says, are you ready? And I answer that straight to you as a casual question.

Yeah, I think so. He said, what he is, the thing doesn't matter with your radio. Or the moment is here.

And now we have to go do IT. And we just go to keep that in mind. And moment is here, go get IT done.

And then we will figure out what we should do next will Better next time. But today, go. And he wasn't being like critical SHE was he was coaching, encouraging something to stuck with me. Yeah, today is the day.

So you you pick a time and you say, this is IT and you go and IT does a lot of this stuff know you can get too much into the competitive story, right? And and is too much in the game's ship. And I really think that that's where intellect struggling right now.

You're in the game's ship. They are in the in their heads around that, and they should be telling their own story, why they great, why they did things, what they were trying to achieve and how they are gone to get there. If they stuck to that, they get the belief in the permission to do things from.

But because you know, they they're coming around trying to do the competitive thing, then they have this issue. So I M D, I think they they um fixed fast what they needed to fix in a market adjustments that shows they listen to the market that they're watching what everybody is saying if the is a really good reaction, was that is that something they should have had to have done. I mean, every company, this public you trade has a fishy duty to maximize return to their shareholders, right? That this laws of this stuff, so they all try to to do that.

So everybody in the manage, when everybody on the team is incentivised to try and achieve the most increased of profits, revenue, the targets, all of that stuff um and every once in a while, they get too aggressive and they have to call me back. And IT can be for internal reasons because you didn't know know the messaging, I didn't know the product. You can be for external reasons, the market shifted between when you decided what you're gonna do and how IT came out.

And IT can be combination of both. So um no, I don't know that I would have been able to manage to stand and say I fixed that, that you would never have on my watch, rising one launch happened on my watching. We had to take over time getting that thing going with the memory and all other things. Yeah, people compare .

this launch to and one in some ways you can. And it's the biggest ry design of the architecture since then. One, but at the same time, I mean, I don't know then one, even with this IT, IT wasn't even close.

I mean, he had six cores competed, twelve threads competing with a quad core quad thread. I five IT wasn't I still think with the issues, right? When that happened, you didn't have and and you know, i've heard people say this to maybe because pile driver was so much weaker IT couldn't to look course, no matter how an optimized was.

But there were couple games at laws were like then five was like a thirty percent regression from then four it's been fixed. But like that still is. Like I and this leads me to, i've heard people there, there's a contingent of mos law's dead fans that I see in the discord, say this quite often at a lot of recent b launches, which is the kind of make fun of a amb.

And go, here we go. Amb is gona launch everything for twenty percent more, they should. And then it's gna need a patch and then they going to drop Prices thirty percent instead of just maybe launching IT two weeks later, three weeks later at the Price that would have been and maybe they could have even held a higher Price longer term.

And I I feel like, you know, in some ways it's good to show of water, but another ways that really seems like would then for and to ten, five, I feel like we almost saw repeat of the same mistakes. And and i've heard people like make this argument that like all this really costing md. And on the one hand, I think IT might have.

On the other hand, I look at their sales in market year and I go they seem to be doing okay. How much of that does affect M D, in your opinion? Because IT does seem like aimed his opinion is let's see what we can get away with in the less you Price drops. Then he got ta wonder, even if you would have nailed at first, would you have not felt pressure to drop Prices as much? Or how much is this really matter overall to like desktop sales because it's not just do IT yourself as everything, how much you think that actually matters every generation?

Yeah, that's that's a great question. So that the chAllenge is, is, is no bell, whether there is no benchmark for this, right? They set the benchmark every single time because there's one other company trying to do the same thing and they're not doing IT well.

So who do they how do they model this? How do you walk game this out with real data to say this is gonna the impact of each other, right? T shirt sales, you can do that. Console sales, you can do that. Phone sales, you got multiple competitors in there so you can watch different strategies.

And the Price place station, the x box goes here.

right? Yeah you know but when you you bringing in something that's redefining performance of a dollar in the category, how do you value that, right? You know you anna, go up. You know you anna, deliver still more to the customer and you don't want to upset them, but you also need to fund everything else you try to like all year.

Remember that A I investment we've to do, oh, you got ten all these multiple things pulling at IT in all different ways and you're never gonna find an executive is going to say, uh, it's not okay to do a Price drop after thirty days when we see what the market reaction is because you can't go up like right because that's the other option of IT is like what I launched at a ten percent lower than the last generation. And if the sales are really good, then will call that intro pricing and raise IT ten percent and then will keep raising IT until, uh, the sales roll off. Like imagine the community reaction to that pricing strategy.

There are companies that do there. There are our suppliers, technology and others that have intro pricing. And after a few years, IT goes up when you think, if you blank age, be cheaper later on.

But I feel like x in place station have been trying to do that a little bit.

Yes, right. Um so you there there's all these different ways. So is is a hot thing to say. You could just nail IT if because that's one person's opinion of what nAiling IT is, you can ask five other guys, you get five other nail IT and someone is going to make a call and they're always going to earn on the side of 服务。 This is the bigger number Better.

So i'm gonna that and that that's how that really short circuit seo IT until you get someone working and say like hi this is you you just missed an audience and and then A A M D look into all the different communities now there not ignoring the communities. Different communities have different um waiting internally. Feed back is all capture and they hear IT.

They just not able to respond to everybody's request same way. You know you think about why there's no thread ripper um replacement yet, right? Not because people are asking for IT, because not enough the right people are asking for IT. Or is IT because they've got example dollars and they can do these three things, and this is number four, right? As wonderful, beautiful as number would be one, two and three type priority and that is what IT is.

Um actually I want to ask you a couple C P U segmentation questions here though PC dog rights and I said should aim to have released the the resent five exert products first and then released their standard products. My personal opinion is that should have been ninety eight hundred x third and then maybe ninety seven hundred and ninety six hundred like non x and then you can roll out of the things if you need to or something like that.

But I think just in general, look at how much of a home around the eight hundred x yes, good lord. It's it's just ridiculous as well selling especially now that I can clock to I mean the multi core performances closer to the ninety nine hundred x than the ninety seven hundred x because of how the alka ard boost clock are. Do you think it's an obviously love this.

What have to do with what win intel is launching and if they can afford to wait. But I would then six, for example, should they just be doing all v cash and then the non x is the non v cash is like I think we're getting close to making the argument. I'm not sure why you have more than a few non v cash miles. Obviously there should be a few. But why arent most of the models v cash at this point?

yeah. I mean, that's a product cost question though. The extra caused to catch the packaging, the testing, the integration, the development, all of that stuff.

There's there's a fundament requirement to build a three d the v cash product and its having the uh original product working and out ready to go, right? So what you're saying is to the being counted right is like a delay all of that revenue you we're going to get for that quarter and wait until this point over here. And you know everybody gets heartbroken over that.

So personally, what I would love to see is the return of the full stacks launch drop, like is the nine C P U from rise and nine to rising three the three tips at the mother's body of system, everything together um that's really hard to coordinate and do you do end up with things getting overlook with the x three the be heralded as the gaming champ if they'd launched them all together? Um like we would have everybody had the time to release, study and learn IT about focus is about storytelling. It's about uh but new enabling the market as quick as possible.

Um I think in the term you going to your segmentation question exes or x three days and non excess of the non v cash, that makes a lot of sense. I think that probably you would want to go to a different brand, I think because you've had yes, so many generations now of x is on these different parts for higher power. You know, he used to mean that IT was a nine five, what CPU and non x was a sixty five six.

And then that went away. And now that there's it's the best in that segment and you have lower versions. But those kind of things, uh, you've trained a lot of people, you know, from retail sales people and best buy through to the selling agents, the distributors and at the ms and the people who write all the marketing copy downstream from AMD in the non climb component channel.

They are all trained and iterating on those things. And if you disrupt that, then is this like relying ching so far we launching 3DV cash to know to say like it's is the best C P U of the rising line， then maybe that would be the way that you would get rise. And ultras, but I am nearly throw up in my protest and that I don't .

really see that. Yeah but you know we probably will you .

know someone is going to take that suggestion now maybe they already thinking, I don't know what IT is. I I think there's a strong case we made now. They fixed uh, the ability to clock up, up to stop this huge performance increase, not just gaming workloads, but several others.

And now they fixed the launch issues and stuff. And that's the other thing is like if you're going back and changing and saying where we do nine, eight hundred and three, the first imagine that with the thirty percent performance drop in some titles that kills that tech really fast, right? Don't want that yet.

We don't launch IT is ready. But but yeah but you know how do .

you find those things out, right? Like you watch that launch and they said we were using this version of windows that has all the fixes, and we didn't know that wasn't coming until this point of time.

That argument to never really knew when you've the effect. So supposed .

to so you need a lot of stars to a line. And IT is a simple logistical problem on paper. But in actuality, IT takes a lot more intense coordination that is really feasible to to get done, which sounds absolutely crazy for the companies that are valued in the trillions of dollars that they can't talk to each other way I have to corporate when I launching my C, P, U.

And I need this version of windows to be available. Can we line this up? Yes, no, right. Is there so many things are going on in that because these ammonius products and trying to get two different companies lined up on one on that introduction has always, always been too hard.

Yeah, I just to finish my thoughts. And I mean, I just don't get why you couldn't do like I don't know, like the ninety nine hundred x, which would have you know, really made, at least done pretty well. I think wherever I was in ninety seven hundred and ninety six hundred, boom, that'll get the bugs out.

And then the ones we know everyone's excited about, nine nine fifty x, which I think could still exist next to an extra version because there are some people that just want all sixteen course of the highest cks. And then in nine and fifty x 2d and ninety eight hundred extra and ninety six hundred and boom， that's the second wave. Once everything fixed.

I just don't see why that's not something that could be done out is a bit but to shoot water at this time. But I can't help but wonder if it's time to just do that with the sex. Yeah.

that's on. Yeah, yeah, yeah. I I would like to see using the the bottle numbers a bit more, you know, like the ninety nine fifty x right? No D V cash.

Ninety nine seventy is the three D V cash. So I don't need to look into what does the x mean and how White to flimm IT. Although I just know a bigger number Better, right? I got twenty more rising points, so I got more performance somewhere. That's that's the difference.

But then wasn't that technically mean, it's supposed to be thirty two course because there's seventy at the end of IT and that's how I was named with thread ripper. I don't know, maybe fifty five x three .

day or something well um pretty my brand in but I did put a thread. Seventy doesn't have to mean thirty two calls right in seventy.

Everyone loves the number seven, fifty, five hundred is posed.

Yeah so this ah this things you could do um I don't know maybe it's ninety nine fifty x is the non x three and ninety nine five C X is the cash, right? You can there's a number different ways you can get cut with this thing. Um there's not a way that everybody he's going to like because of you like you know I just told you that you give me a perfect example how everybody is trained on what the numbers mean.

That's why you can't break IT right? Seventy means studied too cool can do that. Oh.

but I said that and then they launched this thing called the fifty one hundred X T. And apparently IT has sixteen hours and not twelve and I didn't even realize that I had sixteen hours until I saw the reviews. I actually just my brain ignored part of the slide that said sixteen course. I just as I go yeah .

it's got me to so IT this is fantastic so yes, maybe it's a nine nine and forty five or something then you go, wow, that's even Better even spice I love IT is the every Perry chicken in the door I don't know um zander k rights .

and this is is is possible that empty in a forty years move launch then five early knowing that its mediocre performance with lure intel into released an airlink also early and unfinished as they did just in then released the ninety hundred one hundred xd to already destroy til once in for all in desk gaming because I think this is probably a mostly pacifist question um here.

But I I wonder what you think about that too, because I also don't know why area lake launched ed before I was done. Like so it's like they both did IT. So there's something obviously, the three line is microsoft, but still, I wonder what you think about airline launched as well.

Yeah, our lake is they definitely didn't go when they already they were going because they made a commitment to specific people and that they have to deliver on that commitment. They were living through the hate. The day is here.

Let's go make the best of IT um for which you good for for get out there. But that one that one's gna hurt for a while, right? So they got to repair that.

Um there are there are um cycles to buying right that you you GTA be on shelf for uh the holiday season. The biggest buying cycle. The way to do that is to launch in q 3， right if q three launches A Q four under under the tree around the the table. The holiday present got to that time to um yet the patches in get out and flow through the ecosystem everybodies aware of IT and is still the new hotness so that that's why they had to purse because everybody else has you released the CPU into the component channel and started shipping to O E M O N S. A system integrator probably needs fixed to twelve weeks depending on how much things you changed on your platform and you're everything else to build a uh a PC around IT and get that on shelf.

But I know em needs three months, right the four month to be able to build designs, test them and integrate them, get them ready, get all the documentation done, translated in the twenty two languages, train the support staff ah and everything else, build their supply chain and have that ready to go so that they can do that. So if you look at when when did everybody launch and when is the the buying cycle, that's why, right? um. Do the dmd hope to stick a finger until I absolutely he always yeah .

I think they did. They just hope they stuck .

in harder, but I think double fingering point. But that's because intel didn't put their hand up and and get their products out in this more and fast enough way, right? So it's it's they're still going to have products out there and people going to buy IT.

So they showed up. They have deliver on the commitments. Um but yeah, I don't think anybody super happy yet know about that. And that is now the good, honest work of ramping the design win, solving the problems, the data day, blocking a tackling of you, how do we fix former issues, software regressions, performance improvement, key applications, little glee, the little nickles you always find out about, uh, for corner cases of engineering as people build weird and wonderful things and try things you never thought of when you are first designing IT.

Yeah, I mean, but here's the thing that I look at a amazon CPU best sellers. And I got a wrong last time I showed a picture and I was technically was forty nine. I didn't even notice, or maybe I changed last minute when I took the screen.

But I do not see a single area like us. Q on amazon in the top fifty. And when I talk to an om about guys like I see some of these reports that you guys are going to want to black right after that.

What's that about? And I want to say which is, of course, for obvious reasons, but this person told me, uh, oh well, we don't really like IT. We're hoping to just buy more media lake because it's cheaper and hope that customer see ultra and think that means ultra.

So that's how area lake has landed so far. And I know that they're claiming that there is going be some patches here, but I don't know I don't know what you really have to say about that, but I not I am really concerned about intel at this point after this early launch. Like i've got top fifty C P S being sold like the top.

Let's see here, thirteen of the top fourteen on amazon are rising and then irla isn't even in the top fifty. I think half of that is that it's not a lot of supply right now, to be honest. And when I talk to OEM, they're like rap, like a media, like actually like I I not sure.

I mean, it's very sobering thought is if if if the ninety one hundred x three like twenty fitness, say fifteen to thirty percent faster, then ara can gaming. I think panda like might be ten percent faster, fifteen percent faster. okay. But so it's conceivable though that until would take two, three generations to catch up to A M D and game in. I just I don't even know if I have a question, have been very reaction to this and I know it's intent subject, but it's like I don't I think this is worse than one five though I think this is a pretty deep that's been dug.

Well, the trying to fix a lot of things is right. Like if you look back on the last couple of days top C P U on this number one theme that came out from, uh, everybody evaluating into was okay. You're pretty close from the performance.

There's a few wins in here, but holy cow, the power, the temperature, the heat, the is know, this is untenable. You can't keep going around this when you gona do something about IT. So they reacted and they did something about IT and they lost some performance, but they fixed the things that everybody screamed at them about, but they didn't do.

They didn't keep everything else. So the the most technology aware consumers are the DIY PC component buyers, right? The people that build himself, you you understand what a rap lake is, what a media lake is, what A R lake is.

And so that's why you see them not in the top fifty because the the preference isn't there. You go down to this. The people that buy prebuilt pcs, that's where I think you will start to see arrow taking.

Some some sports are still not gonna number one by long show the comments from the O E M of saying, yeah, we want meteor and and raptor is code for you want too much money for the new staff and i'm not gonna a pay IT. So this, you know this this is pricing to performance, power to performance. All of that the they're stumbled on this one, they're picked themselves up, does themselves off and continue going on.

You like if you if you think back to how when the first effects bodow are launched compared to the intel processes, I think the a narrow lake is in a wait at a position relative to five. Then that effect was to the those those old intel CPU. There was the far close room performance, fast close room.

The features are there. They've got the ecosystem. I don't think there are anywhere near a dira as a situation and there is actually a playbook on how to turn around A C, P, U.

Company and make them successful. It's written by lesa. So so they do have examples in front of them of like how do we fix this? And uh I I think that they will figure this out.

It's going to take him longer than they've said. They don't know that because they haven't done IT, and that's that's there. But they need to they need to stop focusing like they are in the moment of they need to do less with more focus.

And really, they're just been trying to to shrink into different areas, but they're not increase in the focus. And now they really go to increase the focus. So it's gona come down to some really articles internally.

What is important, what is the priority. And they going to are on data center, on A I X R raters and on notebooks and client stop is gonna be false on that list. And anytime there's enough money to fund three or four, you gonna see that trick out as this way. It's just the nature of the business.

So again, the the A I this is the by products of the A I revolution and and chase that's going on sucking all of the air out of the room and really impacting the deaf top uh enthusiast because rule sit here going will look CPU cost a lot more than they did. Mother's costs a lot more. Memory costs a lot more.

Graphic cards are ridiculous. Ly more and what they were is not just the inflation IT is Price uh Price increases as well for yet the performance podolia occurs. Wake on so that's that's the chAllenge, right?

Um you're seeing a lot of companies calling softness in the global markets for consumer goods in IoT in in computing. But uh, you so these companies are not onna keep throwing money into soft markets. They're onna go after markets. They're expanding, especially when they're in diagonal and being and wall street is me making up rumors of who's going to acquire them and how like. Well.

speaking of throwing money into a softening market arc, let's talk about intel archer for a second because I just put out of video today is really kind of like a press like I heard wild rumors about that we all know basically a month o that if if not longer R D N A four and R D N A R D A four and black well, we're going to be announced c and baba blaw um and then all these rumors I wear in the last three way oh no.

It's going to be sooner and I put out a video like no c but part of that was also some surprise stuff I got about not really surprised me about battle age of seems like the top card, which have already cancelled the really top one, but the the range one really um that was still supposed to come out, hasn't taped out yet. It's problem. It's supposed to come out now a quarter, one quarter to those questions of if that would even be competitive with R D A four in in the artist, fifty, seventy at that point.

Um but like where do you see are going because it's the information I got over the past week is really just lines up with what i've been here in for a couple years now, which is that it's just too little too late. They need to get core back on track. They need to get back on track and putting a lot of money into I mean, what IT sounds like the b five eighty that supposedly get a paper launch in december IT kinder sounds like a seventy six hundred X T twelve.

Gaby, which is like, well, seven six hundred X T at sixteen launched a year ago and use a six nanometer. Are we really onna pour all this money into a foreign animator, competitor to a six and enemy to product? Like when is he going to stop? I wonder what you think about that?

Yeah I mean, fundamentally, I I enjoy more competition in markets. It's going to be real competition. I think AMD worry about that competition more than in video a in the test top GPU market.

So it's sad but not unpredictable for ar C2Be in the sit uation tha t i'm dis appointed thi ng can be her AMD, more than at her in video. But the execution struggles, you see them everywhere. Some of IT is as business is as usual, because building complex deep uses incredibly hard, and they will now know that.

And you know if if there was funding for the top three things and the top CPU was four, then know best top G, P, U at eight. So that that prioritization continues to to happen. You um I think there's no there's gonna be fans and enthusiast that vocal that would say I M glad I have IT.

There's gonna be um some people who have swayed by the all intel story. O I can have an intel CPU and G P U. And I can move from integrated intel graphics to direct intel graphics.

And that simplifies things for me. You'll be able to find those random people that do that um A M D do trying the same thing. So you know this is in the hubris pathos.

They navigate the the turnaround and reinvent themselves. So this they've walked away from IT and finishing out on the commitment. So what's in flight to maximize the R A Y on the investment for every single dollar they can get without spending additional dollars and throwing good money after bad.

So I personally these situations, I think you should invest in building the best products possible um versus is going through the motions and that you know because when you when you're in a tough situation, soft Marks you won't come edited. Building a well product, even if it's just one, is a Better thing to do for your company. Then IT is to just go and we'll see what happens and not really care about IT.

We're going compete from top to bottom, and none of it's gonna actually .

be that competitive, right? I mean, if that happens, you can go out there and say what you want to do and then when you come, but you gotta be real about where you are when you do IT and say, okay, when next lesson is he doesn't feel like they've had an opportunity to take that next lesson, but we know watching who's moving where and how they're reorganizing this. There's no, no continuation of this activity.

I think the market knows that I think really struggle. People launch IT to say they delivered on a commitment, a few cards out there, whatever we can stay, or what might have been, this would have been fantastic, if only. And then they will fade away, but they will continue to invest in the integrated graphics. Pro map continues on in that fashion.

And because A M B are really blazing a trail there for them by putting the bigger A P use, right, because they then giving intel room to be, room to breathe as well, and say, like I will, you can go Better than just an intel tiny GPU that can light up three displays and play video on one of them and were dogs and excell on the other two into or well now I can do some richer more music experiences. But it's um so they're going to get some continuation there. But I just create graphics.

I think that they prove they don't know how to into great well into the aib system. They don't know how to play with the big game developers on triple a titles, although I will obviate that and say that, that will help them really well in integrated graphs. Gaming, right? Notebook gaming, that is still something people want to do.

So this, this, you keep epm. Those titles tripled or not working on those A P. U. Is gonna critical.

Yeah, yeah. I mean, yeah, it's just in a thing I keep coming back to in the past few months like this argument of like people go in, oh, I hope they just keep trying to compete in the highland of gp s and I like get how much money in video spending on R N D for gp s at this point, it's like you're already starting like three steps behind and you've got the other people like inventing, you know, jet packs to go faster than you.

You like it's like the idea, like A A escape losy you miss the escape ilo. Sy, like they are just accelerating and accelerating one point, one point, one point one way year, one point zero five times one point zero five. It's just gonna grow away from each other at this point.

And I think you've gotta just do what you can do to keep the foundation alive so that when you have the money, you can try again. And anything else, we just for any of thing else that much more important to them that they could be spending money on this piece of content has brought you by server night pcs. They are very knowed american company based in fit fill.

North CarOlina's offer everything from enterprise and business services to individual PC parts, including collectors items that are hard get to prevail pcs to excEllent repair services. And of course, I can test to that last one. Personally, when you in this summer, I don't know where my PC, the one I usually used for making all of this content to shut off and will not turn on, and did the usual trouble shooting of checking, you know, the GPU RAM motherboard.

And eventually around four A M, I had to make a judgment call. Was I really about to strip down the PC from top to bottom, check every part another PC, and then try to reinstall windows, which undoubtedly mean I don't get any sleep and might delay the broken and silicon on from coming out? Or was I just going to get my, you know hot point tablet and call for help when I chose the latter? And i'm glad I did, especially because I called silver er night pcs.

They're me a shipping label within you know hours in the middle of the nighter really morning at that point and they got to work as quickly as possible. And i've got that back by now. I'm using IT right now to make this ad. And really, you should check out the other custom PC they make as well.

They're very easy to get them to make one for youtube because they excEllent bill deal that takes the guest work out of everything you tell them what types of games you even play in addition to the performance in the budget you have or what you're looking for. And they customize and act as expert consultation for you to make the perfect PC just for you. And they give good advice, by the way, i've talked to them, they play PC games.

They are enthusiast. They're not going to do something silly like an OEM would do and put an R T X thirty, fifty, sixty about with a non K I nine just so they can clam there's I nine and R T X graphics included. No, they are actual enthusiast that will give you an actual enthusiast PC like me or any expert would really advise.

And I think of the day this company supported me. And so I would be honored if all of you watching this, if you want some of these types of products, if you supported them in return, which also helps me get even Better support and deals for them in the future. If you do so, one more time support more is laws dead by supporting several ipc today.

But I want do you think about this because I I believe the last time you were on, we were just starting to talk about I think yeah like that big r DNA for was cancelled um and that amy is going for a range. There is a lot of catastrophes about that, although I know you've obviously little of me, but you've much longer than we have been around line up to know that Frankly, every two to three genes emd does not go for the ultra high end. Like this is a pattern whether like high and high and mid range, high and high and ID range, like they do IT all the time.

Six thousand series, arguably, certainly polaris. And three, eight, seventy, I mean, this is five, eight, seventy, was just so good at managed to beat actually in videos s high on carting. I don't think they are really trying to like you talk about, they do one products really well.

Do you think that could be like the eighty eight hundred X T is because IT really does sound like it's going to be seven, nine hundred X T, hundred X T, X rater, surprisingly, maybe forty, seventy. T, I, super red racing. If you could just hit that five hundred or something. I just think that would makes so much more a splash than losing to the fifty ninety by twenty percent. You know what I mean?

yes. Yeah no, I I totally agree and I think that's that's what I want to say. I'm reminding I reminded of the old I think that was ever Green architecture, that radio hd forty eight, fifty, forty eight, seventy against ferme ferme was different class performance, different class power.

You that was where all the leaf blower memes came from but amy really knew that with the forty eight seven and forty eight fifty in terms of performs per dollar and the efficiency and you they didn't and you go in the forms people and ask like, hey, what are tui by when the answer is money no object you buy for me and but if you care about how much money you we only have a limited budget then its A M D all over the rest away and the market share reflected that, right? So go for the meat of the market, go for IT. Uh I personally have any appetite for more thousand dollar plus GPU no matter how fast they are.

So um yeah if if the new cards are sub seven hundred bucks and they are winning in the performance per dollar range and they give me that the great gaming experience that I want, they move the table forward in terms of, uh, you know people were not upgrade from thirty seventy of forty seventy and those are going that's going to be a clean win. I think frame d and I think that something to be proud of no matter what the number on IT is, whether it's eighty eight hundred and eighty nine fifty X T X uber, wow cat I doesn't matter right this going off to that hard as the market um getting back to your foundation, making your business um unit profitable and well respected and delivering excEllent execution. Building back trust then gives you permission to come back and say, well this maybe go for the high and again but likely said the the difference between in video o's investment in GPU everybody else is whole companies now just immense levels of investment.

So if you're not to a if you not gonna go do that same level investment, why would you expect to have the same level of results? And that's it's it's good. I don't know. But a lot of people .

on red IT seem to think you can just map your fingers and have the exact same thing with a quarter of the budget for some reason.

right? There is not even a quarter a than that though. Just looking the size of of aid graphics budget versus in videos is that's not a fair comparison. And I think this is we start see the tight turn on AMD as the plucky underdog upstart come and get there and running up the steps to cheer in front of the statue a moment.

Is the average like them? And you're trillion dollar company, won't you have resources for this kind you to fix this? What is going on over here? One, where you head nothing? There's a little bit of that in the launch.

Two, right? How do you mean you couldn't figure this out? How do you not able to do this? So um the best one in the world, I would love to see AMD competing the high end of the G P U.

The investment is not there. It's not been there in the financial statements. So you can see that it's not been done. Where's IT moved to a eye data center? I think that that um a reset moment, a new architecture can unify some of these things like we haven't seen in videos. Gaming performance hurt by their investment in A I right introducing tensor and all the other associated a developers in their S O C to support ai have not impacted gaming in a negative way. So they found .

a way to make IT work for gaming, even if they had to find a way they did. They found a way.

they found a way. So now AMD has to find the way to, okay, let's go focus on A I build this migration GPU. But then you can see that then there's a moment down the road map where that trickles and kicks in again. So i'm not saying this is a play hold launch. I think that the a empty graphic guys are absolutely meaning to winning gaming at the mid range, and they are deadly serious about IT, but that there is a future benefit to come through here with the architecture advancement says they redesigned and redeploy and refocus yeah .

and I think this is something that also come over the past six months of broken and silicon. A lot like we've always talked about like, oh, a md only has so much money and they're to prioritize what they spend money on. And then when you look at a high end or say, an enthusiastic, really tear GPU, and the issue always had about hearing about this idea of navy for sea, which is, of course, like the twelve chip let insanity they were legitimately considering launching for our DNA for G, I lead a picture of IT just like not.

But I like there's no way that things like less than 4， twenty five hundred dollars and he looks like an M I three hundred and you know that say you beat the fifty ninety by twenty percent, but IT didn't have equivalent upscaling tech. IT didn't have equivalent a lot of these other features that in video has, I think in video with their brand recognition, can maybe just get away with charging the same Price for something that cost less to make. And until amy is the resources to support at that level, at the top level, IT might make more sense to make less and support IT Better, which is an think, an interesting argument that I am really become convinced of, as if they don't do like two, three, four dies like r dina, two is, I think for different configurations. Okay, what are going to do is just to their monolithic will nail the software. We're frame up resources to just nail these tears and it's not really worth going like maybe learnt from mentals mistakes going for the top unless you have everything fixed on the sides.

right? Yeah absolutely to about expectations. It's about like everything you just described and then that comes back into your big A P U.

Story, right? Like you used to be, the AMD would launch five different chips, the cover from fifty dollars to five hundred dollar Price points, which suggest now it's impossible to do. You can can find find chips to cover those Price points.

It's got ta be one or two. So that bottom really relieves the pressure of that in to do that with the A P U story. The high end being so high now that it's really I don't know, it's workstations, it's commerce. I mean, I have forty ninety.

The performance is honestly ridiculous. Like how many people really need that much performance? I know someone's going to say in the comments, I can barely do four k sixteen. I don't know what settings are using. I do, but I can do quite more than four k sixty with reasonable settings.

Yes, absolutely. There is always give me some guy can show you where you mr. Brush roke to paint the corner but the the those the forty nine customer is is a program .

dispose income .

for why exists and why they they get free beats, right? I don't know what IT is, but there's everybody else is not buying those things. You the market share, we don't ever see ninety cards leading on the steam hardware survey, right? What is that? Is because I bad IT gaming? No, because the number of people who buy them is just so small. But you so if sorrow why you gotten get the reputation, once they get what's they nail a great products with the reputation and and they know the software chAllenges, then they will be able to move into that high and competitive again. So it's a reset moment and reabsorb, I think and my .

wrong and I feel like you are leading to its kind of saying this like you kind of hinted with the word like stalling in or buying time due. This is a theory I of the u DNA is something that could allow them to much more effectively compete in high end again, because it's hard to justify a twenty five hundred doa graphics card of all it's used for his gaming. I think we see the G, P, U.

Market is in a downturn right now. But in video, can sell them all into A I, amb. Can't is easily were they to be able to sell something one hundred ex with a different name and forty eight gig bites of RAM as you know, something they can self for ten gram like a video can I mean, it's so easy to justify. So I I wonder you think like if u DNA because i've ve been a lot of people like why do they do u DNA and it's like, I think that the only way they can justify making high and graphics guards again yeah I mean.

this is a recent repeat of the strategy, right? The first explain die was for epic and for rising. And the um reuse really enable us to have the the two market approach back then. So look at IT now.

What are in video doing well? They take an architecture that they put into high and server products and then the d feature d wiss l making to professional grey cards and then take IT down again, making into consumer cards. But it's really all the same thing right now.

To compete with that, you have to follow the same strategy. And you know if it's you deny whatever that is comes in and is fantastic for the A I strategy in the data center as a attach. A card for, you know, given to the instant product lies like this, N M I, four hundred, five hundred, whatever IT comes and IT goes into instinct, and then IT goes into radio for gaming.

And this minimal changes, knee between the three, like you could module clock enabled, cause a memory attached then wow, that's how you get multiple products that win in different markets at the high end when you just tried to build a high and gaming card and then say it's also really good A I is where you get the trouble because everyone doesn't blame you. You have to come out at the other way. So as much as you can, I I and others in the enthusiast market hate IT where we get the code crown ami downs from the server market, from the data in the market. That's what's delivering your best performance.

And your gives you .

thread ripper IT, gives you forty nineties IT, gives you, uh, all of these things together. So if they know you, you gotta be willing to take these these compromises in thought process of how they build the products that to get the the hand competition, so that maybe a fifty ninety isn't three grand, like the old R T X T V was.

But IT might be that .

you might be right. They can rise whenever they want. So who knows .

what it's going to be transported? Eleven, ninety five rights and how are multi generation product road maps define? And how do these change over time? For example, I D A four appears to be a mid range for consideration of a prior generation.

Will the future ship unna looks more like A A reset? What market factors and time scales drive these road maps in general? And i'm really curious. I know we talked about big R D N A for being cancelled, but so much more from people reach out to me afterwards really catalyze what I believe happened. I just like what insight can you give us into like when that decision would have been made, why that decision would have been made or not been made and like when that decision would like have to be made by and like how yeah yeah.

So you know, general terms to be a process or if you have a concept and then that'll go into a set of requirements and then that will go into an architecture spec and then that'll go into a project plan. And each of those kind of like gates, where everybody gets together and says, here we agree to move for increased the resource investment. So your concept is usually very, very, uh, cheap to build for the company.

You know there's a few guys coming together and put there there are some slide way to go, right? Everybody can say, hey, he's the market opportunity. He is the features we need.

This is the technology required. This is why we believe we can do IT. This is the time frame we need to do IT in to be successful.

Is the ask roughly in terms of, you know like what the product cost needs to be, what the suffer investment is, what the target pricing is and you know what the what are the key requirements that we need from everybody has the commitments to move forward that really looks like that and says, okay, yeah, we can fit this into our overall strategy. Then he moves forward into deeper, uh, devolution of the the requirements of what you're trying to do into an architect aspects. So that looks new, looks like that.

So everywhere you go from each one of those stages increases the resources. And that all time of resources assigned the project is, you know that money building up and up and up and up and when you get to um cancelling projects is because you look forward and you say like, okay, the investment changed. What we planned on was x amount and now is doubled.

And the market opportunity, we thought I was here and it's half and now the return on investment doesn't work anymore. So that's when you fine to look at IT and go, okay, that's that's a bad use of money i've got, which is always more ideas than can be executed for products, for request from customers and really depends on is your company market, uh, driven? The most important thing is your biggest customers inputs and what they are telling you for technology partnerships for give me for business opportunities for where they want to go over generations or is IT and internal.

We can see something cool. We can do we have a vision and we're going to go make IT true and we're gona build A A castle out of air and then make IT into reality over time. Um AMD is kind of fifty, fifty on that.

I think maybe a little bit more on the market side, right? So whenever customers are say, you know you pitch you when you get to this concept, you don't just internally look at them. In five years, another group file guys, you go tested the key customers and get the yes is in the notes and like, oh yeah, i'd like that.

And I think if the feedback of we should do this, this and are you going to do that in the other and that, that loop keeps going. And at certain point, when enough customers go, i'm still in, but only if your pricing is here or i'm still in. If you include this requirement for me, and that's hard required, 那 什么 your your competitor will do this for me。 And if you won't, then i'll go with them. And then they start to stack up.

This tends to push one way the other and then IT comes down to like, you know, how can we execute? And then it's are you you relying on something magical happening like a third body enable ment? Like is is everything you're doing reliant upon a third party doing something as well so you can stack on top of IT? And if that's the case, then you really have shaky ground, right? So A I is a great example of that in consumer notebooks, everybody was relying on copilot and then they went away, never breathe go and I got A I P these that don't do A I things um why they sell in like wonder .

why was asking for the right.

So that's how they they make these decisions and know IT can be um the biggest expenditure. Uh the risk way for buys when you're getting ready for production, right? The previous one will be a tape out when you send IT after the foundry, you got ta pay millions to go take the chip out and buy the wave. Do that before that is gonna like how long over time is in the investment? Um so in the most expensive time to cancel products really got silicon back.

You've ordered the wafers and your software teams are starting to right code and you say, um man, this isn't gonna go we need to rise ain or cancel and then is a call of like, you know what are the commitments we made? What's the full back position? How much damage is this overall? Um that's the worst time to do IT.

So you try to do IT ahead of that um and I move everybody over but it's yet. So I can be six months before launch right when you um tape out and maybe nine months these days because IT takes five months to get through a fab for for a design ah so nine months before a launch, you could is really your last go no go. But IT could be a twelve months before launch, eighteen months typically.

Ally, these product cycles and planning cycles are five years. So today, I think the leaders that these companies sit down and they're looking at his what we're doing in twenty five, twenty six, twenty seven, twenty eight, twenty nine, thirty, they know what they going to do to thirty, thirty five, twenty thirty five, but they are focused on resources and planning. They have three year plans on revenue on ox to spend to achieve that revenue capture investments, competitive updates, those are all reviewed quarterly. Um and then decisions usually made um they can be you know something unexpected happens and IT can be two weeks after the big decision was made to continue forward. Actually, we would learn something quick, meaning right down, but typically it's on a quarterly basis.

Yeah and it's interesting you like walking through all that because I remember there the rumors, big guard in a forest cancelled and then I was told, yeah, it's happening and I I think I was like a man the two later, maybe like someone I was I talk to that I get information from we leave at that had a diagram of IT, you know and later than I heard two like a ibs. I don't think they have the diagram.

I think amy, like this is the twelve chip years out works. But I heard that a ibs were very disappointed when empty cancelled big r DNA four. That there is even some denial for a few months of that.

Well, maybe and and you know there probably was some true to that, they were you know you always have like a week to change your mind or something, but like at first, my assumption was and because some people told me I don't know seem really complicated their having issues, but then going to going into too much detail. I say that on a podcast. This happen all the time.

Someone else that amy reaches out to mean is like, no, here's my credentials. Here's what I do. I am the real deal. I don't like what you said, a broken silicon. This is actually what happened, and this person told me, in addition, a lot of other interesting stuff would have worked.

You know, we had problems, but I just would cost a lot of money like, and we were making something with, I mean, i'm look at the direction right now again, it's a one, three, four, five, six, seven and eight, nine, two thousand thirty least four teen chip lets. And they said we're sure would have worked. I just would have taken this much time and this much money and we just didn't think IT was worth IT.

And I think there was also a little bit of this. Also, we realized whether they call IT U D N A R R D A five. However, we were going to do chip less the falling gen at far Better results in simulations than what we were doing with this and where like, so we really going to commit to like this fourteen to twenty chip, let monster.

Now that's costing more than we wanted. Yeah, especially when R D N A five looks like it's going to do more with less. Like why would we do that? And to me, that's what IT seems like happened, you know.

And I think all too often people look at that and go they gave up. It's like, well, I don't know. I mean, maybe they just have something Better that's going to come simple than you expect. And I just eventually, especially once the A I thing took off, they are like, there's no way guys like what and I I like the A I is probably said, yeah, this will be great and we sell for fifty one hundred and they were like. Maybe or something you know .

yeah yeah Price to performance. The I ibs are always keen and excited to go in to do new tech and new exciting things, especially bigger big, big new tech. They can sell the heck out of that really way you just described as no the or why wasn't there so they looked at IT, said, you know how into the next one, okay, we execute fast.

So I don't think it's a bad thing to to not really take a nee, but really do that concept right and do IT justice as well as um save the money. So no, this is not like they haven't just done layoffs, right? So they do very consciously realize their cost profile to their investments. They don't hold people around on things that they don't value, right? They want to make make IT the best return on the dollar um b fish right .

divi insides into how amy decides to forecast and allocate wave for supply across product lines. I've alone assumed to as margin base, but that dusk of consumer could have higher volume and maybe make up for similar or margin species. They are ship with that a center.

But to be fair, that's just an assumption of mind. And I guess really getting up this question, I just I we think about like this all the time. I broken silicon.

Me and Michael, much of IT is like how often like I am thinking like the P S, 5 pro uses four animator and five uses for the animate are four uses for animal。 Is there any insight you can give us until like a decision being made where it's like note, not using the way for for that too bad. It's just not going to have that. Like and how far advanced can that happen? Because I i've heard I don't know, I could be completely wrong. It's like, well, yes, you can share waves between products, but there's a certain degree of like this set up actually has to make this product even if it's foreign n animal because that they prepared for that and like how easy easy to switch between things and how much of IT because I mean mad when I look at them selling, for example, then three back in the day for eight hundred dollars for, what, three hundred million ter square of silicon versus five hundred in twenty eight eight meter square, I think of silicon for a one for, like a one thousand dollar at work, or six fifty or sixty eight hundred X T. I can see why I would be hard to justify those radio guards.

Yeah I mean, won't know wafers are equal um so they do get together and managed the the capital investment and the the forecasting is is intensive here. There's a lot of supply demand planner at all these companies um the winning and yields a lot of the real reasons why you see uh launch strategies the way that they are. Um you have to hit the sales target. So um you you you get customer orders that really helps to firm up know if you get anted.

So and the way one product, but you've forgot guaranteed double the sales over here, you might still just go with that, which is probably by play station. They still make so many of them.

Yeah that's that's part of IT. You got you know this customer agreements for a supply like hey, you when you when you get big design winds, they put in contracts like you can't short me, I get priority if I don't then there's a penalty or we we have to meet to explain and and agree why there's no penalty exception.

So those things happen um mostly it's the um the component channel products that get the um they have the most amount of flexibility, right, of what you can shift demand around on um to move products into different things, especially inside of um this stock product lines like rising is really obvious example, right? You don't want to sell on day one six cores, which you got uh bits of silicon disabled when you could say a right, you said you start up and all the equals and you stop winning all the ones that have the natural fall and you collect to stalk of them. And once you get enough of those to fill in your rise on five definition, you start selling them.

And then that's when you start being and defeat ing by volume sales instead of by natural full out. So is no is a little bit of margin, a little bit of product mix um uh also about know every number that every sales guy gives is broken down into product. So the the guy that managing three system in integrators and a distributor in the pacific northwest is given a target for his sales and he is then allocate that across different products.

He rose that up and says, this is what I think it's going to do, and that's based on the history of what he's ever sold, is to talk to his customer. How many do you, anna bike? And I talk you into this many, this Price.

He rose that up, places his orders in and then the supply managers aggregate all of this and is an ongoing effort because you've got to have the linearity right. You know what I had try to do all of this stuff at the at the end of the quarter. Um you just hamari your inventory, your warehouse, to ship a quarter's worth of products in seven days.

You want to have a nice and Linda, that continuous supply bees got large shipments, uh, very expensive. You not drop shipping things or rare shipping things when you could be put on the boat up for a cheaper and maintained the market so that the wave of supply is usually, uh, forecast I at the beginning of the year for the whole budget of the of like a financial year and we'll be upper, lower bound, but they'll be some accuracy to that. And we based on years past and they say this is what we need and then it's regularly monitored as they go for.

And like I said, this can be five, six months last time on products. So if you're running hot, like you're really doing well in the first half of the year, you might say are we're going to buy more away because we're sell of them, right? Do you know what's the back half of the year? I like but then if somebody comes out and says, well, second half weakness in I O T markets you going to get a car at, let's die that back a little bit, you know may I does not spend that money.

Just wait for the signal and stop cleaning out because you don't want to have too much uh capital tied up in um in certain areas you don't want your customers to have building inventory. You don't want to have aging inventory. You especially don't want to have returns as uh distributed to say, I bought all this based on your sales forecast in january and here we are in August and I haven't even hit your march target yet, so you can have this back.

I wish this a lot of their contract. They they're allowed to do this, then they will return. So all of that um base a really large way IT doesn't just start with I think we should sell more rising lines at high margin, right? IT is a fee back loop very closely monitor.

It's very complex. I think that IT will get Better as we see um agent agents A I take on some of the things quickly of them. The forecasting is very manual, lots of data processing gathering and stuff like that. And A I can help together that data and present views faster so that human can make a decision on, okay, here's the train lines, and this is my estimations.

Well, if we could remember some of that manual forecasts, and we had a worldless pandemic, everyone's like everyone is going to stuck inside. So certainly known you'll want a laptop or a car when they can fly, right? Like, and then then they are all overcorrecting actually want to ask this directly.

Like i'm thinking of, like you're ten year AMD, like I would imagine was a cava probably shared node with the theory x or so like both one ter um and there's and those examples of then with the chip lets and GPS got I O I versus polaris was too. How often did you see that like something like and I don't need physics. I'm just curious that happen once a year.

Have you seen IT twice in your ten year, the empty of like? And again, i'm just giving an example. I'm not saying this literally happened, but like then to selling well, polaris not switch to iowa eye production.

Like how often did that I actually have? Because I think people think that happens all the time. I just don't know .

if that does not as much as you think, because the um the upper and lower bounds and stuff is set to to help manage this. Um also because not every um not every transistor is the same, right. So where a graphics card may be using the high power version of the process to CPU would might be using the low power and the I O die might be using the ultro power right?

The hello they are L P L P P or something like that version and so you can't just say like, hey, I bought all these wafers and twenty eight enemy um not going to build of my fury x is i'm going to build I O dies because they don't they're not the same line. They're not the same way for um if you're early enough, you can say or I forecast to buy x amount of wafers from you, I want to change my mix and I can go do that but that you know obviously long lead time stuff I can't do IT in in kind of a quarter. I can do IT for like a six months you um so I I think there is some of that that goes on.

They do try to reshape um those ly agreements, the purchasing um blends and stuff um obviously uh I I do remember when we will put together the forecast on uh I M four and we were saying like hay for chip sets. So we got two chip sets. We got the the x three seventy, the b three fifty that both commentaries and then we ve got the a three twenty.

And then we've got these uh finding basically enable uh chip sets that never went into the come into the market of D I Y。 They went just to the die. ms. They were just say, hey, they were just know you got no extra they were just literally.

hey.

i'm an thens amb. Mother board, you can work and you know the marginal, those was like ninety percent because they were dollar and they cost nothing to make and the the planet came. Just seems to be like, why are you so heavy on these expensive chips that the low magic, why you just say more of these, uh, little ones, can't you Price them up as on and just do more volume than we have to say?

Now that will look from a product perspective, right? Is missing all of these things and these are the core requirements for our thing. Ah okay, we just I saw the margin. I really like I wanted to sell more of those you do yet that kind of conversation happening.

Not everybody in these companies is deeply experienced um D I Y pixi builder that has the heritage of um from the aslan days to f one X P through to um now and understand what is that gives up, what is a another board. All of these guys have no idea that just really focused on supply demand planning and understanding the lead times and getting those things put together. So you do get um I remember I think one time um I remember more often than not that, hey, we need to buy more wafers.

Can you sell more stuff? Can we stuff more into the channel? Vers is a we've got incredible demand on this product over here. We want to start with you.

I I remember IT more is the other way round of like um pick up the bag, right you got ta deliver and say the sole but that no i'm sure it's uh, different now. So I think probably they have conversations were like we're seeing on president epic demand. We need the wafers.

You're going to show your supply on death. Um sorry, this is what IT is and because like I said before, there's a uh, fetish e responsibility to to demar mize profit on returns on investment. So if you've got a choice of telling ten things that you know, high margin, many things at low margin, you gonna pick the bigger pilot dollars.

It's only if you can look everywhere in the iron, say, like you, yeah, I was today's gained, but tomorrow is loss because I would lose that customer permanent. And they did that man over time. So that was usually how the the conversation went. So there wasn't a huge amount of back and forth because relationships matter, right? You can't just go to everything .

into a bit. You go to hp and be like, you know sorry, we're not making these gp s you ordered because rise and so profitable like you had a country yeah that feels like you're same you so you've seen IT happen where they have switch allocation. But for the most part, it's just make less or make make less or make more of something because we're forecasting IT not make more of this in respect and make less of that at the same time.

Um alright, so basically the final thing I want to ask you about, I mean, previous stands around this company the entire time and I want to ask very directly in video right now that seems like in videos boys to launch a very expensive fifty ninety and fifty eighty in january after announcing that would see yes, how do you see black world paying out? How do you see in videos place in the gaming market right now? And is there anything you would change from your perspective and how and via conducts themselves? Like should they still be aggressively trying to have some good five, four hundred dollars ard is IT makes sense that are going sound like fifty seventies can be six to seven hundred. Um is there anything you changing how they're doing?

And how do you see their place in the market? I mean, there's the champs. They have the the response very maxon as the R R Y. I think yeah if you've got jenson being wide and dying by elan and others all at the same meeting, the people that would never go and um he's telling is E P used with seal bits over dinners then I think there's a credible argument to be made.

Every gaming G, P, U, he put into the market where he makes violent of books could have been a forty thousand dollar platform where he made one thousand dollars. So not to say, oh, google n video. Thanks so much, but appreciate the i'm just saying that that they're still treating the gaming m rad.

I could be radio .

they're still treating is taking this seriously, still investing. They haven't forgotten where they came from. They haven't forgotten who gave them the the moments to start right from the from the very first river days and want to going forward.

I come back just wrote a great little article about how uh in video changed the gaming market by taking on driver to help themselves and making IT Better and high quality and and simply to work with though the games could run Better um so I think that still inside of them all ously, I want lower Prices. I'm not excited about any six seven hundred dollar component. And even as I say here is next to my thirty in twenty twenty x at per system, right? Uh, I I know, but there's a reason why IT hasn't been upgraded in a few years, right? Because wait now right.

So I think the um the real game of battle is gone to be down in the sub seven hundred dollar Price point is five hundred dollars and lower everybody um everybody y's feeling the pain of inflation of the market forces of games is getting more expensive. So is is fifty series stephane class GPU above going to be professional? You know they're really for people who have a business use for them. You could make the case .

this for pro gamers, my deo I put out today, but I was sent a document that i'm not sharing for obvious reasons that said that NVIDIA is direckly len retailers right now, that anything above the 3DT， I should be marketed as professional at places like micros and best buy.

Yeah, yeah. no. I mean, I think if they were to do that would be for two reasons. Number one, it's get on the front foot to manage um the the less uh informed by as right little john y says, hey mom, I need to do in video a cards for four and SHE goes to microsite so they don't try to sell or a thousand or or G P U.

They sell something less and SHE doesn't come back and being like why why isn't that this much where this problem no, those Price points and expectations in terms of support and in terms of what they can do, right? I the general the general public, the general gamers don't um know there's a mismatch of expectations. Calling our professional card is saying yeah and game but is really if you're intensively using uh editing video editing, you're company creating with A I you're making money of this thing is that sound like you is this a family PC that everybody y's using and no, I just the kid game, right?

Don't buy by that uh, unless of course he is already a streamer and in which case probably already got the revenue stream to go by forty nine and whatever eta by himself. And so I I I think it's the defensive manuvers. Um the other part of that would be, hey, that maybe managing some supply um and it's a way to like prioritize the supply to you like they gona have less cards.

Maybe I don't know, I think they'll have a healthy supply. But there are also giving a reason for their planners to to anna respond to is like, hey you you know you when you sell these seventies and you give them to microsoft's not just kids playing video games, right? It's apple coming to buy a professional cars, enable their business.

This is actual revenue that we want. We want to drive mindshare here. We want to continue to have people buying in video cause for professional products. You know, they're investing in this, uh, studio driver X A, right? That's not something you see too .

much on the side drive.

right? They really focused on like Alice, you know it's the dobe certified driver with the creative driver on new videos. okay? You got to consume a kind it's focused on stability and enabling features in premier pro and stuff like that.

But I think that segmentation is is increasing um and I think that is partly to help them manage allocation with adding boards um as well the a ibs because some of those companies are pure gaming companies and when they position IT as a festival card against, say, we look and prioritizing my supply on the companies that serve both of these markets, professional and gamer, your company branding is purely gamer. You only talk to them. You talk about exports cafe as you talk about for tournament.

You don't talk about um A I and video creation, all these other things that I am pivoting into when you come on and join with me. And that and that marketing, then the supply can ease up. But if the the given themselves tools to manage expectations, not just with consumers but with distributors, with partners, with resellers and with builders themselves.

so I wonder what you think about this idea is an idea brought up before. But like you know, IT feels sometimes to me, like in video competes well, they're certainly competing in GPU, but they go balls to the walls when they want to. And I think that in video saw the two ninety x, for example, beat the and then the feel well, and the seven hundred and seventy, before that really know, time almost matched.

They are dying in as well. I want a matter of drivers and they went, okay, A M D can compete if we just sit around and then that's when they just boom, went for with maxwell, more gas, surprising, they accelerated the release on twenty eight and enemy, I mean, they just went for IT and they that's, in my opinion, really when they crush M. D.

As maxwell they just kept ammori with, they could have charged eight hundred for the ten eighty if they charged seven hundred and five hundred and the ten ti, cuba thousand they made IT, they didn't let up. They didn't let me be up for air for a while there, and they've been living in them up for air lately. I think they've been in some room. But I wonder what you think about this idea of, like there is actually an opportunity here that jensen probably okay with.

Like I think I almost get the feeling that Johnson right now it's gaming and is like, hey lucking, if you want to kill yourself to take twenty percent G P U market and get to forty percent or something, go for because we're making so much more money in A I at the same time because amy is not a one trillion dog company yet oh no, maybe you know what what's metaphor I could make? Like, you know, I might not be a loser dinner, but stick to someone who is chicken every day by the lobster. Like, you know, like, so I wonder what you think about like this actually with our DNA four is an opportunity and amy should not make IT six hundred.

This should make a five hundred because I actually in videos fine with that this year. Do you think that's a good argument? Or do you still think it's just same old, same old? Not really. They should just stick with trying to do everything they're doing everywhere and do only what they have to do.

Yeah um that's that's a different thing at GPU first company and a CPU first company. right? So in video, maxwell, they they invest heavily.

That's coincidentally also went google aim to them and said he building a tenser unit. I want to do way. I I need hardware. I can't build IT.

You're the only guys that can, right? And if you throw, how did max will come out is so good? Well, is as they baked in from day one, had when they went to concept, a customer saying, I need this and i'm onna pay this for IT and I went, okay, so now we can do all the things that we could previously hesitant about because we've got that locked in, right? That confidence paid off over a decade like we've never seen, right? fantastic.

As the N. B. Turnaround rise is being invidia just off you then the million folk to whatever kind of korean freier. A M D is right. So just the year, they they for a few years there where they were so dependent on gaming revenue, then they absolutely went for grounding.

M pound is no mercy, no quarter, every single launch, delivering execution and and and covering all the basis now that the ground pound has shifted to data center, then they are not focus so much right there. But I think the people in video, I still have that mental, all those guys holding out and there's new people coming in, but I think there's still that, that believes that they want to win there. So when A I N D I I think that if A M D um says, you know what, let's let's go hard and do five hundred dollar uh graphic cards, I don't think in video will chase them.

They'll say we're in the headlines with the ones ever won to talk about every day. Went on sunday, still on monday. Is gonna work for us. And i'm already taking the margin hit to go enable the G, P, U. Market instead of selling everything into data center.

Why would I take more of a margin hit for them to get five points, right? Because I think the latest that was like like ninety percent GPU share for a video. Now it's it's incredible. Yeah depends how you count that.

I mean in some IT depends on the market. And I mean, here in china, it's like ninety five .

right but right so dependent account way you look but you let let's be um speed generous and say M D has fifteen and the other eighty five is in video. If A M D goes to twenty five, fantastic. Nearly doubling their market.

What a win for them did in video. Feel IT right? Well, that seems like you think, no.

think this seems like you think amy does have an opening because of how much in video doesn't care anymore.

I guess what i'm reacting, I think they care, but IT doesn't raise to the level of reacting about IT is like, you know um mosquito bites you do you slack yourself harder than the mass was? So he just flick up so I IT IT is going to be interesting to see um and and how it's portrait.

I think that the the influences of media gonna have a portion in this um but like do you talk your margin on your G P products that could have been A I just to hold five points of market share to combat some headlines? Or do you just sit back and say, but I sold more in a quarter than these companies that compete with me do any year right at high I margin and i'm out of supply like which is a Better response in a story to everybody, uh, that they care about. So yeah, so little is it's gonna be interesting, the sea. But that the way that the two ways andy can play that right is, well, if the video I don't care, then I can Price high, because doesn't matter what I do .

OK not saying. I'm sure we'll do five hundred. I think six hundred is very .

likely for andy, right? I mean, if the way I would approach this is I would come in um with a solid Price performance, uh, improvement from the previous generation. You know make IT look uh that every frame you get is you know ten percent, fifteen percent lower Price and what you got from the previous gen of AMD and there is competitive with video, maybe not uh, more expensive, but you know two, three percent right in there. And then watch what happens in the market and be ready to pull the lever to .

to point this ending.

If you see a real opening like I, you set on a bunch of leave IT alone. If the sales are kinds slow, will stop rolling off, then O, K, let's move with IT and and continue to drive IT and definitely be asking your product team can give me a at eight fifty right in six months time, something right, so that I can bring that in at my original Price point and push the eighty eight hundred down even more right? Just do that Price point, replace and drop and everything else and just push that down.

So this um there's a number of strategies they can employ here will see what they wanted do. Um I think there's serious about what want to take share back. The perception was when when an until launch this GPU that the really wasn't in video fans and A M D fans that were in video people and not in video people, because the only person who have lost here for A M P, so A M P after to go counter that they were or no, actually, we're really good at gaming.

We got these great TPU. We got all this fan base. Radio is everywhere, and we're gonna win the hearts and minds of that community and and show IT off with the eight eight hundred next to whatever. It's cool.

Yeah I mean, I would say a couple things like number one, um you know it's like it's really comes down.

I I really think even at five hundred dollars from what i've heard, the dies are very big anyway like they can probably still make good margin and the question will become, is IT going to be good at good Price performance six hundred? Or actually is there that extra large? And where at five hundred dollars, actually we're still making good margins? So let's go for the kill.

But I would say to AMD, i'm sure they're thinking about this um is if you think you have a kill on your hands here, be ready for the follow up because my special is jenson wouldn't care for one generation and he doesn't I mean, he's got A I IT is a matter. He will care the next gene. There will be another maxim moment if he thinks he needs IT to just regain the market market.

So I would say if amb and I I think this is something I kind of hear behind the scenes, is amb a little scared to go for twenty thirty percent market here or more because they know once they go for that, the gloves come off again because jenson does not like losing. So if you're if, say the anything, if you're going to go for IT, make sure na is also looking good because it's onna need to really try to, you know harness what you've built and and think I was such a disappointment that like r dina one was surprisingly very good, very good petite for I think what expectations were. R D N A two, a lot of people think was just Better than the amp.

here. I just didn't have production to take market share. And then R, D, A, three fumbled is not terrible, but IT fumbled.

And I think it's like, well, now you got to start over again because you need three genes in a row of just feeling like you're at the same level before. I think you can really take minds. And so I don't know.

I think a lot of IT would also come down to is like A M D. I mean, you ready to really go for IT as well? And I think that's work going to learn as when we see what they do.

I don't know. I had A A pretty visual reaction to you saying they afraid to go for. And because the guys that I up and road into battle with rising against, now we want.

we want those are different teams. To be fair.

I know how much they different now I would I would do some .

poking so you think they're they're .

out for blood. Then I think that there's some quite confidence you know jack Jones running that now across client and graphics. I would um I would say they strapped up the gloves and and ready to write and I definitely don't think they are afraid. I will will say the you know maybe the what's a winning scenario outcome is not what the enthusiasts market thinking is, right?

If you just put together a plan and you say we're going to do conservatives this and uh, aggressively, this is our target market for market share at a Price point over a number of units and a given them of time and you deliver more than that, you'd be every metric then you can't call that failure even if you're at, say, eighteen percent of the G P. U. Market, right? That the certainly the case, remember we started from AMD rise and started from like five percent, yeah right?

When we got to ten trumpets everywhere, but ten percent really you think that's great. IT really depends on your advantage point. If the plan was to get from five percent to seven percent in the generation and you deliver a ten percent, your heroes, your awesome, you wonderful, because you beat the plan and deliver on the commitments, and everybody outperformed.

So IT really depends on what they trying to do with the A E with this new card. How are they trying to go to IT? Sounds me like they are going after the harder market.

That sounds like an aggressive, aggressive strategy. Um how will they get there? I don't think they're onna be uh shy. I think they're na come in hard and fast, but I think we're going to see the standard um intro pricing maybe is little high and dialed in um and go from there. Used to be we could go to the the media and and test Price points and stuff, but now everybody just puts that straight out on the way you have to go with what you go and and go from there.

Um I can't think of a Better way to end this pot, guess. So I think i'm going to end IT right there actually. But like I always say to my guess, you know, before I let you go course, thank you for coming on.

But please, anything you would like to plug? There's been things before. Is there anything now?

No, i'm i'm excited to stop my new role with with the MIPS and coming days and help them with their of the product strategy and marketing strategy to help build the next generation of A I accelerators and data intensive processing processes with the customer silicon ecosystem. Um i'm really excited to see what's gonna happen with uh all the fee s announcements around uh, trying gaming technology and A I in the client space.

I think next year it's onna be a really, really uh favorable year for all these technology bets, the payoff. Um so I think it's the great time to be a gaming enthusiasts thinking through earlier, like the technology we have available today, the performance that features just absolutely fantastic choices compared to what I was twenty years ago is unrecognizable. So when we're talking about, you know, is that really possible for intel? L A turning around A M D.

Did IT can uh can in video hold off the A M D attack on GPU? Intel couldn't on the CPU side what you but there's so many unknown, uncertain ties out there. So I I am thankful for you guys for appreciate and help me come back on great questions. Thanks so much to all of you who listen, and thank you for your time. I really appreciated.

Yeah and one one thing i'll just said until like I do think people need to take a step back every now then I go I mean, I remember when dirty frames per second average was okay. And like the four, eight seventy review, i'm pretty sure was like those reasons like thirty on average, I think things everybody was dropping in twenty four friends per second sometimes and then IT went to sixty.

And even like ten eighty reviews, look at up four sixty average. They were fined with a drop into forty eight frames, and now people are complaining if I locked at one forty four or hire. And even councils are running some games that want to want to hurt.

This is where people jump in. Oh, not that many games. The fact councils running any game, that one to any hurts is such a are cry from where we were ten years ago.

It's just I know gay means really even Better, especially in terms of performance. But yes, thank you for coming on again. James prior, everybody, you can find him in the description as well if you want to see what he's doing.

And thank you for watching everybody you are listening remembers the scribing more laws that on youtube support us on patron. And have a good week, everybody. This podcast was brought you by the youtube channel and website.

More laws dead. Mos law is dead and broken silicon are tradeMarks of the creator tom that guys me and I am indeed the creator, editor, writer and show runner a mois lazed that podcast videos, articles and other media. However, it's not just me. More sla dead is a team with broken sicond o hosted by my brother dan, audio added by jara cortez, renters being done by the industrial designer john fully cml special assistance is also provided by carbon cry Carry no sugar as well.

Find all of information at W W W dot moralized dot com on the about such support page in the event you do want to hire me for consulting work, for audio work, hire e joon fleet for industrial design work, or your interested in working with carbon, cry or Carry. No, so you got IT as well. You can also find our long term sponsors on that page.

If you want to show them some love for putting food on our tables, or you can also mail a some love, you can send letters or hard redondo to the the following address moralize dead pill box six zero six three two and national tennessee, that code three seven two zero six. Although, to be honest, the best way to show more is laws dead. Some love is to support us on patrons.

On patrons are what makes more laws, dead content truly possible every month and really every day, depending on who you're talking about. Me, jd, dan and john flip are working tirelessly to provide a steady stream of content that we could not keep doing unless we knew work was possible without being reliant on sponsors dictating every little thing we put out. Don't get us wrong.

We love our sponsors, but we love directly working for you, our fans, much more. If you have any extra money, even a couple three dollars a month, consider supporting us directly on patron. Those couple of monthly dollars will get you access to the exclusive podcast dyring voting on subjects of future podger episodes, the ability to ask, ask questions, and of course, access to the most losted discord full of like minor people who I am sure would love to meet you.

I am one of them. Additionally, higher tears get access to early and three episodes of broken silicon, the ability to ask questions in all broken silicon episodes and loose ends, live streams ahead of the recording and the entire back catalogue of mos laws at podcast. In addition to having things in the credits of videos and podcasts, depending on the deer, with other perks available as well.

And hey, if you cannot afford to support us directly every month, please do share mos last ted videos and pocket with friends and family and on social media and websites like redit and get broken silicon, a five star review on apple podcast or your preferred APP of choice. All of this does really help us so much. But like I said, this podcast will not be possible without IT the patrons directly, providing predictable and reliable support every month.

And so now IT is time to give their personal things to the greatest of the fails. The following supporters are at the ten gigg hurts or higher support levels. Brad beven, jua, fs, Z, J, anio, d. Dick, the r ern close, genral, Daniel, hi. J Z S, J B one S N S charmers, the very year of welcome ave toka jenson wang, gregory sacroc strick even dangle, impress A I blocks s poo how boa sherd holden bolly sem law and sympathetic league the boss house john athan, Christine and franco for a dick shake two twenty three jake Martin's lucky needs resign meat and port tim rob Cliff er travels coin maths two Taylors different coats greg Michael mai paddy cakes gers unite m able chief mark michell Tommy I should mark RAID maker James Anderson.

Jammy waiters call out a Cameron genre with pxi poo original r tec om my shona when wing windstar coral and our James I rather little german G S M H shame millton Melody warrior BRT Jones shut clinton mcDavid, even dic j if shot modern hagi just and bustle joe foot, brian canon harden janson and gema Joseph Kelly slush by him to gun tails two, two, nine, nine mile vell vea. This is a seven, seven, seven one, the forbidden juice bryant write R B race or A C. Alex favor, dr.

J. M. Free Michael cozy, john swenson, ystwith joe alan s. Gentleman Matthew maro, brian peace bruti who did had forty two the guff he guk corns to six, seven, one pental winter.

How Richard Angel of cake, john with bombs k or make a dodge mr. Tiger san OS curry king three, nick susman, ia koa among vacua city farm snake and right hill chee balls. And of course, thank you to see here for the music.