Speculation: Ryzen 3000 series

[Mopetar]

Why would AMD add more cores when its products are already selling well at the current prices?

Then why do we have as many cores as we have now?

If you apply your reasoning to the past, AMD and Intel already had products that were selling well at then current prices.

When you've worked out the reasoning behind that, try apply that same reasoning to the current situation. I think you'll find that it explains why AMD is going to seek to deliver additional value.

 

[Mockingbird]

If something costs $1 to make and you can sell 100 of it at $4 dollars, you make $300 of profit. If instead you lower your price to $2 and can sell 1000 of it instead, you make over three times as much profit.

The only circumstances under which you don't sell at $2 is if you can only actually make 100 of that thing. AMD's chiplet strategy is going to allow them to make many times the processors that they could before. This should be obvious.

...and what makes you think that if you cut the price by half, demand will increase by ten times?

That is probably only true if the product is horribly overpriced in the first.

 

[maddie]

Has there been any info one way or another if the two-chiplet designs need to have the same # of active cores?
Zen 1 had to be symmetrical, but if Zen 2 doesn't, allowing even higher effective yields could only help.
Instead of throwing out a chiplet with 5 cores, maybe you stick it in with 7-core and boom, R7.

The fact that they communicate through the I/O die might change things. For sure, flexibility is one of the top design mantras at the present AMD.

 

[OTG]

...and what makes you think that if you cut the price by half, demand will increase by ten times?

That is probably only true if the product is horribly overpriced in the first.

It could easily be argued that Intel CPUs have been horribly overpriced for a LONG time.
Remember: the 6950x launched at over $1700, and the 9900k is selling for well over $500.

It was insanely overpriced for its development cost, but AMD had no competition so people who needed it, bought it.
You could look at Ryzen/Threadripper/Epyc as pulling CPU prices back to where they *should* be, in a competitive marketplace.

 

[maddie]

...and what makes you think that if you cut the price by half, demand will increase by ten times?

That is probably only true if the product is horribly overpriced in the first.

Like the 9900K?

I could say a lot more but won't.

 

[PotatoWithEarsOnSide]

I have to indulge in some ribbing here.

But, but, but, margins.

Meanwhile, Intel's margins on all of those lost sales fall to 0%.

 

[Mopetar]

I have to indulge in some ribbing here.

But, but, but, margins.

I realize that you're not the person who's making the ridiculous argument, but for anyone who doesn't know there's a difference between a product margin and a company's profit margin. It's perfectly possible (and typically normal) to be able to increase profit margin even though product margin has decreased.

As @PotatoWithEarsOnSide indicated, economy of scale is important. In addition to the manufacturing cost of a single CPU, which will decrease as you make more of them, a company will have certain costs for R&D, marketing, etc. whether they make 100 units or 1000 units. Increasing the overall revenue by lowering margins on a particular product can easily increase the profit margin of the company.

 

[PotatoWithEarsOnSide]

If I was at AMD I would be asking "How do I get all that pent up demand, from 2600K/4770K owners, to actually play their hand?"
The answer would be to offer them something far in excess of what they can currently get, and at a price that'd make them bite your hand off.

Edit: I certainly wouldn't be trying to get 8700K-9900K owners to buy because they are likely content with their lot just now. Show them that history will view their purchases as bad buys, such that next time they go with AMD, even if that isn't for a while yet.

 

[OTG]

If I was at AMD I would be asking "How do I get all that pent up demand, from 2600K/4770K owners, to actually play their hand?"
The answer would be to offer them something far in excess of what they can currently get, and at a price that'd make them bite your hand off.

Also from Ryzen 1!
I bought a 1600 with a Taich X370, and I want something with enough cores that I can put this ridiculous VRM to some real use; a 1600 at 3.8 ghz barely warms up the heatsinks.

Come on AMD, gimme something with enough cores that I never have to close a program again!

 

[Mopetar]

...and what makes you think that if you cut the price by half, demand will increase by ten times?

The particular example figures I used are merely to illustrate the general point. You can reduce the price of a product while still increasing the overall profit margin of a company.

History is littered with examples of this happening all the time. A particularly good one is the Model T. Ford sold well over ten times as many cars we he cut his prices in half.

 

[Kenmitch]

By the way, this data point as for a German retailer AFAIK. Is this the same worldwide? Do you know, or are you making assumptions and extrapolating?

View Amazons best selling cpu list. R5 2600 tops the list.

 

[jpiniero]

If I was at AMD I would be asking "How do I get all that pent up demand, from 2600K/4770K owners, to actually play their hand?"

Offer a processor that's decently faster in games.

 

[Kocicak]

...and what makes you think that if you cut the price by half, demand will increase by ten times?

The reaction is determined by the price elasticity of the demand. And in this particular case we have a huge pool of Intel owners who need some clear incentive to jump ship. The reaction can be actually pretty dramatic.

Please can you comment on your knowledge of microeconomics? Or business strategy? Have you studied these topics somewhere?




You cannot bait another member with this off-topic posting.
" Please can you comment on your knowledge of microeconomics? Or business strategy? Have you studied these topics somewhere? "

Both of you need to stop this back and forth between you. Now.


esquared
Anandtech Forum Director

 

[mattiasnyc]

I'm confused, you're making my point for me?
The 2990WX costs roughly what Intel's 10-core CPU did, two years ago.

It's possible I misunderstood you.

 

[mattiasnyc]

It seems a lot of people are assuming that in this coming year all that will be available and or produced will be Zen 2 based products. But if AMD continues to offer lower-end Zen 1 based CPUs then some of the arguments I see seem to go away.

No need to cripple or salvage 8 core chiplets to create a quad or hex core chip for example. Just continue to sell the current chips with those configurations and price new 8 core chips accordingly (and vice versa).

 

[Mockingbird]

It could easily be argued that Intel CPUs have been horribly overpriced for a LONG time.
Remember: the 6950x launched at over $1700, and the 9900k is selling for well over $500.

It was insanely overpriced for its development cost, but AMD had no competition so people who needed it, bought it.
You could look at Ryzen/Threadripper/Epyc as pulling CPU prices back to where they *should* be, in a competitive marketplace.

If you have to use an outlier like the $1700 processor that's not affordable for >99% of the people to make an argument (about a mainstream product), then it's not a good argument.

 

[Asterox]

View Amazons best selling cpu list. R5 2600 tops the list.

It is newer model so it is expected. But in reality according to sales statistics "old R5 1600" is AMD best seling CPU.Mindfactory is only one but no doubt very significant/trends sales example.

https://forums.anandtech.com/thread...d-possibly-9600k.2554023/page-4#post-39591667

So in the end, for AMD it is very logical with cheep Ryzen 3000 6/12 CPU try to break old sales figures.

Intel cant sell/not now or in the future 6/12 CPU for same $ or around 150$."Intel has a little problem", or about 90 000 more workers+hm 12 FAB-s to keep on living or working.

 

[Kocicak]

...
The last thing any producer wants is for their product to be treated like a low margin commodity. Everyone in the tech game wants to be in Apples position: Having very high brand perception, so you can command higher margins.
Slashing your prices and margins is pretty much giving up. It's not smart business, it's the last resort of a struggling business.

I am not sure if you noticed, but AMD few years ago was very close to bankrupt. They had bad products and very low market share. Now they have good products AND THEY CAN GROW AGAIN. Growth is at this point so important that whole AMD can operate with 0 profit, all they need is to finance their growth and development of the products that will ensure the growth. Only after they grow back and stabilise, they can start to care about profitability again.

AMD is at the moment ALL ABOUT GROWTH.

 

[maddie]

It seems a lot of people are assuming that in this coming year all that will be available and or produced will be Zen 2 based products. But if AMD continues to offer lower-end Zen 1 based CPUs then some of the arguments I see seem to go away.

No need to cripple or salvage 8 core chiplets to create a quad or hex core chip for example. Just continue to sell the current chips with those configurations and price new 8 core chips accordingly (and vice versa).

GloFlo isn't the biggest fab in the world. If Zen2 is very successful, they'll have their fabs full with the I/O which is a large die for Rome and probably TR2.

 

[OTG]

If you have to use an outlier like the $1700 processor that's not affordable for >99% of the people to make an argument (about a mainstream product), then it's not a good argument.

The 10/12-core level of performance is now $650, brand new.
Lots of people will buy a product at $650, who wouldn't at $1700.
Lower price yielding higher sales.

If AMD launches a 12-core R7 at $329, there are even more people who will buy at that price.
I don't know how I'd rationalize a $650 CPU, so I wouldn't buy one.

A $329 dollar, 12-core CPU?
Now I'm interested.

 

[mattiasnyc]

The 10/12-core level of performance is now $650, brand new.
Lots of people will buy a product at $650, who wouldn't at $1700.
Lower price yielding higher sales.

If AMD launches a 12-core R7 at $329, there are even more people who will buy at that price.
I don't know how I'd rationalize a $650 CPU, so I wouldn't buy one.

A $329 dollar, 12-core CPU?
Now I'm interested.

The thing is though that AMD has had a fairly linear pricing when looking at core counts, within each line of CPU (i.e. "mainstream" vs "HEDT"). So the current 12c being $650 translates almost exactly to the 16c part's price.

So a 12c mainstream CPU at $330 would put the 8c part at $220. This would be roughly equal performance per core as the 12c CPU. It seems very low to me, just as does a 12c part at only $330 considering where Intel's prices are.

To me the question isn't just where in AMD's stack of CPUs a particular CPU sits and how that then affects all the other AMD CPUs, it's about how that entire product stack relates to Intel. After all, we are dealing with a competitor as well.

So as a hypothetical, let's say the 9900K is the leader of Intel performance and is sold at $500. And let's say AMD has an 8 core chip that matches it. What is a consumer willing to pay for such a part? $500? Clearly, as that's what Intel is selling their parts at. So, why sell such a CPU at $330 when $400-430 would;

- make AMD more money per CPU
- still undercut the 9900K
- allow for a possibly generally wider spread of CPU prices in AMD's entire stack

At a hypothetical $400 for an 8c part it would still allow AMD bragging rights at the $500 level with the 12c part at about $500-550. They'd basically be telling everyone "Hey, for the same price as a 9900K you get 50% more cores from us".

We probably agree that there's an equation at work here where we (well, AMD) pits volume against sales price and then it'll shake out maximum profit. I'm just guessing along with others that pricing will be slightly higher, which to me is pretty much what we've seen.

 

[OTG]

At a hypothetical $400 for an 8c part it would still allow AMD bragging rights at the $500 level with the 12c part at about $500-550. They'd basically be telling everyone "Hey, for the same price as a 9900K you get 50% more cores from us".

But the 9900k is only that expensive because AMD doesn't yet have a competitor.
The second AMD releases that 8-core, the price for a 9900k drops to match it. Ok, it lags a bit because Intel, but pretty damn soon AMD has to cut prices again because if prices are close, people will go Intel.
Eventually Intel won't be able to cut the price any further, and AMD starts to really pull ahead, but it'll take a while to get there.

AMD can just put the chips on sale with decent (but not Intel) margins, rack up sales that would have gone to Intel's HEDT line, and get lots and lots of AM4 boards into the wild. And do it starting at launch day.
I really think that time is the key.
Intel is having difficulty with production, but it won't last forever, so AMD needs to grab as much marketshare as possible before Intel recovers.

 

[PeterScott]

I am not sure if you noticed, but AMD few years ago was very close to bankrupt. They had bad products and very low market share. Now they have good products AND THEY CAN GROW AGAIN. Growth is at this point so important that whole AMD can operate with 0 profit, all they need is to finance their growth and development of the products that will ensure the growth. Only after they grow back and stabilise, they can start to care about profitability again.

AMD is at the moment ALL ABOUT GROWTH.

Again, slashing prices to grow is an act of desperation, not sound business acumen

Look at Apple when it enters a new market and has no market share. Do they slash prices to grow share? Never.

1: They aim to build better products, and sell them at higher prices/margins.
2: Channel higher margins into more R&D, to improve the product.
3: iterate the above, while the market comes around to recognizing high quality product and pays more for it.

AMD is not desperate. IMO Lisa Su is a great CEO, she will aim to iterate on making better premium products and charge premium pricing for them.

AMD will be pursuing the Apple strategy, not the commodity brand, desperation strategy.

Can we put a pin in this? There have been over a dozen back and forth yes/no/yes/no... and no one is going to change any minds on the other side. Lets wait 6 months an see where the pricing lands.

 

[DrMrLordX]

The financials only reinforce my point, not yours. Compute and Graphics is nearly everything they do, but you will notice when they mention, the first driver is always "Desktop", not professional, not data center. AMD makes the vast majority of their money from consumer, not professional market.

I will make it simple for you: Vega20 shows you everything you need to know.

AMD will sell only 20k of these cards to the consumer market, at $700 apiece. Even if they sold all of these cards as pure profit (which is ridiculous), that's only $14 million in total for the entire lifespan of the product. In reality, BoM + R&D costs will drive profits for those 20k cards much lower. AMD did not release Vega20 only to sell 20k cards, plus some unknown/irrelevant number on the pro market. Vega20 wasn't even supposed to be a consumer-facing product at all until they realized that there would be over two years between Vega10 and Navi, and that AMD had better get on the ball about fixing that.

Think about it: AMD launched its last high-end consumer GPU in 2017 as RX Vega 64. Since then, they've had nothing. NOTHING. Navi might not even be any faster than Vega10. It'll use less power and be cheaper, sure, but Navi10 might not be faster! What does that do to RTG as a force in the consumer market?

AMD/RTG has ceded complete control of the consumer high-end graphics market to nVidia. Since Vega10, all we've had is Polaris refresh madness. And now 20k Vega VII cards. Which is next to nothing. Vega VII might not even be any faster than RTX 2080. And do you know why? It's a pro card, wearing consumer clothing.

A consumer-centric RTG does not leave the consumer market in the lurch like that. A consumer-centric RTG does not almost-exclusively design their next-gen, high-performance GPU (Vega20) for the pro market. RTG isn't making most of their money off the consumer market, and they know it. Most of their consumer-facing revenue came from Polaris being sold as a mining card. That windfall is gone.

That is all the evidence I have. If that is not good enough, then too bad.

To tie it all back to what I was saying originally, then, you can't use any of RTG's supply/demand pressures when examining the situation that faces Zen2. Zen2 faces a completely different market landscape.

 

[Anarchist Mae]

Again, slashing prices to grow is an act of desperation, not sound business acumen

Look at Apple when it enters a new market and has no market share. Do they slash prices to grow share? Never.

1: They aim to build better products, and sell them at higher prices/margins.
2: Channel higher margins into more R&D, to improve the product.
3: iterate the above, while the market comes around to recognizing high quality product and pays more for it.

AMD is not desperate. IMO Lisa Su is a great CEO, she will aim to iterate on making better premium products and charge premium pricing for them.

AMD will be pursuing the Apple strategy, not the commodity brand, desperation strategy.

Can we put a pin in this? There have been over a dozen back and forth yes/no/yes/no... and no one is going to change any minds on the other side. Lets wait 6 months an see where the pricing lands.

AMD isn't entering a new market, they're and old player who have a recent history of under performing and hot CPUs. Their position is nothing like Apples.

Many of us think it's possible that AMDs new chiplet strategy will allow them to sell cheaper because their product is either significantly cheaper to produce than at any previous time, or they are willing to trade product margins for product volume while maintaining healthy profit margins, or some combination of the above.

Do you think this is not a possibility?