TR Forums

NoOne ButMe wrote:

AMD is in a better position for a price war in the target markets by Zen.

Intel can completely buy AMD with the profits Intel garners in a single quarter.

You are delusional.

chuckula wrote:

which is that there is an assumption among the AMD faithful that there's a big cash cow sugar daddy lurking just around the corner who will buyout AMD and provide AMD with unlimited funds to do whatever they want.

Well, sure. I mean, we all can "agree" that AMD "needs cashflow" so, obviously, clearly, undeniably, QED that's just going to happen, like somehow. Because.

Glorious wrote:

NoOne ButMe wrote:

AMD is in a better position for a price war in the target markets by Zen.

Intel can completely buy AMD with the profits Intel garners in a single quarter.

You are delusional.

If we all gathered our empty beer bottles and returned them for deposit even we could probably buy AMD.

just brew it! wrote:

NoOne ButMe wrote:

Not that AMD should care about them. Who cares about losing a 50% margin sale to consumer when make many more sales where you sell chip for 5x cost of making.

Have you looked at AMD's share in the server/datacenter space lately? The enterprise market doesn't turn on a dime; even in an optimistic scenario, it will take a while to turn that ship around even if Zen kicks butt (which is nowhere near guaranteed).

JBI is completely right about this. Intel dominates to such a degree that being in and around the server side of things so much you forget AMD even offers server processors. I guess somewhere there's a workload that makes sense for AMD processors but then you'd be stuck with their stone age platform.

Convert wrote:

JBI is completely right about this. Intel dominates to such a degree that being in and around the server side of things so much you forget AMD even offers server processors. I guess somewhere there's a workload that makes sense for AMD processors but then you'd be stuck with their stone age platform.

The only time I still encounter one now days is when I fire up a small M1 instance, zone 1, in the Virginia data centre on EC2. Even that is getting more rare as the instances that are usually being retired at Amazon are AMD based ones.

Deanjo wrote:

Convert wrote:

JBI is completely right about this. Intel dominates to such a degree that being in and around the server side of things so much you forget AMD even offers server processors. I guess somewhere there's a workload that makes sense for AMD processors but then you'd be stuck with their stone age platform.

The only time I still encounter one now days is when I fire up a small M1 instance, zone 1, in the Virginia data centre on EC2. Even that is getting more rare as the instances that are usually being retired at Amazon are AMD based ones.

Yup, their percentage share of new server deployments has gotta be in the low single digits at this point, and their installed base is withering away as old servers get retired.

My previous employer had (and still has, AFAIK) an AMD server (dual 6-core Opterons, 64GB RAM, RAID-6 storage array) that a former co-worker and I built several years back. After the company got sold, the new owners brought in racks of Dell equipment and the Opteron box was relegated to a secondary role. It was probably being utilized at only about the 5% level by the time I left last year.

I do my part and buy AMD when it's the right choice for my server deployments. My entire TSM infrastructure is Opteron now, it just made more sense at the time for the price point I was shooting for.

There are those that blindly buy Intel every time but sometimes that just doesn't make sense.

Waco wrote:

I do my part and buy AMD when it's the right choice for my server deployments. My entire TSM infrastructure is Opteron now, it just made more sense at the time for the price point I was shooting for.

There are those that blindly buy Intel every time but sometimes that just doesn't make sense.

"No one ever got fired for buying Intel"

And, If you ignore Intel's stock and T&D try can engage in a price war. If they want to stop AMD fro getting 10% of the sales they'll need to cut ALL of their prices.

If AMD is getting sales from Intel going lower (as long as it covers production/etc) is making them more money than before.

Say, the average Xeon is $100 to make, ship, test, etc. and, Intel sells 5 million (made up number) with these selling for 1000 each. Or, 5 billion in revenue and 4.5b in profit. AMD's zen costs $150 to do everything and they sell for $500. For the about same performance point that reaches needs of say, 80% of the market.

If Intel drops to compete with AMD, say to $600, they "lose" 2 billion dollars a quarter. If AMD gets even 1% more marketshare, AMD is making more tha previous.

Intel is doing everything possible to keep it's earning staying around flat each quarter. They care about their stock price because for better or worse they have to.

An Intel that doesn't care about it's stock could slaughter AMD in a price war. However, Intel does care. So, it's hands are a bit tied.

AMD doesn't need a huge part of the market to be profitable on CPUs, just a little. If their 40% is correct, they should be fast enough to get that little bit at the least.

Of course, AMD's ability to execute that, well. That I call into question. If the products arrived as promised should be fine however.

What is wrong with you?

You are seriously claiming that AMD will hold an advantage in a price war because:
1) Their part will cost more to produce
2) Their gross margin is vastly less
3) Their profit margin is vastly less
4) ...Why am I even numbering these? Every aspect of their situation is drastically inferior to Intel's!

Either you have no idea what a price war is, or you are just incoherently rambling.

Glorious wrote:

What is wrong with you?

You are seriously claiming that AMD will hold an advantage in a price war because:
1) Their part will cost more to produce
2) Their gross margin is vastly less
3) Their profit margin is vastly less
4) ...Why am I even numbering these? Every aspect of their situation is drastically inferior to Intel's!

Either you have no idea what a price war is, or you are just incoherently rambling.

I think he's operating under the idea that because AMD has managed to scrape by making losses for a while that means AMD can continue losing money for forever and that Intel will go bankrupt due to some type of "price war" where AMD will somehow get third parties to fab huge expensive parts for no money and then sell those parts for very little money because.. uh... well because. Additionally, all of this will happen on-schedule and Zen is guaranteed to be a successful chip.

I can see a situation where Zen is a good server chip, but I'm not seeing a situation where AMD wins by playing the price war game.

Yeah, I don't buy that line of reasoning.

Zen, if it delivers on time, and at a reasonable price, could make some headway into server sales. That GPU, being so close to memory, will be exceedingly fast for acceleration duty.

Glorious wrote:

What is wrong with you?

You are seriously claiming that AMD will hold an advantage in a price war because:
1) Their part will cost more to produce
2) Their gross margin is vastly less
3) Their profit margin is vastly less
4) ...Why am I even numbering these? Every aspect of their situation is drastically inferior to Intel's!

Either you have no idea what a price war is, or you are just incoherently rambling.

No, I am basing it off a few different things:
1. AMD only needs a small portion of the market even with higher production and lower sales prices to become profitable
2. Intel values it's stock's price.
3. Intel cannot engage in a price war without lowering it's stock price a noticeable amount or stopping R&D for their fabs.
4. Intel is more worried about ARM than AMD and won't squander it's R&D money keeping AMD from gaining high margin marketshare as they cannot beat ARM even with a 2 node advantage with Intel having better nodes. Much less a same node.
5. The funding that Intel had in reserve has been put forward for their purchase of a 17 billion dollars which despite their previous 50b+ of assets required them to take a loan out for the first time in over 20 years.

That's my train of though.

It isn't that AMD is in a good position currently and Intel a bad one. Quite the opposite. It's that Intel has a lot more to worry about than AMD and their stock price is one of them which requires them to keep their gross profit even at least. They dropped about 3 and a half billion dollars in R&D this year. How much more can they cut to keep their earning staying about flat?

And, due to the R&D flatline and drop ARM will soon be able to have a node situation in which they're essentially even in terms of power characteristics. Possibly even ahead of Intel to 7nm.

What costs Intel more money? Giving 10-15% of the market to AMD, or cratering their ASP for 50%+ of the market?

Once more, if you want Intel to act like a private company or one that doesn't care about it's stock, it could totally crush AMD here. I think it will act as a public company and in the interests of it's stock price, as it has shown to do this year be artificially bolstering it's net income.

I get where your coming from but I think you could have simplified your post.

Basically Intel has no incentive to worry about small market share gains by AMD as they stand to lose more money elsewhere. In the stock market these days people want instant gratification. If Intel does anything to hurt the valuation of their stock (and dividens) in the short term, lots of stock holders may get scared and sell even if in the long run its beneficial to Intel's share holders. Intel has been a fairly steady stock for a long time, people love steady. The last thing Intel wants to do is ruin that. True buying power for public companies is your stock being desirable, they still have that.

Intel has to like where AMD is right now... just relevant enough to keep doors open so that Intel can't be technically a monopoly. Yet weaked enough that Intel still has control over the market and can push AMD back down if they wanted to. It's for this reason that I hope AMD can put out something that blows Intel out of the water and take them by surprise... but I'm not holding my breath.

Soooo,

Intel doesn't want to mildly upset it's shareholders, so AMD has an advantage because it has legions of extremely unhappy creditors?

Intel, which could dump BILLIONS into atoms for YEARS in order to ward off ARM with no meaningful effect on stock price or R&D, couldn't possibly thwart AMD in a price war in a single product cycle? AMD, for which Zen is essentially Do-or-die (Die in the sense that if zen isn't a sufficient success AMD is not going to be the same sort of company it has been for the last ~10 years now).

Intel, which is terrified of ARM moving up from the bottom, will just readily cede the easily defended, high-margin big processor market to AMD, just, like, because. That Intel will practically welcome AMD making this a two-front war when it'll cost them very little to push them out entirely?

That Intel, which has a little long-term debt, is vastly inferior to AMD, which is drowning in short-term?

---

Look, AMD *might* have a chance. They don't have an advantage.

Glorious wrote:

Look, AMD *might* have a chance. They don't have an advantage.

This pretty much sums it up. If Zen lives up to its hype, AMD will live to fight another day. If Zen falls short and/or is late, AMD will continue to circle the drain. There is no "AMD gains the upper hand" scenario here, unless Zen exceeds all expectations and surpasses AMD's hype; and that's about as likely as me winning the lottery tomorrow. And I don't even play the lottery.

Glorious wrote:

Soooo,

Intel doesn't want to mildly upset it's shareholders, so AMD has an advantage because it has legions of extremely unhappy creditors?

Mildly? Intel's data center group had an income of 1.8b on 3.9b of revenue. client computing has 1.6b on 7.5b.

Intel's gross income was 2.7b AND that includes the fact that they cut about .7b a dollars from R&D per quarter compared to previous 4-5 years.

So, care to guess what chips are mostly sold in that group? I'm going to guess the low to mid range chips. The ones that AMD's Zen should be able to compete with.

Intel already loses money net on every other part of it's business (net income is 2.7B, those two groups make a total of 3.4B). If AMD can hit Ivy Bridge level's of performance, how much will Intel have to cut prices?

20%? 30%? More? (500 dollars? 1000 dollars? More? NO! It's just 100 dollars! Yes, that's it, just one hundred dollars!)

The cost of losing their bottom 10-15% of their server sales is well under half a billion dollars or their low margin business. It will cost them more to cut prices and keep AMD out of the market. A lot more.

Intel, which could dump BILLIONS into atoms for YEARS in order to ward off ARM with no meaningful effect on stock price or R&D, couldn't possibly thwart AMD in a price war in a single product cycle? AMD, for which Zen is essentially Do-or-die (Die in the sense that if zen isn't a sufficient success AMD is not going to be the same sort of company it has been for the last ~10 years now).

Not anymore, Intel has almost all of it's short term reserves taped.

And, Intel squandered all the money it could have spent stopping AMD now in a manner that did absolutely nothing to stop ARM. Now if Intel wants to stop AMD they have to crater their profit and margins. Battling ARM never did that.

just brew it! wrote:

Glorious wrote:

Look, AMD *might* have a chance. They don't have an advantage.

This pretty much sums it up. If Zen lives up to its hype, AMD will live to fight another day. If Zen falls short and/or is late, AMD will continue to circle the drain. There is no "AMD gains the upper hand" scenario here, unless Zen exceeds all expectations and surpasses AMD's hype; and that's about as likely as me winning the lottery tomorrow. And I don't even play the lottery.

Yes, I've never said AMD gains the upper hand. I'm saying AMD can afford a price was because if their product is decent they'll be making more money than before in almost every case.

Unless Intel can shut AMD down to 1% of sales or something insane like that. And, that would probably cost Intel 1.5+ billion dollars.

How far do you think Intel's stock price will fall if their net profit craters and there are no "one time write offs" or such to account for it?

NoOne ButMe wrote:

Unless Intel can shut AMD down to 1% of sales or something insane like that.

In the server space I think they're already there. Zen has a long uphill battle ahead of it if AMD is to regain relevance in the datacenter.

NoOne ButMe wrote:

Yes, I've never said AMD gains the upper hand. I'm saying AMD can afford a price was because if their product is decent they'll be making more money than before in almost every case.

I don't believe it is that simple anymore. The server market is too broad and specialized today, comprised of dozens of segments. The "value cost" segment is only going to be a small slice of the server market pie. It isn't the same 10 years ago where specialized was simply a subset that meant, PowerPC, RISC and Itanium chips. Today Intel custom designs chips for the largest customers multiple times a year. They also cobble together more standardized modular designs upon request depending on what larger customers want. AMD doesn't have the resources for that nor the customer base volume to even fund custom chips anymore. Custom chips and "value" chips are not words that mix.

NoOne ButMe wrote:

How far do you think Intel's stock price will fall if their net profit craters and there are no "one time write offs" or such to account for it?

I think that's as much a pipe dream as Zen beating Skylake in both power & performance. Intel competes in the entire server pie, and the "value focused" server market is almost certainly less than a third of the pie. Budget markets are the least profitable segment, so Intel would naturally see only a minor impact if AMD managed to win market share in that category.

Half of the problem is datacenters are especially concerned about perf per watt more than ever before. For AMD to begin winning even budget sales they need to make a chip that not only is slightly better cost/performance than Intel, but it also must be near power/performance parity or it won't matter how much AMD drops prices. A lot of datacenters are constrained by the power input to the building, and so can only buy servers that fit within the same power envelope as the server being replaced... and they are going to buy whichever chip gives the most performance at that given power level. (They certainly wouldn't buy servers with worse performance than what they were replacing, either) A steep price discount to make up for a lower power/performance ratio is simply not remotely viable in that specific market segment.

Kougar wrote:

NoOne ButMe wrote:

Yes, I've never said AMD gains the upper hand. I'm saying AMD can afford a price was because if their product is decent they'll be making more money than before in almost every case.

I don't believe it is that simple anymore. The server market is too broad and specialized today, comprised of dozens of segments. The "value cost" segment is only going to be a small slice of the server market pie. It isn't the same 10 years ago where specialized was simply a subset that meant, PowerPC, RISC and Itanium chips. Today Intel custom designs chips for the largest customers multiple times a year. They also cobble together more standardized modular designs upon request depending on what larger customers want. AMD doesn't have the resources for that nor the customer base volume to even fund custom chips anymore. Custom chips and "value" chips are not words that mix.

segments, yes, however, the specifications of the actual CPUs end up mostly similar.

NoOne ButMe wrote:

How far do you think Intel's stock price will fall if their net profit craters and there are no "one time write offs" or such to account for it?

I think that's as much a pipe dream as Zen beating Skylake in both power & performance. Intel competes in the entire server pie, and the "value focused" server market is almost certainly less than a third of the pie. Budget markets are the least profitable segment, so Intel would naturally see only a minor impact if AMD managed to win market share in that category.

Which is cheaper to profits?

Half of the problem is datacenters are especially concerned about perf per watt more than ever before. For AMD to begin winning even budget sales they need to make a chip that not only is slightly better cost/performance than Intel, but it also must be near power/performance parity or it won't matter how much AMD drops prices. A lot of datacenters are constrained by the power input to the building, and so can only buy servers that fit within the same power envelope as the server being replaced... and they are going to buy whichever chip gives the most performance at that given power level. (They certainly wouldn't buy servers with worse performance than what they were replacing, either) A steep price discount to make up for a lower power/performance ratio is simply not remotely viable in that specific market segment.

AMD's chips certainly won't be viable on a high performance power level. However, for performance that requires lower clockspeeds, AMD could very well be competitive.

Will they be, I don't know. It certainly is possible however. The biggest differences between Samsung 14nm and Intel 14nm is that Samsung's is physically not as close to 14nm and it doesn't let you make parts with higher performance bias as easily.

The power characteristics should be about the same, and, AMD has pushed into a lot of stuff specifically lowering the clockspeed to lower the power consumption further. HDL on Excavator being a good example. Given they can be just as dense as Intel on the space level, that opens a lot of places for them to compete as they will almost certainly be competitive with pref/dollar.

Few installations are looking for the maximum performance at any cost/power.

Lol, all this assumption based on enterprise customers (where the money is at in servers) pay retail prices.

Here is a hint: They don't, not even close.

I guess my simplification of his post was yet again not simple enough.

The way I read his post was as follows.

Intel is king... even a short fall from the thrown hurts more than a long fall as a peasant.

Intel stands to lose so much more if they deviate from steady actions in an attempt to strike AMD. Intel has bigger fish to fry. Yes Intel WANTS AMD around picking up it's crumbs. Without AMD, Intel would be classified as a monopoly.... so AMD wouldn't likely go away the government wouldn't let them close their doors. They would be selling off their IP to someone else who may actually pose a threat to Intel's sweet position for desktop CPUs.

Status quo is working out good for Intel.

I would agree with the sentiment that AMD doesn't have an advantage, just a chance. Intel sort of has it's hands tied though if it wants to be steady stock.

Deanjo wrote:

Lol, all this assumption based on enterprise customers (where the money is at in servers) pay retail prices.

Here is a hint: They don't, not even close.

Oh, no, definitely not. I've been consistently shocked at the prices I can get when I order stuff at work.

Deanjo wrote:

Lol, all this assumption based on enterprise customers (where the money is at in servers) pay retail prices.

Here is a hint: They don't, not even close.

Depends on the volume. For high volume, it still leaves plenty of room for AMD to sell at 2/3r's of Intel's average cost for their highest volume tier and make plenty of money. Compared to now.

I'm going to assume 2 million units per quarter sold in that tier, and, that's a pretty substantial hit to Intel given they reduce AMD's price edge to only 4/5th of Intel's.


Of course, those are prices I think anyone here would salivate at paying for to get some high end Xeon's. =]