Merrill Lynch's Next-Gen console prediction

[Mintmaster]

Mintmaster have you read the thread? You seem not to know what I'm ranting about.

I guess not when I saw this:

What I'm arguing is the bogus BOM numbers here, and how a firm that can put RSX at $75 would put Cell at $248. Do you see my point? My case? It has nothing to do with 360 vs PS3; it's all about PS3 vs PS3 - costwise. These numbers are ridiculous.

I thought you were talking about this, given all your references to ML and how they changed their mind:
http://webpages.charter.net/spartan85/ML_Consoles.pdf
I figured you were complaining about Cell @ $160 compared to RSX @ $100.

So for the purpose of saving me time of reading 12 pages, what was your opinion on the ML report that the thread is about?

As for why Cell won't cost less than RSX, I've given you many reasons:
A) Poorer yields, due to:
- Extra low CPU error rates needed for Cell
- High clock speed (Sony knew that whatever speed they chose, it'll get easier down the road, so 3.2GHz is intentionally not a super easy target)
B) Similar size (see calculations in previous post)
C) Similar volume, because non-PS3 cell sales will not even compare to PS3 sales.

We don't have the data to quantify (A), so it's unreasonable to say one number is way off. CPU's need to be rock solid. They cost a lot more than GPUs of the same size. In video cards, it's unlikely that the GPU costs more than 1/4 of the MSRP, as you have the memory, colling, PCB, connectors, assembly, packaging, etc. to worry about as well.

Use this to compare prices of CPUs to GPUs. A 100 mm2 Prescott is ~$180. Say $90 is profit. I doubt ATI's RV515 GPU, which is also 100 mm2, costs much more than $40, or it would be suicide.

 

[Alpha_Spartan]

I have a new slogan for the Internet: "The Internet...where EVERYONES an Expert! Get connected today!"

BTW, I think I found this Merrill Lynch character:

Busted! That'll teach him to stop making up numbers. You're now exposed Merrill! Stop making shit up!

 

[Carl B]

I guess not when I saw this:


I thought you were talking about this, given all your references to ML and how they changed their mind:
http://webpages.charter.net/spartan85/ML_Consoles.pdf
I figured you were complaining about Cell @ $160 compared to RSX @ $100.

Mintmaster what do you mean all of my references? I think I said one or two things on that matter, completely on the side and not in the vein of the main body of my debate whatsoever.

So for the purpose of saving me time of reading 12 pages, what was your opinion on the ML report that the thread is about?

As for why Cell won't cost less than RSX, I've given you many reasons:
A) Poorer yields, due to:
- Extra low CPU error rates needed for Cell
- High clock speed (Sony knew that whatever speed they chose, it'll get easier down the road, so 3.2GHz is intentionally not a super easy target)
B) Similar size (see calculations in previous post)
C) Similar volume, because non-PS3 cell sales will not even compare to PS3 sales.

We don't have the data to quantify (A), so it's unreasonable to say one number is way off. CPU's need to be rock solid. They cost a lot more than GPUs of the same size. In video cards, it's unlikely that the GPU costs more than 1/4 of the MSRP, as you have the memory, colling, PCB, connectors, assembly, packaging, etc. to worry about as well.

Use this to compare prices of CPUs to GPUs. A 100 mm2 Prescott is ~$180. Say $90 is profit. I doubt ATI's RV515 GPU, which is also 100 mm2, costs much more than $40, or it would be suicide.

The Prescott core is by and large estimated to cost Intel $40 to fab, and here's a link to such for you: Link

I honestly don't think the yields between GPU's and CPU's at the same die size diverge *that* much; maybe we can get some stats on that. As for the 3.2GHz, I'm sure it has as much to do with power and cooling concerns as anything else. Did you check out that schmoo plot I referenced earlier in the thread?

I've given you my reasons already why I feel Cell will be lower priced than RSX, so I'm not going to repeat the same line over and over. (Ok maybe a little)

If you're asking me to answer the problem I have with ML's numbers, it's the same ones I have with Citi's: the numbers just seem all around too divergent and slipshod. You don't need to dwell on GPU's vs CPU's to see the fallacy of their analysis - you have low yields expected, based on the premise of perfect chips and an immature process, and on those grounds alone one can reverse engineer the cost and assume that *if* Merrill (or Citi) even had anything realistic to work with in the first place, defect-tolerence on the SPE's and a yield rate that should approach industry norms soon before or soon after the PS3 launches should lead to a much more 'in line' cost with it's die-size peers, if for nothign other than 'in line' yields.

If you came here to debate on the behalf of the position of ML because they are a 'reputable' firm, then where does Citi stand in your eyes with their divergent numbers? Do you believe at the same process with the same die size Cell (absorbing a defect and in the middle of it's schmoo capabilities) would cost 60% more than the RSX? How about Citi's three times as high numbers? What about Envisioneering Group's $30 estimates?

If you are going to arbitrarily defend one firms numbers, how do you choose which one?

To end, the XeCPU is reported by IBM to get 'good' yields. Since I imagine it would be easier to get Cell chips to 3.2 GHz than a tri-PPE cored chip, and because the Cell chips can tolerate a die defect now and then, and because the fab process is the same, I imagine that yields should at least be similar on one level or another.

 

[Sis]

I look at it this way say the PS3 is $500 who many people will buy one?

Are you seriously restating what the report is saying?

B.) Xbox 360 for $250 (with a HDD)

And then you conveniently ignore that this is a wrong interpretation already well discussed; 250 is for the Xbox 360 without a HDD.


You may want something really badly, that doesn't make it reality. Sony may very badly want to release at the 299 price point--just like PS2. It doesn't mean they can. And just the same: Sony may very badly want to release at the 399 price point, but it doesn't mean they can.

Honestly, I think we'll see a $399 PS3 up against a 250/350 Xbox 360. This doesn't change the fact that if you look at the core model, that's 250 compared to 399, a $150 difference. Add up that it appears not to hurt MS as much financially, and you have Sony at a disadvantage--which is what the report says.

.Sis

 

[Inane_Dork]

Is there any point in putting the HDD on there when it won't be included with the PS3? It seems to me the best comparison would be to compare the PS3 costs with the costs for the core 360 console.

The HDD is included because it's included in the more popular SKU for the 360. It's possible that Sony will have a 2 SKU setup with one of them having an HDD, so just in case, they included it. If not, then just skip that column. No biggie.

 

[nelg]

xbdestroya,

That chart is for PPE clocks only, IIRC.

 

[Carl B]

xbdestroya,

That chart is for PPE clocks only, IIRC.

Well, it's actually SPE clocks.

 

[expletive]

Are you seriously restating what the report is saying?


And then you conveniently ignore that this is a wrong interpretation already well discussed; 250 is for the Xbox 360 without a HDD.


You may want something really badly, that doesn't make it reality. Sony may very badly want to release at the 299 price point--just like PS2. It doesn't mean they can. And just the same: Sony may very badly want to release at the 399 price point, but it doesn't mean they can.

Honestly, I think we'll see a $399 PS3 up against a 250/350 Xbox 360. This doesn't change the fact that if you look at the core model, that's 250 compared to 399, a $150 difference. Add up that it appears not to hurt MS as much financially, and you have Sony at a disadvantage--which is what the report says.

.Sis

Just to add on here, with Sony its a zero-sum game between cost and selling price. You can say sony will sell at whatever you wih but there is direct impact to losses as that sales price goes down in order to match the 360.

What Sony wants to do is obviously sell it for as much as they can without it affecting sales numbers. Maybe there is no limit. Maybe they will sell them all regardless. However, if that isnt the case it seems in terms of price, at least early on, they have to rob peter to pay paul.

 

[mckmas8808]

What Sony wants to do is obviously sell it for as much as they can without it affecting sales numbers. Maybe there is no limit. Maybe they will sell them all regardless. However, if that isnt the case it seems in terms of price, at least early on, they have to rob peter to pay paul.

Yeah and if you keep paying Paul with Peter money, Peter will soon realize it and just seperate itself from the robber. Thus Paul will start to receive less and less money over time.

Honestly, I think we'll see a $399 PS3 up against a 250/350 Xbox 360. This doesn't change the fact that if you look at the core model, that's 250 compared to 399, a $150 difference. Add up that it appears not to hurt MS as much financially, and you have Sony at a disadvantage--which is what the report says.

Well yeah but the report had the PS3 at $500. Something neither you nor me believe will happen as of now (a.k.a. we agree on this). And to me that basically means that in no way can Sony compete in any way with the X360 in price. Not it get close. I simply disagree with that thought. Does the 360 cost less to make? Of course. But will it be $250 less? I don't think so imho.

 

[Mintmaster]

xbd, the only thing I'm disputing now is this:

I've given you my reasons already why I feel Cell will be lower priced than RSX, so I'm not going to repeat the same line over and over.

The only argument I've seen is that:
A) Cell has fewer transistors and is on the same process
B) You made some obscure argument about volume.

Regarding A: # of transistors is irrelevant if you have the die size and yields. The former will be about the same for both. I've already shown you the simple calculations, and you've said nothing about them. The latter we don't have, but I see no reason for Cell to have better yeilds than RSX. There's a stricter testing for CPUs, and higher clock speeds do more for the CPU than the GPU.

Regarding B: Cell will not have an appreciably higher volume than RSX, for the last time. Both are on equal footing. The number of Cell processors sold outside of PS3 is almost meaningless.

As for your link to Intel's cost for a Prescott, remember that $40 includes all the celerons they sell, which are actually cache-defective P4's. Moreover, my number for ATI's RV515 is quite high. TSMC will take their profit as well, so the actual chip can't be more than around $20.

 

[Nesh]

Just read the article.Things may proove as I have expected.

Ken has the same attitude Sega had with the Saturn.And just like the Saturn PS3 is extremely costy compared to XBOX260.To make things even worse for Sony, MS can make easier price cuts.Just as Sony could, back in the PS1 days forcing Sega to reduce their price.PS1 was more competitive both at price and cost.

Now Sony is Sega, PS3 is the Saturn,MS could be Sony, and 360 could be the PS1 of this generation

 

[one]

Sorry if this is already posted,

This is how Merrill Lynch forecasted the PS2 launch in August, 1999.
http://news.zdnet.com/2100-9595_22-515432.html?legacy=zdnn

Report: PlayStation to land Jan. 23
By Robert Lemos, ZDNet News
Published on ZDNet News: August 13, 1999, 5:00 PM PT

Financial firm Merrill Lynch predicts that the PlayStation 2 -- Sony Corp.'s next-generation game machine -- will hit Japanese shelves on Jan. 23 at a hefty price of 45,000 yen (U.S. $391), according to a report published this week.

Even at that price, Sony will barely cover the cost of the parts needed to build the machine, ZDNN reported in a previous article. In addition, the machine will have to contend with Sega Enterprises Ltd.'s own game machine, the Dreamcast, which has been available in Japan for almost a year.

While Dreamcast machine will hit U.S. shelves Sept. 9 at a much more affordable $199, analysts have predicted that the PlayStation 2 will become the giant in the market.

Merrill Lynch's outlook is also optimistic. The report pegs initial shipments of the PlayStation 2 at 400,000 units with a total of 1 million consoles sold during the first three months of 2000.

Aggressive plans
Those numbers mesh with the agreements forged between Sony and its partners. For example, Sony has contracted Toshiba to manufacture the console's main processor to the tune of 1 million processors by Dec. 1999, and another million in the first quarter of 2000.

The aggressive plans will drain the company's coffers, according to the Merrill Lynch report, which predicted that Sony's PlayStation division will post a $110 million loss in the year ending March 2000.

Those losses will be temporary if demand picks up, transforming into a $130 million profit by the following year. High software prices will contribute to the turnaround, with the 10 initial titles at an average price of 8,000 yen (U.S. $70) each.

The report said to expect more details at the Tokyo Game Show in September.

Now let's compare them with the actual stats. The exchange rate at the point of the above article was 1$= 115 yen (45,000 / 391). The rate today is 1$ = 117 yen, so not very different.

PS2 launched in Japan on March 4, 2000 for 39,800 yen. The average price of launch titles is 6,800 yen (or lower as the most expensive was at 6,800 yen).

For production shipments,
http://www.scei.co.jp/corporate/data/bizdataps2_e.html

2000/03/06 720,000 units (Japan only)
2000/03/15 1 million units (Japan only)
2000/03/31 1.41 million units (Japan only)
2000/05/24 2 million units (Japan only)

BTW, PS2 was launched in the US on October 26, 2000 at $299.99 with an HDD bay.

 

[Carl B]

The only argument I've seen is that:
A) Cell has fewer transistors and is on the same process
B) You made some obscure argument about volume.

Regarding A: # of transistors is irrelevant if you have the die size and yields. The former will be about the same for both. I've already shown you the simple calculations, and you've said nothing about them. The latter we don't have, but I see no reason for Cell to have better yeilds than RSX. There's a stricter testing for CPUs, and higher clock speeds do more for the CPU than the GPU.

Regarding B: Cell will not have an appreciably higher volume than RSX, for the last time. Both are on equal footing. The number of Cell processors sold outside of PS3 is almost meaningless.

As for your link to Intel's cost for a Prescott, remember that $40 includes all the celerons they sell, which are actually cache-defective P4's. Moreover, my number for ATI's RV515 is quite high. TSMC will take their profit as well, so the actual chip can't be more than around $20.

I think you and I are at an impass of sorts. You keep mentioning my commentary on Cell and volume... but I haven't given any. What in the world are you debating me about with your 'point B'? Have I mentioned Cell's volume even once with you? When did I say it would have an 'appreciably' larger volume than RSX? Any comments I made regarding Cell in TV's was simply to aggravate Powderkeg, and I only made one. In my prmary argument I simply used the point that East Fishkill will be producing Cell in addition to Nagasaki and OTSS to indicate that Cell will have more collective expertise fabbing it (and yes more volume to an extent, but not a focul point). RSX has only Nagsaki and OTSS.

On your other point, well honestly I just don't know - do GPU's have better yields, so to speak? If you could link me to any information indcative of such, I would certanly take the time to read it.

I'm not sure why you started debating with me in the first place t be honest, or what we're debating about at this point, but if it was simply to counter my assessment of the Cell costs vs RSX costs to be overly optimisitc, well we'll agree to both understand each others positions.

But now with all of that said, what I want to know is to what extent do you agree with the BOM numbers of ML and Citi, respectively? Becase even though I can undertstand you disagreeing with me, I can't full well believe you wouldn't see how their numbers seem *highly* suspect . Beyond their numbers not agreeing with each other in the least, either...

 

[Nesh]

Sorry if this is already posted,

This is how Merrill Lynch forecasted the PS2 launch in August, 1999.
http://news.zdnet.com/2100-9595_22-515432.html?legacy=zdnn


Now let's compare them with the actual stats. The exchange rate at the point of the above article was 1$= 115 yen (45,000 / 391). The rate today is 1$ = 117 yen, so not very different.

PS2 launched in Japan on March 4, 2000 for 39,800 yen. The average price of launch titles is 6,800 yen (or lower as the most expensive was at 6,800 yen).

For production shipments,
http://www.scei.co.jp/corporate/data/bizdataps2_e.html

2000/03/06 720,000 units (Japan only)
2000/03/15 1 million units (Japan only)
2000/03/31 1.41 million units (Japan only)
2000/05/24 2 million units (Japan only)

BTW, PS2 was launched in the US on October 26, 2000 at $299.99 with an HDD bay.

You dont know how much costy to manufacture PS2 was back then compared to how costy PS3 will be to manufacture though.Also add the aditional manucaturing cost of the BR disc which could reduce revenues per game sold and the fact that developers need more revenues per game sold since games are a lot more costy to create and Sony could end up in trouble.

 

[one]

You dont know how much costy to manufacture PS2 was back then compared to how costy PS3 will be to manufacture though.Also add the aditional manucaturing cost of the BR disc which could reduce revenues per game sold and the fact that developers need more revenues per game sold since games are a lot more costy to create.

Do Merrill Lynch know those details? I don't think so. What I know is, Microsoft own no fabs while Sony own some.

 

[Nesh]

Do Merrill Lynch know those details? I don't think so. What I know is, Microsoft own no fabs while Sony own some.

Probably he doesnt but neither do we

You are right about your last comment though.Thats something I ommited.

 

[Powderkeg]

Sorry if this is already posted,

This is how Merrill Lynch forecasted the PS2 launch in August, 1999.
http://news.zdnet.com/2100-9595_22-515432.html?legacy=zdnn

I see something you apparently didn't notice.

They said the $391 would barely cover the cost of the parts. Now, I don't have proof, but I'm fairly sure that that figure is pretty close to right. We all know the PS2 sold for a loss at launch, and I don't think a <$91 loss per system is unreasonable.

 

[one]

I see something you apparently didn't notice.

Eh? They said the launch would be January 2000. Don't you see the price is closely related to the launch date?

 

[scooby_dooby]

Sorry if this is already posted,

This is how Merrill Lynch forecasted the PS2 launch in August, 1999.
http://news.zdnet.com/2100-9595_22-515432.html?legacy=zdnn


Now let's compare them with the actual stats. The exchange rate at the point of the above article was 1$= 115 yen (45,000 / 391). The rate today is 1$ = 117 yen, so not very different.

PS2 launched in Japan on March 4, 2000 for 39,800 yen. The average price of launch titles is 6,800 yen (or lower as the most expensive was at 6,800 yen).

So 7 months before launch they predicted the launch price within 12%, not bad.

They also nailed the X360 launch price, and 2 SKU's well before the announcement BTW.

 

[scooby_dooby]

You implied there that CELLs outcome could be like the EE, which are in two completely different leagues.

I said it may very well end up with the same fate as the EE, something you need to realize is a very real and distinct possibility before you start calling these reports "garbage" and insulting people who try and point that possibility out.

If you listened to analysts, and news reports in 2000, EE and it's super-computer like power was going to power many things beyond just game machines. It didn't happen.