Do you think there will be a mid gen refresh console from Sony and Microsoft?

[rntongo]

"Meaningful" is pretty subjective, so I'd rather think in terms of absolute cost reductions or historic comparisons.

You can still save in areas other than simple transistor cost with for example reduced power and cooling demands, and smaller cases and lower shipping costs etc, but historically you'd save a lot on the silicon after a shrink plus a lot on power and cooling and have a smaller console to boot.

Cost reductions on consoles where you'd see a device drop to 50% (or less) of its launch price within a few years are a thing of the past. Oh, and that was often with the console maker taking a loss on selling early units too. If you want a sub $250 or $200 console these days you can't just wait a few years (we're already three years into the generation!), you have to make one. Whether it's important to have that is another matter of course.

Edit: just wanted to add that a PS5 Pro would require a more advanced process, but an Xbox refresh might only need N6 - cutting the disk drive, reduced power and cooling etc would all add to a reduced BOM, as might simplifying the memory bus.

Brother you're dancing around the question.I wasnt talking about the higher cost of moving to new nodes compared to previous generations, a fact we all agree on. I was asking if there would be any meaningful cost reductions in the price of silicon from moving to new nodes despite this fact. The empirical evidence says yes since Sony has seen meaningful price reductions in the cost of silicon by moving to 6nm and is moving to N4P next year.

Cost reductions on consoles where you'd see a device drop to 50% (or less) of its launch price within a few years are a thing of the past. Oh, and that was often with the console maker taking a loss on selling early units too. If you want a sub $250 or $200 console these days you can't just wait a few years (we're already three years into the generation!), you have to make one. Whether it's important to have that is another matter of course.

I agree with you on this but you're missing some key information. The PS4 was a cheap console to make from the get go(cheaper than the PS5, PS3, PS2) and launched at $399 after Sony got burnt with the PS3 which launched at $599, So there was more room for significant percentage price reductions as newer process nodes(which we both agree were not as long or as expensive as they are today) became available(The Xbox One had much cheaper SoC but higher costs were with other things but it quickly dropped in price as well to a more realistic selling price of $399 7 months later). As well the PS4 and Xbox One didn't launch during a global pandemic and chip shortage. So if Sony is able to give us a $349 discless PS5 in 2024 and a $499 discless PS5 pro then they've done an excellent job for a machine that launched at $499 during a pandemic and global chip shortage. So I think you need to factor in these two things: Higher initial cost of the machine as well as pandemic and chip shortage.

Edit: just wanted to add that a PS5 Pro would require a more advanced process, but an Xbox refresh might only need N6 - cutting the disk drive, reduced power and cooling etc would all add to a reduced BOM, as might simplifying the memory bus.

I dont think its only the PS5 pro thats going to be on N4P next year. I think they're going to move to N4P across the board including the base PS5.

 

[rntongo]

His 10% number might just be a lesser expensive cooling solution+case size+weight+shipping costs for the PS5

The figure is higher than 10% and its primarily as a result of moving from 7nm to 6nm which resulted in savings in the cost of silicon and lower power consumption enabling them to save money on cooling. Sony was/is getting 20% more PS5 chips per waffer than before and 50% more chips per waffer than MSFT can for the Series X which remained at 7nm. It all comes down to meaningful price reductions brought about by moving to a new process. I dont understand the subtle denialism and dancing around the fact.

You can read more about it here

So I also doubt its only the PS5 pro that will be moving to N4P next year

 

[function]

Brother you're dancing around the question.I wasnt talking about the higher cost of moving to new nodes compared to previous generations, a fact we all agree on. I was asking if there would be any meaningful cost reductions in the price of silicon from moving to new nodes despite this fact. The empirical evidence says yes since Sony has seen meaningful price reductions in the cost of silicon by moving to 6nm and is moving to N4P next year.

But "meaningful" is so nebulous a term it's not particularly "meaningful" to use it in this context unless you can define what meaningful is.

These days, a lot depends on what the node is engineered for. N6 was to shave a bit off cost and power and probably to help towards capacity on more profitable nodes. N5 and N4 are more performance driven.

I agree with you on this but you're missing some key information. The PS4 was a cheap console to make from the get go(cheaper than the PS5, PS3, PS2) and launched at $399 after Sony got burnt with the PS3 which launched at $599, So there was more room for significant percentage price reductions as newer process nodes(which we both agree were not as long or as expensive as they are today) became available(The Xbox One had much cheaper SoC but higher costs were with other things but it quickly dropped in price as well to a more realistic selling price of $399 7 months later). As well the PS4 and Xbox One didn't launch during a global pandemic and chip shortage. So if Sony is able to give us a $349 discless PS5 in 2024 and a $499 discless PS5 pro then they've done an excellent job for a machine that launched at $499 during a pandemic and global chip shortage. So I think you need to factor in these two things: Higher initial cost of the machine as well as pandemic and chip shortage.

Xbox One cut out the very expensive Kinect 2 device which allowed the BOM to drop so fast so early. X1 also had a larger / more expensive die than PS4.

Pandemic and shortage driven prices have long since returned to more normal levels, but console price remains high.

I dont think its only the PS5 pro thats going to be on N4P next year. I think they're going to move to N4P across the board including the base PS5.

We'll see about the PS5, but at the end of the day AMD chose to stay on N6 for lower end RDNA3 products for a reason.

 

[Sonic]

10% is less meaningful than the 40 to 50% reduction in cost for processors, multiple times, that previous generations have experienced. What you consider to be meaningful the rest of us consider to barely be able to reduce the costs.

 

[rntongo]

10% is less meaningful than the 40 to 50% reduction in cost for processors, multiple times, that previous generations have experienced. What you consider to be meaningful the rest of us consider to barely be able to reduce the costs.

But meaningful enough for Sony who is getting 20% more chips per waffer compared to the 7nm Oberon and 50% more chips per waffer than Xbox Series X.

 

[Tkumpathenurpahl]

The figure is higher than 10% and its primarily as a result of moving from 7nm to 6nm which resulted in savings in the cost of silicon and lower power consumption enabling them to save money on cooling. Sony was/is getting 20% more PS5 chips per waffer than before and 50% more chips per waffer than MSFT can for the Series X which remained at 7nm. It all comes down to meaningful price reductions brought about by moving to a new process. I dont understand the subtle denialism and dancing around the fact.

You can read more about it here

So I also doubt its only the PS5 pro that will be moving to N4P next year

The transition to 6nm hasn't impacted power consumption, just the quantity of dies per wafer. The rest of the size reduction is because Sony started cautiously with about 17kg of copper in each PS5, and they've continued to improve the efficiency of their cooling solution the entire time.

As for the transition to 5nm for the base console, I expect that to follow a year or so after the Pro. 5nm is still relatively expensive, and only has finite capacity, so it makes more sense to utilise it for a Pro console with a premium price tag while the base model ticks along just fine at 6nm.

There's also the matter of not wanting to confuse the market with a Pro and Super Slim launching at the same time. They can make a big song and dance about the Pro, while sticking with the cheaper node for the base model until the following Christmas season when they get to make another fuss about an even slimmer base model.

 

[Sonic]

But meaningful enough for Sony who is getting 20% more chips per waffer compared to the 7nm Oberon and 50% more chips per waffer than Xbox Series X.

Unfortunately it isn't meaningful enough for Sony because the price of PS5 hasn't been reduced to $399.

 

[rntongo]

Unfortunately it isn't meaningful enough for Sony because the price of PS5 hasn't been reduced to $399.

They chose to increase their bottom line because of the high demand for the consoles but their long term plan is to bring down the price and increase software and services revenue. Just wait till next year

 

[rntongo]

The transition to 6nm hasn't impacted power consumption, just the quantity of dies per wafer. The rest of the size reduction is because Sony started cautiously with about 17kg of copper in each PS5, and they've continued to improve the efficiency of their cooling solution the entire time.

I agree with most of what you've written as someone that has watched tear downs of the different PS5s you're right. As well, I think the article by the analyst was right as well.

As for the transition to 5nm for the base console, I expect that to follow a year or so after the Pro. 5nm is still relatively expensive, and only has finite capacity, so it makes more sense to utilise it for a Pro console with a premium price tag while the base model ticks along just fine at 6nm.

There's also the matter of not wanting to confuse the market with a Pro and Super Slim launching at the same time. They can make a big song and dance about the Pro, while sticking with the cheaper node for the base model until the following Christmas season when they get to make another fuss about an even slimmer base model.

I find your argument for releasing a 5nm base console a year after the pro very plausible though but not anything beyond a year. If its more economical for Sony to keep the PS5 on 6nm for 1 more year while the PS5 pro is on 5nm then no doubt they'll do it. But the Pro is going to sell out regardless because of GTA 6, especially if they release it in alongside GTA 6. They dont have to worry about that one at all with how much consumers like buying the best new thing with faster performance. But Sony iirc has never called any of their consoles "slim" or "super slim" even when that is exactly what they were. They simply call them PS5 or PS3 or PS4 and stop producing the bigger models. They'll just simply stop producing the bigger PS5(current slim) so there wont be any confusion for customers. I remember the PS4 pro and PS4 slim released around the same time and their sales were good.

 

[Nisaaru]

But meaningful enough for Sony who is getting 20% more chips per waffer compared to the 7nm Oberon and 50% more chips per waffer than Xbox Series X.

And how much more do they pay for the wafer and R&D to adapt the design? Which brings us back to my previous point that the real cost advantage might not come from the chip but the lower TDP enabling a cheaper PS5 redesign.

 

[rntongo]

And how much more do they pay for the wafer and R&D to adapt the design? Which brings us back to my previous point that the real cost advantage might not come from the chip but the lower TDP enabling a cheaper PS5 redesign.

You didn't read the article:
"TSMC ensures that their 6nm EUV process is design rule compatible with N7 process, meaning customers can reuse existing designs made for N7 on N6, with a simple conversion process that doesnt alter the underlying high level synthesis"

I dont have the figures for how much Sony or MSFT pay for their waffers unless someone here has some nice equity research on these contracts(which Equity Research Analysts do have). People quote the 7nm 2018 price of $10k and the 5nm price of $16k. So its up to you to assume what price Sony gets 6nm waffers.

 

[Johnny Awesome]

I'll define "meaningful" in the context of the way MS meant it: 50% cost reduction in 3 years like the good old days. Sony hasn't achieved that. Not even close. Therefore, MS were right.

MS wanted a $250 console 3 years after launch. The only way to get there was the S. Whether it was as important to get to $250 as MS thought it was remains to be seen, but they weren't "lying" about the fab issues.

Will they move to a smaller node if they can save money? Of course! But they're not going to see the cost of the X drop by $50 when they move to 6nm and say to themselves: "Wow, I guess we were wrong about silicon savings being insufficient. These 10% savings are so game changing and meaningful!"

That doesn't mean the S was a great idea. We won't know that for years. My instinct is telling me that it was an "ok" idea. Especially if it means an M could launch later with a library of 500 games.

 

[BitByte]

I'll define "meaningful" in the context of the way MS meant it: 50% cost reduction in 3 years like the good old days. Sony hasn't achieved that. Not even close. Therefore, MS were right.

MS wanted a $250 console 3 years after launch. The only way to get there was the S. Whether it was as important to get to $250 as MS thought it was remains to be seen, but they weren't "lying" about the fab issues.

Unfortunately for MS, I’ve seen the series s going for $138 at Walmart, Costco and the like yet it’s still not selling. With the recent price cuts, I can’t even imagine the amount of money they’re losing per unit. If I had to hazard a guess, I’d say it’s going to be greater than the reported $200/unit.

Will they move to a smaller node if they can save money? Of course! But they're not going to see the cost of the X drop by $50 when they move to 6nm and say to themselves: "Wow, I guess we were wrong about silicon savings being insufficient. These 10% savings are so game changing and meaningful!"

I honestly don’t see the point of them going to 6nm. I don’t think they’ll get the cost efficiencies they want due to their split product lines. Those two chips are different and as it stands, the series sales are awful. Will they really get the volume efficiencies or will they just end up losing more money on hardware.

That doesn't mean the S was a great idea. We won't know that for years. My instinct is telling me that it was an "ok" idea. Especially if it means an M could launch later with a library of 500 games.

I think the series s was a terrible idea. It’ll go down as one of the factors that hampered hardware sales. The series s would have been a great idea if it spawned from defective X chips.

 

[Shifty Geezer]

Brother you're dancing around the question.I wasnt talking about the higher cost of moving to new nodes compared to previous generations, a fact we all agree on. I was asking if there would be any meaningful cost reductions in the price of silicon from moving to new nodes despite this fact.

What counts as meaningful? A 10% reduction on a $120 SOC is $12. 10% is a meaningful cost reduction for a business, but $12 is nothing to the cost for the consumer on a $500 device.

 

[Tkumpathenurpahl]

I agree with most of what you've written as someone that has watched tear downs of the different PS5s you're right. As well, I think the article by the analyst was right as well.

I find your argument for releasing a 5nm base console a year after the pro very plausible though but not anything beyond a year. If its more economical for Sony to keep the PS5 on 6nm for 1 more year while the PS5 pro is on 5nm then no doubt they'll do it. But the Pro is going to sell out regardless because of GTA 6, especially if they release it in alongside GTA 6. They dont have to worry about that one at all with how much consumers like buying the best new thing with faster performance. But Sony iirc has never called any of their consoles "slim" or "super slim" even when that is exactly what they were. They simply call them PS5 or PS3 or PS4 and stop producing the bigger models. They'll just simply stop producing the bigger PS5(current slim) so there wont be any confusion for customers. I remember the PS4 pro and PS4 slim released around the same time and their sales were good.

True. And I think the point you raised about a Pro selling like hotcakes due to GTAVI makes an awful lot of sense. It also makes it more likely, in my opinion, that Sony would want as much 5nm capacity as possible exclusively for the Pro. A few months later, the base : pro metric may be more easily calculable.

As for the naming of the devices: good point. Yeah, Sony don't really do the "Super Slim" naming convention, but the media and broader public still do. So I think there's still value in launching one device/revision per year and giving the market time to chatter about it on social media etc

I think the series s was a terrible idea. It’ll go down as one of the factors that hampered hardware sales. The series s would have been a great idea if it spawned from defective X chips

IMO it was poorly executed purely because of the lack of memory. I'd have loved a Zen3+RDNA2+SSD architecture updated X1X as the XSS, with a big 20-24GB 80CU beast as the XSX. But equal memory in both Series consoles would've been second best. Coulda shoulda woulda though, eh? ¯\_(ツ)_/¯

What counts as meaningful? A 10% reduction on a $120 SOC is $12. 10% is a meaningful cost reduction for a business, but $12 is nothing to the cost for the consumer on a $500 device.

Are you implying that the occasional 1% VAT reduction doesn't dramatically improve your life?! I am shaken.

 

[rntongo]

What counts as meaningful? A 10% reduction on a $120 SOC is $12. 10% is a meaningful cost reduction for a business, but $12 is nothing to the cost for the consumer on a $500 device.

Yes, to the business. As well Sony could have kept the new slimmer digital PS5 at $399 but they chose to up it to $449 in order to increase their profits due to the high demand for their hardware and issues meeting that demand. I strongly believe if they release the pro next year then consumers will get a better price on the base PS5 and feel the benefits of the cost reductions as well.

 

[rntongo]

I think the series s was a terrible idea. It’ll go down as one of the factors that hampered hardware sales. The series s would have been a great idea if it spawned from defective X chips.

Exactly, so I think MSFT is hesitant about releasing a 3rd pro console even if it made sense with titles like GTA 6 coming out soon(with for now a console only launch at the start). 3 consoles in generation is a bit too much.

True. And I think the point you raised about a Pro selling like hotcakes due to GTAVI makes an awful lot of sense. It also makes it more likely, in my opinion, that Sony would want as much 5nm capacity as possible exclusively for the Pro. A few months later, the base : pro metric may be more easily calculable.

As for the naming of the devices: good point. Yeah, Sony don't really do the "Super Slim" naming convention, but the media and broader public still do. So I think there's still value in launching one device/revision per year and giving the market time to chatter about it on social media etc

Yes this makes sense.

IMO it was poorly executed purely because of the lack of memory. I'd have loved a Zen3+RDNA2+SSD architecture updated X1X as the XSS, with a big 20-24GB 80CU beast as the XSX. But equal memory in both Series consoles would've been second best. Coulda shoulda woulda though, eh? ¯\_(ツ)_/¯

Agreed. Another odd thing was MSFT told some highly regarded media houses like DF that the Series X was their pro console(DF only revealed this years later) without informing the public. So their base next gen console(Series S) had less RAM than their previous gen pro 4K console(Xbox One X at 12GB) and less RAM than their "pro" console(Series X 16GB). But the games of the gen on the consoles were primarily being designed around 16GB(14GB after OS). They really put themselves in a pickle with that because their pro console(with larger GPU) has been performing like Sony's base console and because of the lower memory in their base console, they've had some dev issues. I wonder if developers will be able to better utilize the Series X once the PS5 pro is out. I think adding a 3rd pro console would just confuse the public with the Series S, Series X and then the Pro. Imo they should just have program to push their Series X to the limits, if there's any juice to squeeze in it.

 

[Nisaaru]

Unfortunately for MS, I’ve seen the series s going for $138 at Walmart, Costco and the like yet it’s still not selling. With the recent price cuts, I can’t even imagine the amount of money they’re losing per unit. If I had to hazard a guess, I’d say it’s going to be greater than the reported $200/unit.

Do I understand this correctly you think they lose more than 200$ on a Series *S*?

 

[BitByte]

Do I understand this correctly you think they lose more than 200$ on a Series *S*?

I don’t know if it’s the series S that was losing $200 a unit or the x but the news of the losses was reported when the msrp was $299 for the S and $499 for the x. Last I checked, inflation is still well above 2% and now the X can be found for $350 and S for $138. That’s why I guessed that they’re losing greater than $200/unit. If their competitors are selling at full price and we know Sony’s gross margin is like 6% ish, it’s hard to say that MS is not losing a bunch of money on hardware. I mean, Sony raised the price of the ps5 digital to ensure they’re not losing money on the unit.

I know that usually console manufacturers sometimes sell hardware at a loss and make it back in software sales but software sales for the Series is very bad. I don’t know what they’re going to do. You can’t have the series s struggling to sell when it’s priced less than the switch lite.

Loss per unit source

 

[Tkumpathenurpahl]

Agreed. Another odd thing was MSFT told some highly regarded media houses like DF that the Series X was their pro console(DF only revealed this years later) without informing the public. So their base next gen console(Series S) had less RAM than their previous gen pro 4K console(Xbox One X at 12GB) and less RAM than their "pro" console(Series X 16GB). But the games of the gen on the consoles were primarily being designed around 16GB(14GB after OS). They really put themselves in a pickle with that because their pro console(with larger GPU) has been performing like Sony's base console and because of the lower memory in their base console, they've had some dev issues. I wonder if developers will be able to better utilize the Series X once the PS5 pro is out. I think adding a 3rd pro console would just confuse the public with the Series S, Series X and then the Pro. Imo they should just have program to push their Series X to the limits, if there's any juice to squeeze in it.

Aye, it's a tricky position. It may, however, be an opportunity for them to test the waters with a rolling generations model.

Perhaps something like using full 60CU SoC's, pushing the clockspeed as high as 5nm will allow, and pairing them with 18gbps GDDR6 without the split bandwidth memory? That should simplify their manufacturing process, as they could use the same SoC's for the Series X, albeit with disabled CU's and slower, split bandwidth memory.

Developers could still target the Series S if they wanted to (great for indie developers of less technically demanding titles) but I think it would greatly benefit MS to have the Series X as the new baseline for 3rd party AAA's.