StarTopic Future Nintendo Hardware & Technology Speculation & Discussion |ST| (New Staff Post, Please read)

[Hermii]

Nintendo gets more and more reasons to not design their hardware with third parties in mind.

 

[ShadowFox08]

Switch 2 for 2022 not being a thing makes more sense now than before. From transistor chip shortages, to Nintendo likely still selling well (as long as they have the 1st party software), to PS4 and xbone still be supported in games for another year, to saving costs. .

Spoiler

Although I'm 100% ready for a fall 2022 release

Y’all crazy. Look I understand we might get an MSRP bump at launch over $300, but no way in hell will they try and sell a $500 Switch successor.

You think enough people would buy a higher SKU model with 512GB for $500 . Sure it's $650 on steam deck, but Switch 2 probably won't be coming out until March 2023-Nov 2023, so price will drop. Not to mention Nitnendo will likely get a better deal from Nvidia, and it doesn't have to be the same SSD.

 

[Terrell]

Who says it will be sold as a successor?

At this point, they’ll want it to sell precisely because production is already more expensive than necessary due to the chip shortage. The alternative of iterative upgrades (see: PS4 Pro, New 3DS) only typically sell at about a 20% rate of the hardware line they’re a part of as opposed to a successor, which it is able to be due to the disparity in compute power it will be able to achieve from Switch. Nintendo isn’t spending all this money on R&D (and likely extra for manufacturing due to current circumstances) for a system with this kind of capacity to cut its legs out from under it at retail.

 

[ReddDreadtheLead]

Switch 2 for 2022 not being a thing makes more sense now than before. From transistor chip shortages, to Nintendo likely still selling well (as long as they have the 1st party software), to PS4 and xbone still be supported in games for another year, to saving costs. .

Spoiler

Although I'm 100% ready for a fall 2022 release

You think enough people would buy a higher SKU model with 512GB for $500 . Sure it's $650 on steam deck, but Switch 2 probably won't be coming out until March 2023-Nov 2023, so price will drop. Not to mention Nitnendo will likely get a better deal from Nvidia, and it doesn't have to be the same SSD.

Transistor chip shortage? There’s no shortage of transistors lol

 

[Skittzo]

At this point, they’ll want it to sell precisely because production is already more expensive than necessary due to the chip shortage. The alternative of iterative upgrades (see: PS4 Pro, New 3DS) only typically sell at about a 20% rate of the hardware line they’re a part of as opposed to a successor, which it is able to be due to the disparity in compute power it will be able to achieve from Switch. Nintendo isn’t spending all this money on R&D (and likely extra for manufacturing due to current circumstances) for a system with this kind of capacity to cut its legs out from under it at retail.

Or they'll sell it as both.

Position it as an iterative upgrade at first and then discontinue the base model a few years down the line so that it becomes the de facto successor.

 

[Thraktor]

Or they'll sell it as both.

Position it as an iterative upgrade at first and then discontinue the base model a few years down the line so that it becomes the de facto successor.

I'm kind of expecting a GameBoy Color situation, where Nintendo don't actually explicitly call it a successor (and count it alongside Switch sales), but for all intents and purposes it's a successor.

 

[Skittzo]

I'm kind of expecting a GameBoy Color situation, where Nintendo don't actually explicitly call it a successor (and count it alongside Switch sales), but for all intents and purposes it's a successor.

Yeah this is what I think too. They'll let journalists decide what exactly to call it but for all intents and purposes it'll essentially be both an upgrade and a successor.

 

[Terrell]

I'm kind of expecting a GameBoy Color situation, where Nintendo don't actually explicitly call it a successor (and count it alongside Switch sales), but for all intents and purposes it's a successor.

Considering how long it existed before it was replaced, we should call Game Boy Color what it really was: a stopgap.

Or they'll sell it as both.

Position it as an iterative upgrade at first and then discontinue the base model a few years down the line so that it becomes the de facto successor.

OK, so what’s the impetus for doing this? Why has this not been a part of the game industry business model since the beginning If it’s so advantageous to do it this way? What are the potential financial positives for Nintendo that offset the potential financial negatives that I mentioned? What technological limitations will be imposed to make it “part of the Switch family”, if any? How does a distinction without a difference like what you’ve described benefit anyone?

I’ve never received any legitimate answers to these questions, and it’s not because I stop asking them.

 

[Dekuman]

I'm kind of expecting a GameBoy Color situation, where Nintendo don't actually explicitly call it a successor (and count it alongside Switch sales), but for all intents and purposes it's a successor.

Yeah this is what I think too. They'll let journalists decide what exactly to call it but for all intents and purposes it'll essentially be both an upgrade and a successor.

I kind of doubt a company as particular and controlling as Nintendo would do this.

Also bringing up Game Boy Color brings with it a particular set of circumstances (shelving of the 32-bit Atlantis included) that I don't think quite fits to Switch. For one Game Boy tech was fairly old by the time Color came out and GBC was effectively the same Game Boy tech with more VRAM, doubled clocks and doubled WRAM and even then Nintendo didn't miss the fact that it was color in the branding. While they counted its sales towards GB's total, i would argue it was positioned an upgrade even if successor may not fit.

A similar positioning for the next Switch would have made more sense if it was based on the X1 design from 2015 with everything more or less doubled with some kind of 4K upscaler for TVs. Given what we know of the Dane SoC it seems like a true successor, unless that tech gets shelved like the Atlantis and we go back to the worst timeline of a die shrunk X1 (lol).

I'm not saying they will explicitly engage in generaitonal talk like sony has, but if they do a low key upgrade , it will be more like Apple's strategy of clear cut generations within its product lines but one that is largely detatched from what others are doing. So a Gen 2 Switch will be clearly marketed as such. Not sure if they will include or exclude its sales as part of the Switch Lifetime but that is i think a less important conversatioon.

 

[Skittzo]

OK, so what’s the impetus for doing this? Why has this not been a part of the game industry business model since the beginning If it’s so advantageous to do it this way? What are the potential financial positives for Nintendo that offset the potential financial negatives that I mentioned? What technological limitations will be imposed to make it “part of the Switch family”, if any? How does a distinction without a difference like what you’ve described benefit anyone?

I’ve never received any legitimate answers to these questions, and it’s not because I stop asking them.

I'm pretty sure I've answered a number of those over at install base. And to be clear this idea isn't very different from what Sony has done since they entered the console business, it's just new for Nintendo home consoles.

OK, so what’s the impetus for doing this?

Nintendo has never had a single platform with one software environment before. Continuing the Switch family keeps current Switch owners engaged and allows current Switch developers to tap into that same base.

Why has this not been a part of the game industry business model since the beginning If it’s so advantageous to do it this way?

...it has. Again this is not very different from what Sony does, the only real difference is marketing strategy. Plus Nintendo did it with the GBC...

What are the potential financial positives for Nintendo that offset the potential financial negatives that I mentioned?

What negatives? The positives would be the ability to price it like a premium product at launch which the enthusiasts would gladly pay. And again it would help solidify the Switch concept as a family of devices that everyone with any model is a part of, which helps older Switch games and developers retain a more engaged audience.

The current climate for consumer electronics would seem to indicate that there is very little risk of this product selling poorly (i.e. not selling out for months). I don't see any negatives here.

What technological limitations will be imposed to make it “part of the Switch family”, if any?

None? We're all talking about the same device here, this distinction is purely about marketing and messaging. Dane is a known factor at this point, that's not going to change.

How does a distinction without a difference like what you’ve described benefit anyone?

I've answered this already.

 

[Thraktor]

Yeah this is what I think too. They'll let journalists decide what exactly to call it but for all intents and purposes it'll essentially be both an upgrade and a successor.

One thing I wonder is about pricing and profit margin. Typically if you're releasing a new generation console you're either selling it at about break even (Nintendo), or at a loss (MS and Sony), and work towards increased profitability over the course of the generation. I believe MS and Sony would usually move from a moderate loss to a small profit, and I would imagine that both the PS4 Pro and Xbox Series X were sold at a small profit. There's no point taking losses on hardware late into a generation, and I'm sure Nintendo made reasonable profits on both the DSi and new 3DS when they launched.

The Switch is in a weird place, though, as it was launched at around break-even, and is still being sold 5 years later for the same price, so they're obviously making a very comfortable profit on the hardware by this point, with or without the chip shortage. If they're releasing a 'pro' model of the Switch, then logically they wouldn't want to sell it for a significantly lower profit margin that the current model, which means designing something that, at say $350-$400, sells for a comfortable profit.

If they're releasing a new generation device, though, they're going to want to price much closer to break-even at launch to bring in early adopters, and then rely on cost reductions to increase profitability over time if they can. They could try launching a new generation with higher margins, but that really didn't work out well for them with the 3DS, so I wouldn't expect a repeat any time soon.

Which means that the distinction would actually be a pretty important one in terms of the hardware they can use in the device. A $400 device designed around a similar profit margin to the current Switch would look very different to a $400 device designed around selling break-even as a new generation console. In fact, I would suspect we're actually comparing a higher-priced (maybe $400) less powerful "pro" console to a lower-priced (possibly $350) more powerful new generation, given the target markets of each device.

One of the questions behind this would be how long do they expect to keep making Switch style consoles? If they release a "pro" model, then given the age of the Switch, I would imagine a new generation would be only around 2 years or so behind it, but it really doesn't seem like they would release a Dane-powered pro model in 2022, then a Switch 2 with another new Nvidia SoC just two years later. It would be like if the DSi were closer architecturally to the 3DS than the DS. If they do release a "pro" model in the next year or so, then honestly my expectation is that we wouldn't even see a Switch 2, it's probably intended to extend the life-cycle of the Switch while they prepare something completely different.

If they release a GBC-style midway point then I'd still say it's a sign that they've got something different planned, which incidentally was the case with the GBC, as the GBA came out only around 2 and a half years later. Perhaps they'd get a bit more time out of it than a "pro" model, but that really depends on how customers react. If they simply see it as a pro model (which is, thanks to Sony and MS, now pretty much the norm for gaming consoles), then adoption might not amount to enough to keep things going for very long.

I know that trying to apply my own sense of logic to Nintendo's decisions is a futile process, but to me it seems logical that Nintendo should want to keep the Switch family of systems around for as long as possible. Moving from separate home consoles and handhelds has been a very good move in many ways, but it does amplify the risk that goes along with new system launches. Previously if they launched a new console that was unsuccessful, like the GameCube, they could still remain profitable if the other half of their hardware business (the GBA) was doing well. This gave them the freedom to try things like the Wii (and unfortunately also the Wii U). They don't have that leeway any more, and trying to reinvent the wheel after the extremely successful Switch would be extraordinarily risky. So, logically, they should release a successor to Switch with the same form-factor and with backwards compatibility (ie a Switch 2, whether they call it that or not).

Therefore, to me, it wouldn't really make sense for them to release anything but a Switch 2 (in function if not in name) at this stage. With the earliest launch 5 and a half years after the original Switch, and 6+ years a possibility, it's too late for a Pro, unless it's a short term solution while they try something completely different (which would be crazy following the Switch). A Switch 2 released around 5 and a half to 6 years after the original, with a slow cross-gen transition of software over to the new console, seems a very sensible approach to transition over to a new generation as the original Switch winds down. It also would seem a waste to me to take a generational leap in architecture, with both A78 CPUs and Ampere, and use them up in a pro model with a small marketshare and only a handful of exclusives.

So, if it were me, I'd take what technology I had available to me, and make the best device I could for a roughly break-even ~$350 selling price, treat it as a new generation console, and try to really push new buyers onto it.

There's one big gap between that thought, though, and the information we have available to us. Namely, 8nm. For a $350 or $400 device with a healthy profit margin, an 8nm SoC would be fine, and an 8nm Ampere SoC could definitely give us nicer looking Switch games, plus 4K output thanks to DLSS within a Switch form-factor and thermals.

For a device launching perhaps in 2023, at around break-even for over $300, though, 8nm is a pretty serious restriction. It's old. It's far from state of the art, much further, in fact, than the 20nm TX1 was when Switch launched. The first TSMC 20nm chips launched in early 2015, so Nintendo was behind by about 2 years, and 20nm was actually state of the art when it launched, at least (not a great state of the art, but the best node available in 2015). FinFET 16nm/14nm based devices had only started arriving around a year before Switch launched, so 20nm, while not ideal, was far from ancient.

Even with the most optimistic launch of late 2022, though, 8nm is getting pretty ancient. The first Samsung 8nm chip was the Exynos 9820, which launched in late 2018, a full 4 years before our best case for the new Switch. Even then, 8nm is an evolution of Samsung's 10nm process, which saw its first chips in early 2017, only shortly after the original Switch. To put this in perspective, if Nintendo were to use a 2-year old manufacturing process (like they did for the original Switch) in late 2022, they'd be using TSMC 5nm, which is an absolutely enormous leap over Samsung 8nm.

Now, it's true that the progression in manufacturing technology has slowed a lot in the past few years. Post 28nm, we haven't seen the automatic jumps to new process nodes on a regular cadence, for a variety of reasons. It's also true that the high demand over the past two years means we haven't seen the same cost reductions we may otherwise have. Still, 8nm is a much more conservative choice of manufacturing process than the original Switch, and the choice of manufacturing process is especially important for a device like the Switch, where power efficiency of the SoC is the single biggest design limit on performance.

If we look at comparable devices, TSMC's 6nm would seem the most obvious choice at the moment. It's an evolution of their 7nm EUV process (which has been in commercial products since 2019, and is itself based on the non-EUV 7nm which was seen in products back in 2018), so it's hardly bleeding edge, and it seems TSMC is pushing new 7nm clients over to the 6nm variant. It's also a significant improvement over Samsung 8nm in both transistor density and power efficiency, to the point that I'd probably expect 50% more performance if you were to tell me Dane were build on TSMC 6nm vs Samsung 8nm.

The TSMC 7nm/6nm nodes are also used in devices at a similar price range to the new Switch. The Steam Deck is the obvious example, and although it doesn't come with a dock, it does use a huge 16GB of LPDDR5, and is made in much smaller volumes than Switch. If we look at phones, the mid-range 6nm Snapdragon 788G (which, incidentally, uses A78 CPU cores) could be found in phones from around the $300 price point (eg this, this and this) a year before the soonest we expect the new Switch to launch. We've also seen Mediatek's 6nm Dimensity 810 SoC in the €200 Poco M4 Pro 5G, which, as the name suggests also contains a 5G modem, not to mention a 50MP camera, while being sold for a profit.

Basically, the idea of them releasing a proper successor, at close to break-even pricing, doesn't really coalesce with the rumour that Dane is being manufactured on Samsung's 8nm process. Kopite7kimi is the source of the 8nm claim, though, and he's about as reliable as we can get for Nvidia rumours. It is still a rumour, and it's possible he could be wrong, but it would seem quite unlikely (particularly as he correctly claimed Orin would also be 8nm in the same tweet). They could in theory go with a larger chip on the cheaper 8nm node, but as I've said before, in practice there's a limit to this as the power consumption becomes too high for portable mode at base clocks.

It's also possible that they don't agree with my logic, and either Dane is being used in a "pro" model which will be followed by something very different in a few years, or they're releasing a new generation device at a comfortable profit margin. Both of which seem like pretty risky ideas when you've only got a single hardware line to rely on, but then again I don't think I've ever successfully predicted Nintendo's movements in the hardware space before, so who am I to know?

 

[ILikeFeet]

They'll let journalists decide

never do this

 

[Skittzo]

One thing I wonder is about pricing and profit margin. Typically if you're releasing a new generation console you're either selling it at about break even (Nintendo), or at a loss (MS and Sony), and work towards increased profitability over the course of the generation. I believe MS and Sony would usually move from a moderate loss to a small profit, and I would imagine that both the PS4 Pro and Xbox Series X were sold at a small profit. There's no point taking losses on hardware late into a generation, and I'm sure Nintendo made reasonable profits on both the DSi and new 3DS when they launched.

The Switch is in a weird place, though, as it was launched at around break-even, and is still being sold 5 years later for the same price, so they're obviously making a very comfortable profit on the hardware by this point, with or without the chip shortage. If they're releasing a 'pro' model of the Switch, then logically they wouldn't want to sell it for a significantly lower profit margin that the current model, which means designing something that, at say $350-$400, sells for a comfortable profit.

If they're releasing a new generation device, though, they're going to want to price much closer to break-even at launch to bring in early adopters, and then rely on cost reductions to increase profitability over time if they can. They could try launching a new generation with higher margins, but that really didn't work out well for them with the 3DS, so I wouldn't expect a repeat any time soon.

Which means that the distinction would actually be a pretty important one in terms of the hardware they can use in the device. A $400 device designed around a similar profit margin to the current Switch would look very different to a $400 device designed around selling break-even as a new generation console. In fact, I would suspect we're actually comparing a higher-priced (maybe $400) less powerful "pro" console to a lower-priced (possibly $350) more powerful new generation, given the target markets of each device.

One of the questions behind this would be how long do they expect to keep making Switch style consoles? If they release a "pro" model, then given the age of the Switch, I would imagine a new generation would be only around 2 years or so behind it, but it really doesn't seem like they would release a Dane-powered pro model in 2022, then a Switch 2 with another new Nvidia SoC just two years later. It would be like if the DSi were closer architecturally to the 3DS than the DS. If they do release a "pro" model in the next year or so, then honestly my expectation is that we wouldn't even see a Switch 2, it's probably intended to extend the life-cycle of the Switch while they prepare something completely different.

If they release a GBC-style midway point then I'd still say it's a sign that they've got something different planned, which incidentally was the case with the GBC, as the GBA came out only around 2 and a half years later. Perhaps they'd get a bit more time out of it than a "pro" model, but that really depends on how customers react. If they simply see it as a pro model (which is, thanks to Sony and MS, now pretty much the norm for gaming consoles), then adoption might not amount to enough to keep things going for very long.

I know that trying to apply my own sense of logic to Nintendo's decisions is a futile process, but to me it seems logical that Nintendo should want to keep the Switch family of systems around for as long as possible. Moving from separate home consoles and handhelds has been a very good move in many ways, but it does amplify the risk that goes along with new system launches. Previously if they launched a new console that was unsuccessful, like the GameCube, they could still remain profitable if the other half of their hardware business (the GBA) was doing well. This gave them the freedom to try things like the Wii (and unfortunately also the Wii U). They don't have that leeway any more, and trying to reinvent the wheel after the extremely successful Switch would be extraordinarily risky. So, logically, they should release a successor to Switch with the same form-factor and with backwards compatibility (ie a Switch 2, whether they call it that or not).

Therefore, to me, it wouldn't really make sense for them to release anything but a Switch 2 (in function if not in name) at this stage. With the earliest launch 5 and a half years after the original Switch, and 6+ years a possibility, it's too late for a Pro, unless it's a short term solution while they try something completely different (which would be crazy following the Switch). A Switch 2 released around 5 and a half to 6 years after the original, with a slow cross-gen transition of software over to the new console, seems a very sensible approach to transition over to a new generation as the original Switch winds down. It also would seem a waste to me to take a generational leap in architecture, with both A78 CPUs and Ampere, and use them up in a pro model with a small marketshare and only a handful of exclusives.

So, if it were me, I'd take what technology I had available to me, and make the best device I could for a roughly break-even ~$350 selling price, treat it as a new generation console, and try to really push new buyers onto it.

There's one big gap between that thought, though, and the information we have available to us. Namely, 8nm. For a $350 or $400 device with a healthy profit margin, an 8nm SoC would be fine, and an 8nm Ampere SoC could definitely give us nicer looking Switch games, plus 4K output thanks to DLSS within a Switch form-factor and thermals.

For a device launching perhaps in 2023, at around break-even for over $300, though, 8nm is a pretty serious restriction. It's old. It's far from state of the art, much further, in fact, than the 20nm TX1 was when Switch launched. The first TSMC 20nm chips launched in early 2015, so Nintendo was behind by about 2 years, and 20nm was actually state of the art when it launched, at least (not a great state of the art, but the best node available in 2015). FinFET 16nm/14nm based devices had only started arriving around a year before Switch launched, so 20nm, while not ideal, was far from ancient.

Even with the most optimistic launch of late 2022, though, 8nm is getting pretty ancient. The first Samsung 8nm chip was the Exynos 9820, which launched in late 2018, a full 4 years before our best case for the new Switch. Even then, 8nm is an evolution of Samsung's 10nm process, which saw its first chips in early 2017, only shortly after the original Switch. To put this in perspective, if Nintendo were to use a 2-year old manufacturing process (like they did for the original Switch) in late 2022, they'd be using TSMC 5nm, which is an absolutely enormous leap over Samsung 8nm.

Now, it's true that the progression in manufacturing technology has slowed a lot in the past few years. Post 28nm, we haven't seen the automatic jumps to new process nodes on a regular cadence, for a variety of reasons. It's also true that the high demand over the past two years means we haven't seen the same cost reductions we may otherwise have. Still, 8nm is a much more conservative choice of manufacturing process than the original Switch, and the choice of manufacturing process is especially important for a device like the Switch, where power efficiency of the SoC is the single biggest design limit on performance.

If we look at comparable devices, TSMC's 6nm would seem the most obvious choice at the moment. It's an evolution of their 7nm EUV process (which has been in commercial products since 2019, and is itself based on the non-EUV 7nm which was seen in products back in 2018), so it's hardly bleeding edge, and it seems TSMC is pushing new 7nm clients over to the 6nm variant. It's also a significant improvement over Samsung 8nm in both transistor density and power efficiency, to the point that I'd probably expect 50% more performance if you were to tell me Dane were build on TSMC 6nm vs Samsung 8nm.

The TSMC 7nm/6nm nodes are also used in devices at a similar price range to the new Switch. The Steam Deck is the obvious example, and although it doesn't come with a dock, it does use a huge 16GB of LPDDR5, and is made in much smaller volumes than Switch. If we look at phones, the mid-range 6nm Snapdragon 788G (which, incidentally, uses A78 CPU cores) could be found in phones from around the $300 price point (eg this, this and this) a year before the soonest we expect the new Switch to launch. We've also seen Mediatek's 6nm Dimensity 810 SoC in the €200 Poco M4 Pro 5G, which, as the name suggests also contains a 5G modem, not to mention a 50MP camera, while being sold for a profit.

Basically, the idea of them releasing a proper successor, at close to break-even pricing, doesn't really coalesce with the rumour that Dane is being manufactured on Samsung's 8nm process. Kopite7kimi is the source of the 8nm claim, though, and he's about as reliable as we can get for Nvidia rumours. It is still a rumour, and it's possible he could be wrong, but it would seem quite unlikely (particularly as he correctly claimed Orin would also be 8nm in the same tweet). They could in theory go with a larger chip on the cheaper 8nm node, but as I've said before, in practice there's a limit to this as the power consumption becomes too high for portable mode at base clocks.

It's also possible that they don't agree with my logic, and either Dane is being used in a "pro" model which will be followed by something very different in a few years, or they're releasing a new generation device at a comfortable profit margin. Both of which seem like pretty risky ideas when you've only got a single hardware line to rely on, but then again I don't think I've ever successfully predicted Nintendo's movements in the hardware space before, so who am I to know?

IIRC the 3DS did not have a comfortable profit margin at $250, it was more or less sold at their typical minor profit. The price cut really harmed their financials for a few years. Whatever the reason they really did seem to overspend on the 3DS BoM.

But when it comes to whatever Dane is, I think the point I'm trying to make is a "pro" style upgrade and a generational upgrade don't have to be different concepts. Nintendo's primary underlying strategy with the Switch was to unify the software environment for all of their products, so you can play ~99% of the same library on everything they release. So far this has extended to the base Switch, the Lite and the OLED model.

There's no reason they can't then sell a 4k model that has the same software environment. It will be able to play games the others can't play but the hybrid is also capable of playing games the Lite can't play, so it fits with the current patterns.

It wouldn't be positioned as a pro, and it wouldn't be positioned as a Switch 2. It would be another member in the Switch family, one which would- in the long run- serve the purpose of a generational upgrade. Maybe 3-4 years after launch when no more base Switches are being made.

The reason they would want to do this is to have a much higher point of entry when it comes to revenue. I do believe they'd go for a comfortable profit margin for this concept, maybe as high as $450 or even $500.

The reason it works is because the current environment for consumer electronics is insatiable. People would buy a Switch that can play all your existing games in 4k plus new ones, especially early adopters who tend to be enthusiasts. Even at $500 I can't see it not being sold out in most major markets for months, at least if it is launched relatively soon (i.e. within the next ~15 months).

The long term goal then would be to slowly position it as a new base model that will be 2025's default Switch, for instance. This would involve lowering the price or releasing a slightly lower end model with the same Dane chip, but for example with less storage.

This is how they avert the increased risk that comes with having only one active platform- they don't deactivate it until the next platform has a userbase that can sustain itself. This is done by positioning not as the "next platform" but as just another member of the current platform family.


I also think this strategy does jive with the idea that it's using an 8nm process. It doesn't need to be cutting edge when it will still be enough of an upgrade to greatly improve existing Switch games which will be the primary driver early on.

 

[Dakhil]

To put this in perspective, if Nintendo were to use a 2-year old manufacturing process (like they did for the original Switch) in late 2022, they'd be using TSMC 5nm, which is an absolutely enormous leap over Samsung 8nm.

Another possibility is Samsung's 5LPE process node or Samsung's 5LPP process node.
Although EUV lithography instead of DUV lithography is being used, one potential advantage of using Samsung's 5LPE process node or Samsung's 5LPP process node instead of using TSMC's N5 process node is that Nintendo and Nvidia don't need to completely redesign Dane from scratch since I imagine Nintendo and Nvidia would still be using Samsung's IP when designing Dane, assuming kopite7kimi's still correct about Dane being based on Samsung's 8N process node.
Another potential advantage is that Nvidia would have an easier time securing capacity for Samsung's 5LPE process node or Samsung's 5LPP process node in comparison to TSMC's N5 process node since I imagine Samsung's 5LPE process node and Samsung's 5LPP process node are not as high in demand as TSMC's N5 process node, although probably in much higher demand in comparison to Samsung's 8N process node, especially with the possibility that Nvidia decides to have mid-range and/or entry-level Lovelace GPUs fabricated using Samsung's 5LPE process node or Samsung's 5LPP process node.
Of course, one potentially serious caveat to using Samsung's 5LPE process node or Samsung's 5LPP process node is that the yields for Samsung's 5LPE process node or Samsung's 5LPP process node is probably not as good as the yields for TSMC's N5 process node, especially if there are grains of truth to Business Korea's report. And I'm inclined to believe there are large grains of truth to Business Korea's report, considering a Xiaomi executive mentioned that the reason the Snapdragon 780G was discontinued relatively quickly was that there was not enough capacity, which does seem to strongly validate Business Korea's report.

I'd have said Samsung's 7LPP process node's also a possibility. However, Andrei Frumusanu mentioned that the binning of the Exynos 990 units Anandtech received are apparently, with the binning of most Exynos 990 units out there being as bad, or worse. And of course, Samsung's 7LPP process node is 20-30% less power efficient in comparison to TSMC's N7 process node.
The only chips publicly revealed and rumoured to be fabricated using Samsung's 7LPP process node is the IBM Telum processor, IBM Power10 processor, and HW 4.0, respectively, which I imagine are relatively low volume chips as far as fabrication is concerned. And I could be wrong, but I imagine yields are not as much of a concern for relatively low volume chips (e.g. IBM Power10 processor, etc.) in comparison to relatively high volume chips (e.g. Exynos 990, etc.).
So I imagine Samsung's 7LPP process node having terrible yields is too much of a huge deal breaker to be a serious consideration, especially when considering that Dane probably has a high performant mobile GPU.

And although the yields for Samsung's 5LPE process node or Samsung's 5LPP process node is probably not as good as the yields for TSMC's N5 process node, I imagine the yields for Samsung's 5LPE process node or Samsung's 5LPP process node are better than the yields for Samsung's 7LPP process node, considering there are many products equipped with the Snapdragon 888 released.

If we look at comparable devices, TSMC's 6nm would seem the most obvious choice at the moment. It's an evolution of their 7nm EUV process (which has been in commercial products since 2019, and is itself based on the non-EUV 7nm which was seen in products back in 2018), so it's hardly bleeding edge, and it seems TSMC is pushing new 7nm clients over to the 6nm variant. It's also a significant improvement over Samsung 8nm in both transistor density and power efficiency, to the point that I'd probably expect 50% more performance if you were to tell me Dane were build on TSMC 6nm vs Samsung 8nm.

I believe the Kirin 990 from Huawei is the only chip publicly revealed to be fabricated using TSMC's N7+ process node, which I don't think is surprising, considering TSMC's N7+ process node isn't IP compatible with TSMC's N7 process node. (And AMD mentioned Zen 3 isn't fabricated using TSMC's N7+ process node.)

The Steam Deck is the obvious example, and although it doesn't come with a dock, it does use a huge 16GB of LPDDR5, and is made in much smaller volumes than Switch.

Gabe Newell mentioned that selling the Steam Deck at the MSRP of $399.99 is painful. So Valve's definitely selling the Steam Deck at a loss. And nobody knows if Nintendo's willing to sell at a loss or not.

 

[Dekuman]

So if Dane is on Samsung 8nm and we feel it's dated by 2023, either the source got it wrong or Nintendo is falling back into their bad habits again

 

[ShadowFox08]

Transistor chip shortage? There’s no shortage of transistors lol

I mean semiconductors.

So if Dane is on Samsung 8nm and we feel it's dated by 2023, either the source got it wrong or Nintendo is falling back into their bad habits again

I don't necessarily think the source got it wrong, but a change of plans could be/is happening. I do think a 2023 switch model will be in the 5-7nm node range, but we'll see.

 

[Mr Swine]

I'm kind of expecting a GameBoy Color situation, where Nintendo don't actually explicitly call it a successor (and count it alongside Switch sales), but for all intents and purposes it's a successor.

Call it Nintendo Switch not 2.

Bam, where is my check Nintendo?

Joking aside, I wonder what they can call it if it’s going to be a successor (why not Switch 2?)

 

[lexony]

Best names would be NES 7 or Gameboy 5.

 

[Hermii]

I mean semiconductors.


I don't necessarily think the source got it wrong, but a change of plans could be/is happening. I do think a 2023 switch model will be in the 5-7nm node range, but we'll see.

Except it’s extremely expensive and time consuming to change process node.

I’m in the 8nm camp, even with a 2023 release date. Yeah its dated compared to other chips coming out, but Nintendo doesn’t compare itself to those. And with Orins feature set, it will provide a nice generational leap.

 

[Kenka]

Gabe Newell mentioned that selling the Steam Deck at the MSRP of $399.99 is painful. So Valve's definitely selling the Steam Deck at a loss. And nobody knows if Nintendo's willing to sell at a loss or not.

I might be beating a dead horse but given the prevalence of hardware sales in Nintendo's earnings (close to 30% of their revenue), I'd say there is a 90% chance that they will go for a model they will sell at breakeven point or for a slight profit (if not for a comfortable margin).

The good thing is, as you mentioned it, that affordable hardware with more elements in their BOM bills than the succ are sold at a profit. Thus, Nintendo has already the possibility to launch a successor with a better SoC at a profit and might just be waiting for the moment their projected cumulated margin on software and on hardware matches their current one. Since their digital sales only go up, so does their margin on software. At some point, the decrease in hardware margin (if any) after the launch of the succ might be compensated so.

I think we might even be able to project when that happens based on former earning releases. Does anyone know how much profit Nintendo makes off hardware?

 

[Jafin]

Gabe Newell mentioned that selling the Steam Deck at the MSRP of $399.99 is painful. So Valve's definitely selling the Steam Deck at a loss. And nobody knows if Nintendo's willing to sell at a loss or not.

I think the Wii U and 3DS were sold at a loss for a while. 3DS was most likely after the price cut when the initial sales weren't what they were hoping for. Wii U was probably because of the extra cost of the gamepad instead of a standard controller. Selling at a loss isn't out of the question, but I don't think they'd do it unless they felt they absolutely had to.

 

[ILikeFeet]

So if Dane is on Samsung 8nm and we feel it's dated by 2023, either the source got it wrong or Nintendo is falling back into their bad habits again

In comparison to the switch, it's a great jump. In comparison to everything else, it's old. Pricing is also a major consideration

 

[Dakhil]

So if Dane is on Samsung 8nm and we feel it's dated by 2023, either the source got it wrong or Nintendo is falling back into their bad habits again

Or to play devil's advocate, outside of price, Samsung's 8N process node is the best choice when yields and securing enough capacity are concerned when compared to Samsung's 7LPP process node, Samsung's 5LPE process node, and Samsung's 5LPP process node.
Unlike Samsung's 7LPP process node, Samsung's 5LPE process node, and Samsung's 5LPP process node, I haven't heard reports and rumours about Samsung's 8N process node having serious yield issues when mobile SoCs are concerned. And considering that Dane's probably a relatively high volume chip, having good yields is extremely important.

Of course, TSMC's advanced process nodes are much better than Samsung's advanced process nodes across the board in terms of performance and power efficiency. However, securing enough capacity for TSMC's advanced process node is extremely difficult and probably extremely expensive since demand for TSMC's advanced process node is absurdly high, coupled by the fact that Nvidia probably doesn't have as much preferential treatment from TSMC as Apple, Mediatek, AMD, and etc., considering Nvidia's only #6 in the list of TSMC's largest customers. In fact, unlike TSMC's top 3 largest customers, Nvidia has to spend significant amounts of money in prepayments to TSMC to secure enough capacity for TSMC's N5 process node. Hopper and Lovelace (specifically high-end Lovelace GPUs for now) are currently rumoured to be fabricated using TSMC's N5 process node. And I don't know if Nvidia considers the margins for Dane high enough to warrant using TSMC's N5 process node for fabricating Dane, at least without making Dane at least considerably more expensive.

 

[Simba1]

At this point, IMO most likely we will have full next gen Switch launch in any way instead of upgrade, do Switch Pro but Switch 2,
there is chip shortage so I dont see new hardware from Nintendo this year, also Switch is still selling great in any case, so I dont see point of relesing Switch upgrade in 2023. for instance instead of full next gen Switch (Switch 2), offcourse current Switch models would sale and recive games at least 2-3 years after Switch 2 launch.

 

[Skittzo]

At this point, IMO most likely we will have full next gen Switch launch in any way instead of upgrade, do Switch Pro but Switch 2,
there is chip shortage so I dont see new hardware from Nintendo this year, also Switch is still selling great in any case, so I dont see point of relesing Switch upgrade in 2023. for instance instead of full next gen Switch (Switch 2), offcourse current Switch models would sale and recive games at least 2-3 years after Switch 2 launch.

The current chip shortage will be lasting until 2027 at least, so it means nothing in regards to launch timing.

Also Dane releasing in 2022 is a lot more desirable/impressive than Dane releasing in 2024+. The hardware we're hearing about is not likely to change regardless of when it's launched, so waiting an extra few years does nothing but make it less and less cutting edge.

 

[Simba1]

The current chip shortage will be lasting until 2027 at least, so it means nothing in regards to launch timing.

Also Dane releasing in 2022 is a lot more desirable/impressive than Dane releasing in 2024+. The hardware we're hearing about is not likely to change regardless of when it's launched, so waiting an extra few years does nothing but make it less and less cutting edge.

It will not be same intensity of chip shortage until 2027.

Its not point about whats more impressive but whats best for Nintendo, and chip shortage and Covid effect on Switch "Pro" launch until now in any case,
thing is that in 2023. Switch is entering its 7. year on market so at this point it makes more sense for Nintendo to have full next console launch instead of next gen upgrade.

 

[Skittzo]

It will not be same intensity of chip shortage until 2027.

You don't know this. Also, it's not "intense" when it comes to things needed to make Switches, considering they're still making ~23-24M a year. They just launched a brand new model in October 2021, why would they do that if they thought the chip shortage was a reason to postpone new hardware launches?

Its not point about whats more impressive but whats best for Nintendo, and chip shortage and Covid effect on Switch "Pro" launch until now in any case,
thing is that in 2023. Switch is entering its 7. year on market so at this point it makes more sense for Nintendo to have full next console launch instead of next gen upgrade.

I'm not sure I understand. There's not going to be any meaningful difference in the hardware itself if they call it a pro or a 2, the hardware is a somewhat known quantity. If they release this hardware in 2022 it'll be a lot more capable of receiving games made for other consoles in the recent past and near future than it would be if they release in 2024+.

 

[Simba1]

You don't know this. Also, it's not "intense" when it comes to things needed to make Switches, considering they're still making ~23-24M a year. They just launched a brand new model in October 2021, why would they do that if they thought the chip shortage was a reason to postpone new hardware launches?


I'm not sure I understand. There's not going to be any meaningful difference in the hardware itself if they call it a pro or a 2, the hardware is a somewhat known quantity. If they release this hardware in 2022 it'll be a lot more capable of receiving games made for other consoles in the recent past and near future than it would be if they release in 2024+.

Its logical assumption that there will not be same intensity of shortage, it could be better or worse also, making assumption that next 5 years will be exactly same like now regardless chip shortages is quite a reaching.

Actually it's very important how Nintendo will market it and positioning it, also real next gen launch also could mean bigger changes and some new features compared to current models instead if we talking only about "upgrade" (counting that Nintendo went from "Pro" plans to full next gen launch).
I was talking about 2023. launch not about 2024., I dont see any new hardware this year.

 

[Dekuman]

Or to play devil's advocate, outside of price, Samsung's 8N process node is the best choice when yields and securing enough capacity are concerned when compared to Samsung's 7LPP process node, Samsung's 5LPE process node, and Samsung's 5LPP process node.
Unlike Samsung's 7LPP process node, Samsung's 5LPE process node, and Samsung's 5LPP process node, I haven't heard reports and rumours about Samsung's 8N process node having serious yield issues when mobile SoCs are concerned. And considering that Dane's probably a relatively high volume chip, having good yields is extremely important.

Of course, TSMC's advanced process nodes are much better than Samsung's advanced process nodes across the board in terms of performance and power efficiency. However, securing enough capacity for TSMC's advanced process node is extremely difficult and probably extremely expensive since demand for TSMC's advanced process node is absurdly high, coupled by the fact that Nvidia probably doesn't have as much preferential treatment from TSMC as Apple, Mediatek, AMD, and etc., considering Nvidia's only #6 in the list of TSMC's largest customers. In fact, unlike TSMC's top 3 largest customers, Nvidia has to spend significant amounts of money in prepayments to TSMC to secure enough capacity for TSMC's N5 process node. Hopper and Lovelace (specifically high-end Lovelace GPUs for now) are currently rumoured to be fabricated using TSMC's N5 process node. And I don't know if Nvidia considers the margins for Dane high enough to warrant using TSMC's N5 process node for fabricating Dane, at least without making Dane at least considerably more expensive.

I also wonder if we're looking at this correctly.
Consumer Maxwell was on 28nm consumer Pascal on 16nm Tegra X1 was on 20nm which was an intermediate process, and as I understood it based on previous comments from people who know more, users of 20nm moved off it pretty quickly to 16nm and there was a glut of capacity that allowed nvidia to use it for the X1

Based on Wikipedia Ampere consumer is on the Samsung 8nm process, which aligns with rumors. The next generation will be Lovelace which is still missing information for consumer. 8nm capacity will likely be more readily available for Nintendo or alternatively if it follows X1's pattern it will land on some intermediate process, but I don't know if that;'s what actually the plan is or if nvidia's plan was always to maximum volume on the 8nm process and maintain it once they move on from manufacturing consume GPUs to a newer process by slotting Dane on it.

 

[Hosermess]

For all the talk about chip shortages, is anyone losing money right now on their consoles? Obviously Nintendo/Sony/MS would prefer the ability to produce more (consumers too) and rake in even more revenue (shareholders too), but if there are only shortages because the demand is crazy high, then I think all lot of the talk and concern is overblown (unless you're a shareholder I guess).

Sure, PS5 hunting sucks. And (unlike MS & Sony who had already announced their consoles before the Covid shit hit the fan), I guess Nintendo could always hold back until they can guarantee the ability produce enough Dane units to hit a certain level of revenue, but I think that if Nintendo predicts they can manufacture enough of them to hit the break even point, (and everything else is good to go), they'd pull the trigger. But who knows.

 

[ILikeFeet]

For all the talk about chip shortages, is anyone losing money right now on their consoles? Obviously Nintendo/Sony/MS would prefer the ability to produce more (consumers too) and rake in even more revenue (shareholders too), but if there are only shortages because the demand is crazy high, then I think all lot of the talk and concern is overblown (unless you're a shareholder I guess).

Sure, PS5 hunting sucks. And (unlike MS & Sony who had already announced their consoles before the Covid shit hit the fan), I guess Nintendo could always hold back until they can guarantee the ability produce enough Dane units to hit a certain level of revenue, but I think that if Nintendo predicts they can manufacture enough of them to hit the break even point, (and everything else is good to go), they'd pull the trigger. But who knows.

their margins have shrunk. whether it's to the point of losses, we don't know, but with the sky high demand of games, losses have been reduced to the point of irrelevance, IMO.

shortages aren't just due to demand, it's ever very known fact that components are difficult to acquire. literally everyone is hit by this

 

[ShadowFox08]

Except it’s extremely expensive and time consuming to change process node.

I’m in the 8nm camp, even with a 2023 release date. Yeah its dated compared to other chips coming out, but Nintendo doesn’t compare itself to those. And with Orins feature set, it will provide a nice generational leap.

Doesn't seem like it was ever finalized to begin with. It's one thing if it was taped out,. but...

If Switch 2 does use a version of the Orion NX chip, it's probably gonna be whatever node size that will be..

Or I could be wrong and it uses 8nm as kopite claims... in 2023.

 

[ILikeFeet]

the only way I can see Nvidia making a chip off 8nm is if TSMC suddenly has spare capacity that Nvidia buys up or Samsung gives out steep discounts for 4nm. the latter sounds more likely to me if this chip gets pushed out further to late 2023

 

[Dakhil]

8nm capacity will likely be more readily available for Nintendo or alternatively if it follows X1's pattern it will land on some intermediate process, but I don't know if that's what actually the plan is or if nvidia's plan was always to maximum volume on the 8nm process and maintain it once they move on from manufacturing consume GPUs to a newer process by slotting Dane on it.

I think that's definitely a possibility, especially since there's also a possibility Nvidia has mid-range and/or entry-level Lovelace GPUs be fabricated using Samsung's 5LPP process node.

Doesn't seem like it was ever finalized to begin with. It's one thing if it was taped out,. but...

If Switch 2 does use a version of the Orion NX chip, it's probably gonna be whatever node size that will be..

Or I could be wrong and it uses 8nm as kopite claims... in 2023.

Jetson AGX Orin is probably already taped out since Jetson AGX Orin devkits are scheduled to become available in Q1 2022, which should be relatively soon. And Jetson Orin NX is probably a binned version of the Jetson AGX Orin going by the specs, so Jetson Orin NX's probably already taped out as well. Considering that Nvidia indirectly mentioned that Orin was fabricated using Samsung's 8N process node, and that max CPU and GPU frequencies for Jetson AGX Orin and Jetson Orin NX are practically the same as the max CPU and GPU frequencies of the Tegra X1, I'm inclined to believe Samsung's 8N process node is being used for the Jetson AGX Orin and the Jetson Orin NX.

the only way I can see Nvidia making a chip off 8nm is if TSMC suddenly has spare capacity that Nvidia buys up or Samsung gives out steep discounts for 4nm. the latter sounds more likely to me if this chip gets pushed out further to late 2023

And I imagine around right now is probably the absolute latest for Nintendo and Nvidia to make a decision to continue using Samsung's 8N process node or migrate to Samsung's 4LPE process node or Samsung's 4LPP process node for fabrication, and ensure Nintendo and Nvidia has enough capacity secured, in the possibility of a holiday 2023 launch.

 

[Klax0nKlax0ff]

What are the potential benefits of Samsung’s LPE or LPP process nodes vs. sticking with 8N? I’m assuming it’s more power efficient but is it realistically enough that Nintendo would be swayed to stray from the currently speculated Orin-offshoot?

 

[Dakhil]

What are the potential benefits of Samsung’s LPE or LPP process nodes vs. sticking with 8N? I’m assuming it’s more power efficient but is it realistically enough that Nintendo would be swayed to stray from the currently speculated Orin-offshoot?

That's definitely one benefit. Another benefit is that chips can theoretically be considerably smaller due to the transistors being smaller as the process nodes become more advanced. (I say theoretically since there are technical complications associated with process nodes becoming more advanced that results in making chips being smaller much more difficult. One example is that the Apple A14 Bionic failed to achieve a real world transistor density very close to the max theoretical transistor density of TSMC's N5 process node, due to the technical complications with scaling down SRAM.)

 

[ILikeFeet]

What are the potential benefits of Samsung’s LPE or LPP process nodes vs. sticking with 8N? I’m assuming it’s more power efficient but is it realistically enough that Nintendo would be swayed to stray from the currently speculated Orin-offshoot?

if they keep the performance the same as an 8nm chip, then they would get a smaller chip that consumes less power. if they keep the power consumption and area similar, then they get better performance

 

[Hermii]

Doesn't seem like it was ever finalized to begin with. It's one thing if it was taped out,. but...

If Switch 2 does use a version of the Orion NX chip, it's probably gonna be whatever node size that will be..

Or I could be wrong and it uses 8nm as kopite claims... in 2023.

It was very close to finalized, and maybe it is now. There’s no way to know.

If they’re changing process nodes this late in the process, they have to scrap/ redo a lot of work.

 

[NineTailSage]

That's definitely one benefit. Another benefit is that chips can theoretically be considerably smaller due to the transistors being smaller as the process nodes become more advanced. (I say theoretically since there are technical complications associated with process nodes becoming more advanced that results in making chips being smaller much more difficult. One example is that the Apple A14 Bionic failed to achieve a real world transistor density very close to the max theoretical transistor density of TSMC's N5 process node, due to the technical complications with scaling down SRAM.)

Here's another good comparison between TSMC's advanced process nodes compared to Samsung's and how they compare with the latest SoC's from Mediatek and Qualcomm.

Whatever happened to Samsung's 8LPA process update they added just last year?

"In the previous process, in the evolution of the 10LPE process, Samsung added an 8LPA. Samsung said that compared to 8LPU, 8LPA achieves a 10% performance improvement and 15% power consumption reduction."

 

[ILikeFeet]

what if Nvidia took the time to design Ampere for Samsung 7nm long ago

 

[Dakhil]

Whatever happened to Samsung's 8LPA process update they added just last year?

"In the previous process, in the evolution of the 10LPE process, Samsung added an 8LPA. Samsung said that compared to 8LPU, 8LPA achieves a 10% performance improvement and 15% power consumption reduction."

My guess is that Samsung's 8LPA process node's being used to fabricate Samsung's 5G RF chips.

what if Nvidia took the time to design Ampere for Samsung 7nm long ago

My guess is that the inventory shortages for the consumer Ampere GPUs would be at the very least considerably worse, considering that Andrei Frumusanu mentioned that the binning of the Exynos 990 units Anandtech received are apparently bad, with the binning of most Exynos 990 units out there being as bad, or worse. And I imagine that yield problems for Samsung's 7LPP process node would increase exponentially as the die size grows.

 

[NineTailSage]

My guess is that Samsung's 8LPA process node's being used to fabricate Samsung's 5G RF chips.


My guess is that the inventory shortages for the consumer Ampere GPUs would be at the very least considerably worse, considering that Andrei Frumusanu mentioned that the binning of the Exynos 990 units Anandtech received are apparently, with the binning of most Exynos 990 units out there being as bad, or worse. And I imagine that yield problems for Samsung's 7LPP process node would increase exponentially as the die size grows.

Do you think they rebranded it then because it kind of reads like a process specialized for RF based technology...
Part of me feels like the 8N process was developed for Nvidia designed around higher power needs in desktop GPU's and all of Samsung's other 8LPP/8LPU/8LPA were catering more to mobile like devices.

 

[ILikeFeet]

My guess is that the inventory shortages for the consumer Ampere GPUs would be at the very least considerably worse, considering that Andrei Frumusanu mentioned that the binning of the Exynos 990 units Anandtech received are apparently, with the binning of most Exynos 990 units out there being as bad, or worse. And I imagine that yield problems for Samsung's 7LPP process node would increase exponentially as the die size grows.

what I'm saying is, since for the longest time, it was thought that Ampere was to be on 7nm, it could be possible that Ampere is designed for both Samsung 8nm and 7nm. if so, maybe it's possible that Nvidia can bring forward their 7nm design since they don't have to fear large die size and high power consumption. this would allow Dane to be more flexible and utilize Samsung's advanced 7nm nodes like 5nm or 4nm

 

[Dakhil]

Do you think they rebranded it then because it kind of reads like a process specialized for RF based technology...

Absolutely possible. This is certainly not the first time Samsung has rebranded a process node, as shown below.

what I'm saying is, since for the longest time, it was thought that Ampere was to be on 7nm, it could be possible that Ampere is designed for both Samsung 8nm and 7nm. if so, maybe it's possible that Nvidia can bring forward their 7nm design since they don't have to fear large die size and high power consumption. this would allow Dane to be more flexible and utilize Samsung's advanced 7nm nodes like 5nm or 4nm

Actually, Samsung's 4LPE process node is now considered a separate process node branch from Samsung's 7LPP process node branch instead of Samsung's 4 nm** process nodes being an evolution of Samsung's 5 nm** process nodes and a further evolution of Samsung's 7LPP process node previously. (There was also mention of Samsung's 6LPP process node, but given that there hasn't been any chips from Samsung or any other company announced to be fabricated using Samsung's 6LPP process node, and Samsung's 6LPP process node's effectively only available to Samsung and perhaps some of Samsung's largest customers, Samsung's 6LPP process node practically doesn't exist.)

So the only realistic option for Nintendo and Nvidia is Samsung's 5 nm** process nodes if Nintendo and Nvidia want acceptable yields. (I suppose TSMC's N6 process node could also be a possibility if Nintendo and Nvidia have designed Dane to be foundry IP agnostic from the very beginning, which I have my doubts about, especially if Dane's based on Samsung's 8N process node, which kopite7kimi has said more than once (here and here).)

 

[Deleted member 645]

Anything new in here?

 

[Dakhil]

Anything new in here?

Outside of some GTC 2022 (Spring 2022) sessions potentially being of some interest, not really.

 

[Dekuman]

@Tron1
Nothing new it seems, We're back to pessimism talk.

 

[Terrell]

I’m out of step here… why are we considering that the Samsung 8nm process node is out?

 

[Skittzo]

I’m out of step here… why are we considering that the Samsung 8nm process node is out?

Wishful thinking IMO.

 

[ILikeFeet]

I’m out of step here… why are we considering that the Samsung 8nm process node is out?

"It's old"