I think the biggest counterpoint here is they have said they expect Switch to last 7 years and we are indeed getting close to that end date, they now only have a single platform to support (so no more faffing around with quick and easy updates on a side platform) and Dane is architecturally different and at the low end will likely be 3X more powerful. That is usually the definition of a successor.
Well that's kind of the thing. The architectural improvements from A78 and Ampere are definitely in line with what I'd expect from a successor, and as we creep past the five year point we're also at a stage where we'd expect a successor, not a "pro" model, to be launched. The rumours we've got on the first page of this thread, though, all seem to suggest a Switch 4K as a "pro" style model, and my argument is that the Samsung 8nm process is also indicative of a "pro" style model (ie designed for healthy profitability at launch, rather than low margins to accelerate adoption). Honestly I don't know what it is at this stage, and although I'm concerned that Nintendo may try an unorthodox approach of a quasi-"pro", quasi-successor that may not be successful, you can rest assured that I have a close to 0% success rate in predicting new Nintendo hardware, so if I'm saying it, it almost certainly isn't going to happen!
For what it's worth, I'm perfectly happy with an 8nm A78+Ampere SoC, and I'll be buying it day one regardless of whether they position it as a Switch 2, a Switch Pro, or something in between. I'm just trying to read the tea leaves, as it were, on their business strategy from the scraps of information we have.
But at the same time, chips redesigns to use newer nodes are not rare for the industry and even Nintendo did one for the current switch. Is true that a last time chip redesign would mean a significant delay, but the theory would be that the product is already delayed.
Redesigning to use newer nodes is common after a chip launches, but not during development. Or at least I'm not aware of any confirmed recent cases where a chip had gone through a large part of the design process on one node, only for a change of plan to redesign it for a different one. In most cases, for CPUs, GPUs, mobile SoCs, etc, I imagine they would probably just cancel the chip altogether, as they would already have a following generation lined up on the next node anyway. It's a bit different for console chips, which are obviously one-off designs, but it would still mean a delay, I assume of at least a year.
If we're to assume that they were developing on 8nm, but since changed nodes, then they would have had to make the change at the earliest mid-2021, as
kopite7kimi was saying in July that it was being made on Samsung 8N. With probably a year to migrate to the new node, then tape-out and manufacturing, that would put early 2023 as the earliest date it could launch, possibly later.
Now that's not impossible, but it would definitely amount to a drastic change of plans. Unless there's any clear indication this has happened, I'm definitely still expecting an 8nm chip.
That's surprising to hear since they definitely were selling it at a loss after the $75 price cut. I guess shipping and packaging costs made up more of the price per unit than I thought.
Well yeah I don't think they designed it with a $500 launch price in mind, but I do think they're likely having discussions internally about driving up the price just based on current demand for any and all consumer electronics.
Switch isn't really all that available in most major markets, even now. In the US Amazon is backordered for every model but like a handful of the regular red/blue one. In Japan it's getting harder to find again, especially the OLED model.
Considering the OLED model has a price that serves as a premium entry point at this stage, the fact that it specifically is essentially sold out globally should indicate that a higher end 4k model that actually has more value in its upgrades would sell similarly well. PS4 Pro and XSX aren't exactly comparable specifically because we're not talking about a strictly "pro" style model here.
And I honestly disagree, I do think if the current climate existed during the launch of the PS4 Pro and X1X they would absolutely have sold out instantly just like the base models would have been doing.
There are sporadic, regional shortages of Switch, but (and perhaps I'm extrapolating too far beyond my local experience), it doesn't seem all that difficult to get one if you want one, and it points to supply being just a bit lower than demand, rather than demand far outstripping supply as it the case for PS5. SwOLED demand has been high, but again still seems to be available if you look. It's in a bit of a weird place as it's priced so closely to the base model, and it's still in the early adopter phase, so I'm not sure what demand will look like in the future. Of course, I suspect they may simply replace the base model with the SwOLED within the next year anyway, which would make the question moot.
I think it’s possible that, with the demand of the switch even before say, late 2019, which was pretty high, they could have designed the product as a 400 dollar product. The product that eventually came out, Aula, wasn’t designed it seems to be a 300 dollar product but a 350 despite not changing anything internally (drastically). Having the better screen being one of the features for it among the others. The dock was probably a last minute thing though, but the switch itself probably wasn’t.
And if course the packaging, the software, the retail, etc., are included in this.
Yeah, I suspect in late 2019/early 2020, if planning a successor they probably were looking at a price higher than $300, between the clear success of the Switch at that point, and just general inflation, I'd say a planned $350 sales price, or maybe even $400, would have been possible.
You might be reading too much into the (rumored) decision of fabricating Dane on the Samsung 8nm. For better or worse, the Switch SOCs are tethered to Nvidia's automotive products*. It started with Drive PX (TX1). When Drive PX 2 (TX2 on 16nm) came along, the Switch SOC followed it to 16nm too. So if Dane indeed will be using the 8nm process, it may not be indicative of Nintendo's succession strategy but simply what Nvidia makes available to them.
* This could change if Nvidia decides to re-enter the mobile/Chromebook SOC market by forging a new partnership with, say, MediaTek.
There's certainly a lot of technology shared between Nvidia's automotive SoCs and their SoCs for Nintendo, but I don't think it makes sense to assume that Nvidia will simply treat Nintendo as some kind of small second-fiddle to their automotive designs. For one, it seems very likely that Nintendo currently accounts for more revenue for Nvidia than the entire automotive industry, and even if that changes in the future (and the automotive side becomes more profitable), the number of SoCs produced for Nintendo will still likely remain an order of magnitude higher than the number produced for the auto industry.
In fact, if we're to assume that Dane is intended for a successor which will fully replace the current Switch, then you'd be looking at possibly 100+ million chips over its lifespan, which would very likely make it the highest-volume production of any individual Nvidia chip for the coming years, perhaps by a large margin. In that case it really doesn't make sense for either Nvidia nor Nintendo to tightly restrict the design around other, much lower-volume chips. Certainly there's no reason to re-invent the wheel, and they wouldn't want to spend huge R&D budgets on redesigning minute aspects of the Ampere architecture, but I can't imagine the manufacturing process would have been off the table, if Nintendo had been willing to pay for the R&D required. And I don't see the cost of migrating to a different node being excessive in the context of 100m chips, particularly when Nvidia has plenty of experience migrating architectures between nodes, and even across different foundries.
Put it this way; when Nvidia started design on the Tegra X1, the idea that it would be used in a Nintendo console selling over 100 million units would have been the absolute best case scenario. But it was still worth migrating from the 28nm process used on their GPUs to the (then) cutting-edge 20nm process to squeeze out whatever performance and power efficiency that they could get. If they're now designing a chip which they know will be used in the successor to a 100m+ selling device, it would seem strange to me that they'd be much more conservative about the manufacturing process.
Of course that's in the context of a full successor with an estimated production approaching 9 figures. If Dane is for a Switch "Pro", and the expected run is closer to 10 or 20 million chips at most, then the R&D spend for a newer node wouldn't make sense, and going for an 8nm chip and keeping things as close as possible to Orin would be the way to go.
Is shrinking Dane down the road a given though? For what—Dane Lite? There's no guarantee such a product would ever be released. It's a budget model mainly for kids and casuals; might as well keep Mariko and lower the price instead. And if Nintendo (partnering with Nvidia?) is to wade further into cloud gaming, there'd be no incentive to replace Lite's SOC.
Even if a Dane Lite is in the cards, Nintendo could use the same 8nm Dane and accept a subpar battery life. Reasonable consumers would understand the trade-off for a budget model.
Well this is it. If Dane is being used for a successor, then yes, it should fully replace the existing Switch over time, and a Dane Lite should absolutely be expected to follow on a couple of years later to replace the OG Switch Lite as the original Switch is phased out completely. If it's a "pro" model to be sold alongside the standard Switch until a full Switch 2 (or something else) appears, then there's no need for a Dane Lite or a die shrink at all.
I think it's probably too late for a pro model, but maybe Nintendo are a lot more literal about being in the "middle phase" of Switch's life than I've thought, and they actually do expect the Switch to last 10 years. Maybe we don't see a successor until 2025/2026, and they're just going to ride this out as long as they can.