So as time as passed I've fallen out of love with the idea of a "Pro" revision. There was a window where I think that had appeal due to worries about the Switch's power gap leading to consumer interest dropoff, and due to a belief that technology for a true generational jump in the technical form factor of the portable Switch wasn't ready at Nintendo price points yet.
Obviously, the Switch is still selling gangbusters, and I guess we could probably say that it's actually currently supply-constrained! It still has a power gap, but its unique portable proposition coupled with the power of Nintendo first party software make it look like it can have a full generational cycle. It's hard to imagine a hardware sales dropoff so extreme that Nintendo couldn't ride a healthy 2020 and competent 2021 for the Switch, taking the console into year 5.
Additionally, I think we're finally seeing the hardware appear that suggests a full Switch 2 successor could materialize at desired performance levels, which is what prompted me to make this post.
To begin with, I think the idea of a Switch 2 targeting XB1 or PS4 performance in a handheld is a good shorthand. I'm not an electrical engineer so this is all internet fanboy level speculation, but I'd generalize that target even further to hitting the 1 TFLOP target for the GPU, and of course increasing the CPU power appropriately.
Currently the Switch can be said to almost reach 400 GFLOPS docked, and almost 200 GFLOPS when portable. So as another rule of thumb, I would generalize a Switch 2 target EVEN FURTHER to aiming to double the Launch Switch's performance.
Actually, this would ALSO jive with Nintendo's current pattern of simply doubling performances every generation: The Wii was two GC's duct-taped together, the Wii U was double the Wii's performance, the Switch is even sort of twice the Wii U's performance.
We've already seen the Switch handle previously unimaginable current-gen ports with aplomb, so in conclusion I think a rough doubling of the Switch's performance, (which almost hits a supposed 1 TFLOP GPU performance target) is a reasonable yardstick to go by.
And by that yardstick, I think we're finally seeing tech in the consumer market that achieve this.
For example, the Surface Pro X launched in Fall 2019 with the claim that it had 2TFLOP GPU. The Tegra X1 ws claimed as the first mobile SoC that hit 1 TFLOP, so this is the sort of "doubling" claim that I've been looking out for. In this case however, MS is using a custom Snapdragon 8cx platform, which is Qualcomm instead of Nvidia. They're using A76 CPU processors and a mobile Adreno 685 GPU as compared to the Switch's A57 CPUs and Nvidia Maxwell architecture.
Again, I'm just an armchair crank with the internet, but on the CPU side I think the Arm Cortex A73 is already supposed to have twice the sustained performance of the A57's used in the Switch. The A76 is another generational improvement on top of that, but this time I think it's more like a 40% improvement vs. the A73/A75. (A LOT of marketing is centered around peak performance unrestricted by power or heat, which is OBVIOUSLY not realistic for the Switch.) So just on the CPU side I think we're seeing consumer-level offerings that are around 2.8X the Switch CPU.
It's easy enough to think about the CPUs, but the GPU is tougher. I have no idea how to compare the Adreno performance claims with the Tegra X1 inside the Switch since they're completely different lines. It SOUNDS like a mobile chipset hitting 2 TFLOPS bodes well for the technology being ready, but it's way guesswork than even I'm comfortable with, and obviously I'm indulging in a metric-ton of what-ifs here.
Ideally, I'd look at Nvidia successor Tegra chips, but these are uber weird things none of which actually are suitable for the Switch. So I'll just look at the GPU architectures they've been putting out. The Launch Switch used Maxwell cores at 20nm. The Maxwell architecture was succeeded by Pascal, but that sounds like most of the benefits of Pascal was jumping down to 16nm with very little innate improvements. The next meant-for-consumers architecture was Turing, which DOES exist in the market: Nvidia GeForce 16 GPUs hit the market in early 2019 as a budget line targeting $150 GPU SKUs, lacking exotic AI and raytracing. These are desktop GPUs, so obviously they're super-powered compared to what will the Switch will use. According to Wikipedia the GeForce GTX 1650 has 3.5X the number of shader processors the Switch has. But naively, imagine a Switch 2 with 384 shader units (compared to the Switch's 256) that are maintain the performance improvement ratio, and that means it should hit 1.083 TFLOPs! And that's considering the GeForce GTX 1650 has a die size of 200 mm^2, so an equivalently shrunk GPU would be 85mm^2 (Tegra X1's GPU size is 118 mm^2).
.... I think my justifications on GPU power doubling is a little thin, but since an Adreno GPU is able to claim hitting 2 TFLOPS, then I would think it's possible in the mobile space for Nvidia to do the equivalent.
Finally, I wanted to find current available tech a these levels because when the Switch came out in 2017, it was using cutting edge 2015 consumer tech. Likewise, if this tech is available in 2019/early 2020, then I could envision a Switch 2 landing in 2021 or 2022. Obviously prices on these components hopefully come down in time, and there are other technical components that need to be considered, like RAM pricing and performance, as well as the question of internal storage.
In conclusion, I've basically completely given up on the idea of a Switch "Pro", and now am focusing on imagining a Switch 2 that is on paper not quite an XB1 or PS4, but in reality is able to perform very much like one with a 3 hr battery life. I guess it could come in 2021, but with the Switch's current momentum, I'm more imagining it coming to market in early 2022 with a little bit more breathing room against the the XBSX and PS5 furor.
And of course, I'd envision it being 99% backwards compatible with Switch so that the generation can transition smoothly, the Switch "1" can move to lower price points naturally and the Switch 2 can have a little more price flexibility in case Nintendo needs to target a higher then $299 price point. If a Switch 2 launches in 2022, the Switch 1 would've had a 5 year lifecycle but still probably be doing really strongly. Perhaps there'd be a risk of cutting out the Switch's sales by releasing its successor, but I'd like to think that a large number of cross-gen games, as well as ever-more-palatable price points, would allow the Switch "1" to trail alongside the Switch 2 for a very long time, somewhat akin to the longer tails that the PS1, PS2, and GBA enjoyed even while their successors were on the market.