I think this alternate history ends up worse for Nintendo in the long run since they wouldn't have made the Switch.

Wii U was a blunder, but I think Nintendo needed that to see where they truly needed to improve. Given Switch's success, it's quite clear the power of the Wii U was not the primary issue. So, unless third parties flocked to Wii U because it was simply easy to port the games, it still wouldn't have been that great

The CPU and GPU in the Wii were pretty weak, so it's possible that Nintendo would have been able to emulate the Wii.

But let's spin the idea of this thread a bit further and have a look what hardware such a hypothetical higher-powered Wii U could have.

There were sadly not many  options for a console CPU: Intel was not really an option due to their prices and unwillingness to do semi-custom chips at the time, IBM was laser-focused on HPC, so their CPUs were a bad fit, and AMD FX had low performance and high power consumption, so that one also didn't fit well, and the Jaguar, which was used in PS4 and XBO, wasn't out yet.

The only real viable option would have been a mobile derivative of the Phenom II/Turion II CPUs, or possibly from the Jaguar predecessor Bobcat. Considering that the Wii U uses a triple-core configuration, I'll do the same in my hypothetical build and choose the Phenom II P820 as a base; it's a triple-core mobile Phenom II with a clock speed of 1.8 Ghz and a TDP of 25W, the latter of which is similar to the one of the Jaguars on PS4 and XBO.

As for the GPU, I will also go AMD, since Nintendo at the time always went with them. Since Nintendo doesn't tend to use the newest chips, GCN, which released a couple months prior, is off the table. This gives me the option between Terascale 2 and Terascale 3, so VLIW 5 or VLIW 4. AMD didn't bring any mobile Terascale 3 chips and only one high-end GPU desktop chip, so I'll use the last mobile Terascale 2 as base for this Wii U version, the Radeon HD 6850M to be precise. This chip should have a performance somewhat similar to the one in the XBO, with higher compute performance per CU but worse clock speeds cancels this advantage out.

To round this out, I'll give it a memory setup similar to the XBO, as the Wii U already had a similar setup anyway. I give it also 8GB of GDDR5 memory, but at a lower clock speed of 933Mhz instead of the 1066Mhz of the XBO, as this is more similar to the one on the GPU, plus 64MB of eDRAM instead of SRAM as the Wii and Wii U used DRAM caches instead - I just doubled the amount compared to the real Wii U.

This brings us to the price, and here I fear the Wii U would definitely go under. Nintendo can't subsidize their hardware like their competitors do, and with the tablet controller I'd give it a price of at least $499, putting the Wii U in a similar position as the XBO with Kinect early on. even without the tablet controller, I still think it would cost $449, making it more expensive than it's competitors while being overall weaker.

Is this assuming a 2012 release as historically? 

In the scenario I posited, this hypothetical alternative would release in 2013 in order to attain parity.

This sounds similarily bad to the actual Wii U. Hardware sold at a loss all the same, same (actually worse) first party droughts combined with being an immediate no-go in comparison to the PS4, because this hypothetical console would be missing out on most multiplatform games from the get-go.

$350 for the actual Wii U, including a game, was a tough sell already. But $400 for a console without a game included, a console that was going to miss out on most multiplats is worse. This would have likely sold less than the Wii U, because an early (and necessary) price cut would be even harder to stomach for Nintendo than it was for the real Wii U. Not to mention that PS4-level of graphics would have led to even more delays for Nintendo's first party games.

Yes, this is assuming a 2012 launch as the trajectory of the Wii sales is unaffected and Nintendo needed a replacement earlier then 2013.

In a 2013 release, the only changes would be the GPU (which then goes from 10 Terascale 2 CU to 12 GCN CU at a slightly higher clock speed), and potentially a slightly higher clock speed or 4th core on the CPU (so based on the Turion II P860 or P920). The performance increase would be rather negligible as I'm sure Nintendo would be able to make good use on the VLIW 4's higher amount of shaders per CU, thus the only real upgrade would be on the CPU.

Also, I made a typo in the previous scenario with the GPU, I meant the Radeon HD 6850M, not 6750M. That GPU is almost twice as powerful, so that typo was pretty significant.

For reference, the real Wii U had basically a slightly underclocked Radeon HD 6450M inside, the 6750M as base would already have been 3 times as powerful while the 6850M as base would have been about 5 times as performant

If Nin isn't more open and friendly with 3rd parties:
Wii 2U - 22M
XB1 - 50M
PS4 - 117M

Assuming Nin also ends up more open and friendly with 3rd parties with this more old school hardware approach:
XB1 - 32M
Wii 2U - 40M
PS4 - 117M

MS and XB screwed up really bad. Really really bad. Nin could have taken huge advantage of this just like PS4 did, but it would require far better 3rd party support.

I think we still end up getting Switch after this regardless, because I think Nin would be wise enough to realize this only worked decently because of how badly MS screwed up with XB1, and that when they right their wrongs like SNY did from PS3 to PS4, XB will likely come back stronger again, so can't rely on another outcome like this.

The question would be does Switch end up similar and as good as it was without what Nin learned from the Wii U failure in our timeline? It very well might have not, in which case Switch launches around 2020, but isn't quite the hit it ended up becoming. Maybe Switch only has 50-75M unit sales at this point in time because of it, which wouldn't be that bad, but certainly wouldn't be the massive success that Switch has become.

If Nin didn't realize why Wii 2U sold decently, and then tries something more like the Wii U as we knew it, Nin bombs gen 9, only to launch Switch early, prior to 10th gen, using outdated mobile hardware just like Switch did, just not as weak overall since it's basically 10th gen at this point.

I suppose the only way not to sell Wii U at a loss while having better specs than it ended up having would be to ditch the tablet, but still make it a lot weaker than PS4 and Xbox One.

Wii U could've had 4 GB RAM like Switch and a more up-to-date RAM at the time. The CPU and GPU could also be something below the other 8th generation processors but better than the actual Wii U and easier for devs to work with.