It's been less than 2 months, aren't all console launches like this?
To some degree but you have nonstandard variables here.
First: it isn't in a holiday season specifically, though Christmas isn't a global large gift buying period just a western world one.
Second: it launched with Breath of the Wild, arguably the biggest non-Mario IP to ever launch a system, a much anticipated release since the last one was at the tail end of the Wii's life and had motion based controls, as such it failed to pull in the core Zelda audience, proven by it's ~3.5M sales. Good for most games but a poor finish by Zelda standards, it's basically their equivalent of Gran Turismo 6, except GT6 sold over 5M units.
Third: the Wii U was largely skipped by even Nintendo's core audience, likely because by the time they actually started getting compelling software on it the writing was pretty clearly on the wall for all to see. As such the fact that BotW, MK8, and Splatoon were announced pre-release to hit the Switch makes a very strong argument to the core Nintendo fan that it's worth coming back.
Fourth: per Nintendo this is both the new handheld and the new console. Sure, they say it's replacing Wii U and not ending the 3DS' lifespan but they've also repeatedly said that all future development efforts would be consolidated on the Switch. So a pretty damn strong signal to their rather large handheld audience to get on board too, a handheld audience that move 70M of an only slightly cheaper (for most of it's life) 3DS for vastly inferior hardware.
I could probably go on, but suffice to say that the Switch situation is entirely unique to the Switch. I've personally never thought there was much capacity for true failure. We're talking about a floor in the ~30M unit range, a most likely mid-point in the 50-70M range, and a more likely than floor upper bound of 100M+ units. 2018 and beyond software, pricing, marketing, etc. will determine where it lands but this was never going to be another Wii U short of Nintendo laying an already rotten egg on year one software (i.e. BotW being a flop instead of an all-time classic, Odyssey being at least as bad as Sunshine if not worse, MK8D being a shoddy port, and Splatoon 2 having busted net code).
Are we acting like manufacturing and holding costs don't exist? Because the latter option is a ton of waste.
lets make a small calculation:
Producing 20 when you can only sell 15 in a giber period means that you have 5 million consoles not paid for. Lets say Switch costs 200 to make. That means tou pay 1 billion upfront for manifacturing and pay extra for a lot of excessive stock costs. You also lose the flexibility of making special budles or ship with updated software roms on those consoles. So in short, have 5 million left over in stock is a horrendious business decision.
We're talking about the launch of a new, consolidated, hardware platform in the company's primary business segment.
1. 5M units unsold in the pipeline isn't a problem as that's barely enough to keep most (probably not even all) major market retailers with on-shelf stock. We've seen how little physical presence 3M Switch units have had. We (those of us following the industry long enough anyway) have seen that same phenomenon with the PSV, Wii U, etc. when they were struggling to move units. Or maybe a better analogy for a struggling Switch, the PS3. At one point Sony had nearly eight figures worth of PS3's in the retail pipeline, seemed to have sorted that one out just fine.
2. Everyone, including Nintendo, has managed to release new bundles with units in the pipeline. Everyone deals with older firmware units in the pipeline after literally every firmware update. These re non-issues.
3. Extra units in the pipeline are only waste if they never sell, period. If Nintendo never sells 20M Switches they're getting fucking sold, so why the hell should the CEO, CFO, etc. of Nintendo care? They sure as fuck won't have jobs afterwards, 5M unsold systems in the pipe or no.
4. Lets review your math here. So a $200 per unit production cost, so $4B to push out those 20M units (plus factory startup, etc. costs, but those are simply cost of doing business). Assume at least a 2:1 tie ratio (pretty shit for a Nintendo console mind you) at $35 profit per (about first party industry standard), so $70 per console. Assume another $30 average per unit in accessory profits (also conservative given where Nintendo priced peripherals/accessories). That means every console sold in this format would produce $200 profit. Sell 10M of 20M units and you've broken even on material costs (not R&D and development) with another 10M units already in the pipeline. Sell 15B and you've turn around a profit of rougly $1B, again with 5M in the pipeline to return on the investment.
Nintendo is incredibly risk adverse in hardware production, to the point of actually costing themselves business. This has been proven with the NES Classic most recently but isn't unique to that. For every Wii U where they would have overproduced and taken and hit they've lost at least as many profitable sales early to mid-life on the Wii and DS.
The only validation for constantly shorting their own hardware on the manufacturing end is that it clears the pipe when they roll out revisions and more profitable builds. As the Switch is basically a fucking Tegra tablet with controllers snapped on the side the later of those two depends as much on their contract with Nvidia as real production costs (just go ask MS about that with the OG Xbox).
The first though is a valid reason. The current Switch is the opening move. Where it goes from here is anyone's guess but I'm betting we'll see as much versioning of the Switch as we've seen of the DS and 3DS. After all, Nintendo didn't fire all their hardware engineers despite moving to this one hybrid device future. After a clear two SKU differentiated entirely by color launch window (1-2 years minimum) they'll likely start announcing alternatives, like a smaller, more portable version without removable controllers. I wouldn't be surprised to see one with an X2 in it a la the PS4 Pro/Scorpio either. There's a lot of potential here for reselling the Switch to current owners via hardware modifications, something Nintendo has always monetized well on the handheld side but something also now introduced and validated by their competition on the console side.