He is half right.
Maxwells tiled based rasterization does give it an efficiency edge as it is able to achieve higher levels of utilization.

As for the Memory. Picking an arbitrary number liks "25.6 GB/s" is completely and utterly pointless and is where he looses any kind of legitimacy.
It's bandwidth is not directly comparable to the Xbox One or Playstation 4. ESRAM or not.

The Switch takes advantage of Maxwells inherent architectural advantages such as colour compression which allows the Tegra chip to "eek" more out of that 25.6 GB/s than the raw number implies. - Plus Tiled based rasterization gives it an efficiency edge there too.

It is easy to get the game running on the hardware.
It is difficult to get it to run at a playable level.

Half the memory is a "Maybe". - We have no idea how much DRAM the Switch OS will reserve for itself and background applications.
The Xbox One and Playstation 4 reserve a few gigabytes. If the switch is more Stringent, it could theoretically only be missing out on 512Mb-1024Mb of DRAM for games.

GPU is a little trickier. The Xbox One and Playstation 4 chips are certainly a league ahead when it comes to Asynch compute.

It is easy to port to the Switch. The hardware, OS and API's make sure of that.

But porting is only half the problem, making it playable is another matter entirely.

You should care. Plenty of people enjoy western releases which could be consumers of the Switch which drives up Platform sales which then also make the switch more attractive to Japanese developers.

Nintendo seem to always do "Their own thing". - Hate it or love it, it is what it is.
But yes, they are the bane of Multiplatforms.

I'd be surprised if it was able to direct port. But the rumors are, it's still rather easy to port to and that's the most important thing for Switch. It'll take a bit of work to port but nothing that will necessarily need an entire team to do a port. That's what I'm thinking.

Yeah, this is nothing complicated.  

• Lightweight game that would run on the hardware natively = easy port
• Less demanding AAA game where certain effects and settings would need to be adjusted down = relatively easy port
• High level AAA game where actual content and level design would need to be scaled back to achieve stable performance = expensive port

That real danger is that the first two on the list might skip the Switch if the last one doesn't show up at all.  Publishers might have the perception that the audience won't be there without the big boys pulling in the consumers.  That's where Nintendo has to offer support and incentives out of the gate.  They should be doing their best to get the easy ports and relatively easy ports on the system, even if some of the expensive ports skip out.

Just buy the switch as if it was a handheld instead of a console, that way you won't feel bad the console has been left behind.

The dream is over...

x1, 25.6 gb/s and maxwell are all confirmed?(for the final product? i dont doubt it could be one of dev kits version)

You guys are a strange bunch!
I get not hyping yourselves up to be disappointed, but to just believe anything negative for the sake of bringing a system down before it's unveiled in just 2 weeks is just weird...

We have actual developers making comments on what it's like working with the devkits, but somehow an ex employee that is going by Internet rumors has more credibility then everyone else. That just seems silly to me and if this guy was so great at his job he'd probably still be employed by Ubisoft, even though Ubisoft is one of the last companies that I would hold gpu optimization in high regards. Their games run like crap on exceedingly more powerful hardware than XbOne and PS4.

I think after Eurogamer's clock leaks, people weren't expecting otherwise

Dammit Nintendo get to work on making the Switch more powerful!!

we've heard this easy to port thing before. and we saw how that turned out...fool me once