AVR Freaks

 I have a project that requires a microcontroller in SOIC package that support I2C and SPI independently.

I guess, in addition to the above comments, I'd ask:

Building 1, or 1 million?

If you are only building a few boards, then who really cares about a few dollars difference in micro's price, or having a few "spare" / "too many" pins.

You could select a $2 - $3 Arduino Nano type board, with a Mega328, power supply, reset switch, LED, USB interface, and I/O pin breakout; that has both SPI and I2C on separate pins, and be working on your first cut prototype!

(I2C is on PortC.4 and PortC.5 pins, SPI is on PortB.2,3,4 and 5 pins).

Easy decision.

JC

SOIC seems to be a really nice package to work with

There are certainly lots of different naming schemes...

Mega328's come in TQFP, which is still a "gull wing" chip, I think.

If you can solder the SOIC, then I think soldering a TQFP would be essentially the same.

(I've not looked at the tech spec's, and pin spacings, etc...).

I agree, VQFN and MLF have the tiny contacts on the edge of the chip without any external leads are both intimidating and challenging.

I've soldered a few of those, and I swear at myself for designing that chip into the system every time, but I've never had a failure on the ones I did solder.

Flux and practice, and a good magnifier, are the keys to success in this regard.

I've never attempted to assemble a board with ball grid, (UFBGA), chips.

I suspect that doing so won't happen in this life time.

There are numerous prior Threads on soldering.

I recall my hesitancy, (fear!), of switching from years of through hole component designs and board builds to surface mount.

I now prefer the surface mount design, and only occasionally do an "old school" through hole design, primarily to use up some old components.

Make the transition!

JC 

That's the right decision. You'll see it's much easier than you think, actually it's faster than soldering a DIP, because you just drag the solder across the pins, check for solder bridges, correct them and that's it. The most important tool is a good quality flux. But since you already know how to solder SOIC, it's the same, just with smaller pitch (as mentioned in #13).

I would like to be strategic in what I invest my time learning. Do you see value in learning this new lineup

 Avoiding Architecture and Vendor Wars entirely, I think I'd say the peripherals on ATtiny814/etc are somewhat typical of the more complex peripherals that ALL microcontroller vendors are putting on ALL of their chips these days, and therefore time spent puzzling out how the peripherals work is likely to be "valuable practice" regardless of final chip decisions.  Perhaps the "usefully familiar" CPU architecture would be helpful.