Hi Rick, I was a bit quick to jump to conclusions, apparently they are not currently working on the IPv6 profile. It does seem to be quite a lot of interest for it out there, so I hope they get around to it soon ;)
It's all about perspective. Looking at the G2 solution the TX power @ 0dBm seems to be 120-135 mA. At 10.5 mA in TX the Nordic solution is an order of magnitude smaller ;)
I wouldn't be surprised if the average current is a hundred times smaller.
Personally I am excited about the improvements made by the WiFi module guys. WiFi offers much higher data rates and is currently more established than BT 4.0, but I think it's going to be a while until we will see coin cell powered WiFi nodes. As a rule of thumb anything over 20 mA is a no go for coin cells like the CR2032.
I can tell you that even the most optimized WiFi solutions use way more power, and cost way more, than BTLE. Nordic BTLE is a much better choice for low cost, coin cell powered sensors and peripherals, and with an IPv6 profile being worked on for BTLE you should be able to connect BTLE sensors to the internet fairly easily.
As for who is going to do this next I won't speculate ;)
Who will be next to merge BTLE and ARM cores in an SoC for this emerging Internet of Things market?
How will Nordic's parts and chose protocols vie with similar efforts from companies pushing Wi-Fi into lower power and cost markets?
The quick answer is that the M0 does everything much quicker than the 8051, while consuming about the same power in active mode (4-5mA at 16 MHz). This means it can stay more often in low power sleep modes, rather than having to run the MCU for a long time to do data processing. Added to that the code size is smaller with the M0 (I believe Arm claim about 30% compared to the 8051), because the 16/32 bit M0 instructions are more efficient than the 8/16/24 bit instructions used by the 8051.