Yeah, lots of people keep telling me that Intel will push Skylake back artificially, but you'd think they would have done it by now. Surely there are a lot of people like us who saw the delay in Broadwell and thought "Well crap, we could get Skylake with a completely new architecture in less than a year so let's wait." However, the truth is that mobile 64-bit ARM chips are quickly catching up to Intel (my iPad Air 2 benches about as fast as a MacBook Air, lol), and can be easily modified and have extremely low power requirements. They're going to be a big competitor going forward. I also think that it's likely Apple might make their ARM play next year, perhaps starting with an iPad/MacBook Air hybrid device. If no ARM MacBooks come out by Skylake next autumn, then I'll feel better about buying Intel. I don't want my apps to quickly become obsolete if Apple switches platforms again. And IMO an ARM iMac would at least come a year after an ARM MacBook Air hybrid device. But that's a lot of speculation on my part.
So isn't the next revision of TB supposed to be 50Gbps, and then 100Gbps after that? And wasn't the move to 50Gbps supposed to be using fiber optics combined with a copper wire for power? I wonder how close that is to being ready. If my math is correct, then I think PCIe 3.0 x16 maxes out at 126Gbps. So in theory later versions of Thunderbolt could nearly run a high-end graphics card at full speed. But by late 2015 PCIe 4.0 will be finalized, which should be twice as fast, so it might be less relevant by then.
I think late 2015/early 2016 is going to be a great time to upgrade, especially to those who are on the fence now and don't absolutely need a newer machine.