I think eyefinity is cool with seamless monitors. But I don't think there is any comparison in engineering prowess here. Maybe ATI will do something next gen to really impress architectually. Right now, they are going for efficiency, and small die as possible, and people confuse this with superior engineering on an otherwise anemic chip.
I understand you are more or less a compensated Nvidia spokesperson so you're going push Fermi's positives, but when you look at this thing from a more neutral set of eyes it's not this engineering marvel you want us to believe it is. It's design makes it slower and more power hungry than two of AMD's smaller dies (which AMD does put on it's fastest card).
AMD's small scalable design philosophy has allowed them to get their entire line up of parts out well before Nvidia could let one trickle in to the market. Now that Fermi is here, what we have is a part that is ~15% faster than AMD's second fastest part, but it does so by being loud and useing a LOT more power. Fermi has some ~800 million more transistors than Cypress, uses at least 170mm2+ more silicon (that's almost an entire additional Radeon 5770 worth of silicon), why isn't it much faster? AMD's fastest part trumps Fermi by a wide margin and even manages to do so while being more power effiecient, cooler running, and quieter despite having two GPU's and more memory. I'm just not seeing this engineering miracle in Fermi that you seem to see...
I think AMD out engineered Nvidia this time around. Good luck pushing this part in forums, it has some serious drawbacks.