Bluetooth is dead, analyst says, because of Intel's leap to independent ultrawideband standard that replicates USB simplicity: TechDirt picks apart the arguments so adeptly that I largely point you to read them. My own analysis is that UWB will absolutely displace a few current technologies and become the core of new ones that require proximity-based high-speed low-power connections. This includes desktop hard drives, camcorder and digital camera file transfer, and all sorts of input peripherals.
The TechDirt folks point out that Microsoft overpriced their Bluetooth offerings, but they don't mention that Microsoft failed to provide the same kind of baseline, baked-into-Windows XP support for Bluetooth that they did for Wi-Fi, which has made Wi-Fi so easy to use in XP. Bluetooth, in contrast, is available as a toolbox set to Microsoft developers, but it doesn't have a uniform XP interface that loads Bluetooth drivers underneath it. (This might be an NDIS problem, too; if Bluetooth drivers aren't standardized with NDIS, a hardware abstraction layer, it can be much more difficult for XP to talk to the drivers.)
TechDirt is totally right that it's premature to write Bluetooth's obituary because UWB-based devices are years away. Further, it's definitely still a possibility that a final 802.15.3a spec will be wrung out that will use the Bluetooth subset 802.15.1 as the basis, and thus keep Bluetooth alive in name, but not in its current fixed hardware implementation. (Tom's Networking analyzes Intel's move and thinks Intel's participation and the whole Multi-Band OFDM Alliance in the IEEE process is at an end.)
Even more interesting is that we have the technology and the standard today in the form of 802.15.3 to offer 11 to 55 Mbps Bluetooth (the data spec, not the hardware), but that .3a is so much more compelling that we'll probably never see the interim .3 turned into equipment.