Now we know why Intel has changed the name of its long-awaited "Light Peak" connection: because it isn't one. Thunderbolt isn't photonics-based, and makes the connection via a good old copper wire. However, it does deliver the promised speed: 10Gbps bi-directionally over dual channels. In sum, you could have 40Gbps of data running over the same wire at the same time, should you ever have the need.
What we don't know is why Intel called it Thunderbolt. My guess would be that someone jokingly said "Thunderbolt and Light Peak" and the Intel office immediately went into Freddie Mercury-based karaoke mode, but don’t quote me on that one. It might have been the Apple office.
After Apple introduced a Thunderbolt connector on its latest MacBook Pros, Intel held an hour-long press conference in San Francisco, which I attended via what used to be the Post Office Telephone System. This made it hard to hear what was being said. I also couldn't see the demos, but I've already seen the real Light Peak in action courtesy of Intel Research, so I didn't really need to.
Basically, we still don't have Light Peak because PC manufacturers thought the optical system cost too much. Sad to say, Intel declined to say how much cheaper Thunderbolt might be, but said its cost was comparable to other advanced connection systems. This is the first reason why you shouldn't expect Thunderbolt to appear in many PCs in the near future: it costs more than USB.
The second reason is that Thunderbolt is much like Intel's PCI Express system but over a wire instead of via an expansion slot. It's going to need a chip on the motherboard, rather than a half-card in a PCI Express slot. However, it does mean you can run a monitor from a port instead of an expansion slot. "To the operating system, [external devices] look as though they're inside the computer," said Intel.
Thunderbolt lets you run better-than-HD video down one of the channels and a faster-than-eSATA data connection down the other channel. (They're not actually dedicated channels, but this might be a typical use-case.) Other protocols besides PCIe and DisplayPort are supported via translation.
The advantage is that you can reduce the number of connections to a thin laptop or portable device -- which is one reason why Apple likes it. But there are two big disadvantages for users. First, you need some way of separating out the different signals to different devices, which means either a break-out box or docking station, or a new monitor with Thunderbolt in and out ports. Second, you need to buy new devices that support Thunderbolt, such as an Avid media processor or a very fast external hard drive. If you are not going to do that, then you're better off using USB 3.0 or the 6Gbps eSATA connection that is often coterminous with a USB port.
During the press conference, Intel said more than once that it saw Thunderbolt as "complementary" to USB, not a replacement. Intel is still working to integrate USB 3.0 into its chipsets, and it says "USB is the mainstream i/o on products". By contrast, "Thunderbolt is targeted at applications that can't be supported very well using existing connections." These are typically in the creative media industries, particularly in sound and video processing (AJA, Apogee, Avid, Black Magic, Universal Audio).
So Thunderbolt is not the new USB, it's the new Firewire (which was the new SCSI). If you still get a kick out of daisy-chaining up to seven devices, Thunderbolt is for you.
I did email Intel a question for the Q&A -- "Do you think Thunderbolt will be more successful than Firewire? If so, why?" -- but as far as I could hear, I didn't get a reply. However, Intel did answer another of my questions: "Has Intel abandoned the idea of a light connection or is a real Light Peak still on the horizon?"
Intel's Jason Ziller replied: "We haven't abandoned Light Peak, and we still believe that optics is in our future. We're still doing research and development in that area, and we'll assess when it's the appropriate time is to bring that into the platform."
All of this makes clear why Intel allowed Apple to lead the way in launching Thunderbolt over copper. First, Apple doesn't have to care about the cost because it already has very fat profit margins built into its prices. It can bundle Thunderbolt whether users want it or not. Second, Apple controls its whole ecosystem, and can increase its lock-in by fitting Thunderbolt to its other devices, including monitors, so they all connect together to the user's benefit. Third, Apple's users in the creative media industries are very likely to adopt it because they can actually derive benefits from it. Fourth, Thunderbolt will soothe the pain of those Mac users who felt Apple should have stuck with Firewire instead of moving most things to USB… and who subsequently felt annoyed that Apple was left behind in the move to USB 3.0.
If Dell had introduced Thunderbolt, the majority of its buyers would have ignored it, and many of the rest would probably have bitched about it. With Apple, however, Thunderbolt means happiness all round. That also includes companies keen to sell high-priced high-speed peripherals, such as LaCie.
http://www.youtube.com/watch?v=kidmWiqKzqY Jason Ziller shows the real Light Peak