Ever since Intel launched its high-performance Thunderbolt interface, it’s been fighting to win market share away from USB devices. There’s never been any question of which standard was faster — first-generation Thunderbolt blew USB 3.0 out of the water — but higher device costs, expensive cables, and limited OEM uptake have limited Thunderbolt’s market penetration. Today, with the announcement of Intel’s Alpine Ridge, that schism is effectively over. Third-generation Thunderbolt won’t just be faster than USB 3.0 or USB 3.1, it’ll offer all of its performance over a reversible USB-C plug.
Thunderbolt 3, aka Alpine Ridge, accomplishes this by integrating a USB 3.1 controller directly on-platform, neatly bypassing the entire “Should I use Thunderbolt 3 or USB” argument altogether.

Here’s what this means, in aggregate: You can now run DisplayPort, Thunderbolt, USB 3, and PCI-Express over a single Thunderbolt port. Even better, the USB 3.1 port in question is USB 3.1 Gen 2, as opposed to the USB 3.1 Gen 1 (aka USB 3.0), port that Apple uses for its new MacBook’s USB-C port. Unlike Thunderbolt 2, which kept total bandwidth the same as Thunderbolt 1.0, but allowed for more flexible bandwidth sharing options to improve performance, Thunderbolt doubles real bandwidth to up to 40Gbps.

Intel doesn’t include PCI-Express on its list of interfaces, but at this point, it could. 40Gbps works out to 5GB/s worth of bandwidth — which means Thunderbolt 3 can provide roughly a bit less bandwidth than an older PCI-Express 2.0 connection. The new standard offers support for up to two 4K 60Hz panels or a single 5K 60Hz display (this makes sense, as a 5K 60Hz panel is roughly 1.8x as many pixels as a 4K panel). This works out to an x4 PCI-E 3.0 connection. While not exactly suited for multi-GPU configurations, it should be enough bandwidth to run a single card at near-full speed in the vast majority of cases.
The combined Thunderbolt 3 / USB 3.1 standard will offer up to 100W for laptop charging if the manufacturer certifies the port to draw that much power and 15W of power for standard, bus-powered devices. 15W is significant because that’s more than enough energy to run most hard drive docks or external peripherals, including low-power portable displays.

The new configurations are also supposed to be flexible. Second device ports offer all the same functionality as the computer port, meaning you can plug a USB device into a monitor, which then connects to the computer via a Thunderbolt cable. Up to six devices can be daisy chained. Intel is also pushing the entire concept of a Thunderbolt external graphics dock for mobile gaming, and even the use of Thunderbolt for networking purposes.
One of the problems with Thunderbolt has been its reliance on expensive cabling. While the standard has never required the optical cables that were originally meant to debut alongside it, the available copper cables have often been expensive. Intel has improved the situation somewhat: Thunderbolt 3 will support passive cabling up to two meters with bandwidth of up to 20Gbps, 40Gbps active copper cables for full 40Gbps connectivity (again, up to two meters long), and 40Gbps optical cables arriving in 2016, with lengths of up to 60 meters.

Could Thunderbolt 3 finally catch on?

I’ve always been dubious about Thunderbolt’s chances of finding wide acceptance, but this latest announcement makes me think claims of Thunderbolt’s demise have been greatly exaggerated. FireWire may have lost its fight with USB, but FireWire never took the step of integrating a USB controller into the same package.
The two major questions remaining are how much Thunderbolt chips will cost and when they’ll ship. Intel has said it expects to begin shipping devices by the end of this year, which implies the new technology will show up on Skylake. If the licensing fees are reasonable, high-end devices with combined Thunderbolt / USB 3 ports could quickly become commonplace.