Gigabit is the new 100 Mbps for this decade, with everyone from AT&T and Google to the Obama Administration onboard with the idea of "ultra-high-speed" – because higher speeds just aren't enough to keep up with the Japans and South Koreas of the world.  However, terabit network speeds are what carriers are going to be boasting about in 2020.

Why will terabit be hot? Consider where cable, fiber, and copper are at now and where wireless wants to go.  The cable industry is currently deploying gigabit to multi-gigabit speeds through DOCSIS 3.1; you can go up to Anchorage Alaska and get a gig if you are in the right neighborhood.  AT&T, CenturyLink, Frontier, Google, and numerous other carriers, both local and municipality, are rolling out gigabit services with a mixture of fiber and copper at the last mile.  Verizon has teased a 10 Gbps FiOS capability with the potential to go up to 40 to 80 Gbps "as the market demands," while it continues to putter around with 500 Mbps as its currently available top residential speed.   Some municipal carriers, such as Chattanooga, have also discussed plans to deploy 10 Gbps if customers want it.

The wireless industry currently measures LTE speeds in terms of multiples of 10s of Mbps, call it between 10 to 40 Mbps on a good day, depending on what country you are in, how close you are to the cell tower, and how many other subscribers are also downloading.   Some networks, with the right phone and enough licensed radio spectrum, say they can get anywhere between 150 Mbps to 300 Mbps on existing networks today, with the potential to move up to 450 Mbps in the near future. 

In addition to licensed spectrum, the cellular industry is planning to rollout LTE-U (unlicensed), enabling the use of 5 MHz unlicensed spectrum to deliver download speeds of up to 300 Mbps. LTE-U would be deployed in a small cell with a regular licensed LTE radio, providing high-speed off-load/download on unlicensed spectrum.  Demonstrations are taking place in China and are planned to take place in the U.S. by the end of this year, but there's a political/technology battle between cable and cellular carriers as to if LTE-U will stomp all over existing Wi-Fi usage.

By 2020, the wireless industry hopes to formalize 5G as a standard, enabling gigabit to multi-gigabit speeds to a cell phone. There's a lot of heavy lifting involved to make 5G work, but Japan and South Korea carriers along with Verizon say they'll have pre-5G high-speed tech rolled out prior to 2020.

Simple math of aggregation dictates how fast you need to go from the edge of the network to the core, regardless of technology. If you are delivering gigabit speeds at the edge, you need at least 10G to 100Gbps (or faster) at the core.  Bumping up to 10 Gbps per second, you need 100 Gbps or more – exactly how much more is subject to debate and network engineering – at the core.

With larger businesses asking for anywhere from gig to 10 Gig connections, 100 Gbps core speeds are essential.  And here's where economics and network needs get tricky.  Currently, the IEEE is working out a standard for 400 Gigabit Ethernet, with formalization expected by December 2017.  How expensive 400 GbE will be and if 100 GbE will get cheaper are open questions at this point in time, but I'd speculate you'd see someone figure out how to make things faster and cheaper.

 Nobody's worked out how a standardized Terabit Ethernet (TbE) connection will work. Outside of university lab trials, CenturyLink and Ciena have demonstrated the ability to deliver terabit speeds on an existing fiber network in central Florida by packing a lot of wavelengths of light together. It's not standard, but it worked well enough for the two companies to issue a press release about the feat. 

Regardless of how it will be done, terabit has to happen within the next five years to support higher speed home broadband networks and the promises of 5G. How fast and how deeply will depend on the balance between cost, network engineering, and customer demand for reliable multi-gigabit speeds at home and work.