- Published: 18 September 2017 18 September 2017
Randall Stephenson makes an extraordinary claim. "You need this gig speed," adding, "Over the next few years you get 1 gig speeds ubiquitously." Randall, unfortunately, has the same attitude to the truth as the average politician, even when speaking at Goldman Sachs.
AT&T's actual plan doesn't go that far but is remarkable. Some percentage of the country - ?5-15% - will not be covered. The "gig" will be 75 to 300 megabits down, most times and most places. That's not too shabby by any standard and a huge jump over the 5 meg to 25 meg speeds most common on LTE networks.
In wireless, I assume he mostly is planning "Gig LTE." In certain test conditions, that does deliver close to a gig. In the real world, the best analysis available indicates speeds will usually be a quarter gig or less. (Article coming.) But "gigabit" is a magic word so I'm not surprised he's using it. Even 75 megabits to 85-95% of the U.S. is A GOOD THING.
I think he's going to do it. The other three six big U.S. and Canadian carriers have similar plans. So do major carriers in Myanmar, India, and many other places.
For the last year, I have been reporting that essentially all towers in the developed world with decent backhaul would be upgraded to LTE. Randall's comments confirm that will usually be at 100-300 megabits with capacity for a substantial amount of video. (3-4 CA, 4x4 MIMO)
"Gig LTE" is now the sweet point for wireless vendors. 3 or 4 CA, 4x4 MIMO, 256 QAM is now the standard configuration and will deliver 100-300 megabits to most. I'm delighted to report that reasonably robust system is becoming standard in countries like Myanmar and Thailand. In India, Bharti is leapfrogging to 64 antenna Massive MIMO, which I believe is the start of a trend.
("Gig LTE," is a marketing term for what has become the norm on new cells. 3 or 4 20 MHz carriers aggregated, 4x4 MIMO antennas, 256 QAM encoding. Bharti in India just announced they've started deploying Massive MIMO, with 64 antennas. So are Softbank in Japan and China Mobile, with many to follow over the next 18 months.)
On the wired side, AT&T is adding about 3M homes passed with GPON each year, which delivers a true gigabit down and 500+ meg up. Within about five years, 30-50% interritory will be able to get 800+ meg downstream, close enough to a gig for most practical purposes. That will include 15M+ with fiber home.
Randall also discussed the need for sub 10 millisecond latency, which I'm addressing in a second article.
Since AT&T and probably other telcos will bring 100 megabits plus near to ubiquity, the subsidies under consideration need to be reviewed.
His predecessor, Ed Whitacre, in 1999 promised broadband to "80% by 2002 and all our customers in the following few years." (Emphasis added.) Actually, they refuse to reveal the broadband figure but it is about 87%.
Here are Randall's comments, courtesy of Seeking Alpha.
think of the bogie -- or anybody who is delivering network services to the consumer or the business, the bogie has to be 1 gig speeds. And I mean 1 gig speeds whether it's fiber to the home or 1 gig speeds on a smartphone that has to be the bogie. And so, we're on a trajectory with our network evolution, LTE evolution that continues to move us towards that destination.
And we're using multiple technological architectures to get there, but obviously LTE there are LTE derivatives, LTEA and LTE-LA and so forth. With MIMO and so forth that take us down a path like in Austin where we have deployed this technology and we are getting 750 meg type speeds on our existing network in Austin. You are going to see us deploy that broadly over the next year or two. You're also seeing us invest aggressively as you said in deploying fiber, fiber-to-the home, fiber-to-businesses, and fiber to all the sale sites. And as we move into a world of 5G, fifth-generation networks, it's going to be operating at millimeter wave spectrum that we secured a nice footprint of this millimeter high frequency spectrum with high bandwidth, the ability to get incredible speeds, we have secured that. You'll see us begin building that out in end of 2018 timeframe. And so, this is how over the next few years you get 1 gig speeds ubiquitously. That is our target. We need 1 gig speeds. There will be used cases that drive the first deployments of this. Used cases for 5G, the most logical one that comes to mind are those that have really low latency requirements. It's one thing to have fast networks, that's great, you need this gig speed. But the other thing you need is really low latency like sub 10 millisecond latency. What applications and uses have those kinds of low latency requirements? Think autonomous cars, think drones, these are the kind of applications that have really low latency requirements.
And so, our network evolution is not just getting faster speeds than our mobile networks. That's critical. It's not just getting lower latency. That's critical. We got to get the latency out of the entire network. So, think about autonomous cars, where you'd have live maps and maps that are being updated in the cloud, you cannot have a centralized cloud infrastructure hundreds of miles away and have autonomous cars. That cloud has to be distributed. I love our position of being able to have the distributed cloud. Our software-defined networking makes us possible. But literally you have distributed cloud, distributed compute architecture, where you have it at every sale site, think 70,000 sale sites at every node, at every central office, multiple distribution points, where you can get the content as close to the autonomous car or whatever application you're thinking as humanly possible. That's what's required to get to this sub 10 millisecond speeds, so that you have what is from a human standpoint and a reality standpoint, instantaneous communication. We do not want an autonomous car that has 25 to 30 millisecond type latency. That's an accident. Those are safety issues. And so, all of this is critical to get these used cases that we all think about virtual reality and augmented reality will require these kinds of low latency requirements. And so, we're building this infrastructure now that we have the capital lined up behind it, we will be going aggressively, the standards, we've been working the standards aggressively. We think it's really important that we have global standards around this technology, and we're very, very close to having that done. And so, I look forward to bringing some of these services online 2018.