Networks of the world, 2019. A first draft.

4G will be almost everywhere with peak speeds often over 100 megabits. 100 megabits will be slow for wired connections in the developed world. Most developing countries - except China - have very few landlines and it's not clear wireless will have enough capacity for much video over the net. Prices in strongly competitive markets will be flat to down because the costs of modems, Internet transit, and nearly everything else are going down. In the more common weakly competitive market, like the U.S., prices will creep up unless the regulator is strong.

Two billion more people will be connected to the net as LTE phones drop under $50. Africa will have more Internet connections than the United States by 2017, more than the 315M population of the U.S. India is growing almost as fast, and China already has twice as many connected as either the U.S. or Europe. The rich countries are today 40-45% of connections; by 2019, that will be down to 35-40%. 5% to 25% of homes will be screwed as rural areas have limited coverage and competition.  

A reporter friend asked what broadband will look like in five years, so I worked this up. Improvements welcome. Some of this will be wrong like any other five year prediction on a fast-moving technology. I'm sure I have some errors covering this much. My sources are strong, however, starting what the companies say publicly and on Wall Street. This is informed by data from Cisco (probably the best), Ericsson, Qualcomm and many academics in both wired and wireless. If you email me with an improvement, I'll thank you. This email address is being protected from spambots. You need JavaScript enabled to view it.

Wireless

  • LTE with peak speeds over 100 megabits down will reach 80-95% in the developed world. A typical LTE Advanced cell will share 600 megabits or more, and some will be over a gigabit. That's by the standard measurement of cell capacity. LTE-A since 2011 has been targeting 1.5 gigabits and 1.2-1.4 has been produced in the lab.
  • In Q1 2015, the highest speed Akamai measured is 149 megabit down, exactly what you'd expect from a cell designed for 150 megabits, the design target for LTE in 20 MHz. Akamai State of the Internet page 42. Japan peaked at 126 megabits and Thailand 106. More typical is Britain peaking at 91 megabits and Germany at 70 megabits. Most developed countries peak between 45 megabits and 90 megabits. I don't know whether that's because the carriers limit peak rates or because the cells don't have 150 megabits to share.
  • The U.S. is the major outlier, with a peak rate of only 18 megabits and an average rate of 4 megabits. The discrepancy in peak rate is so extreme I would want to examine how Akamai gathers their data. It's actually not implausible that the U.S. carriers so heavily throttle their customers. The U.S. has one of the worst mobile networks among rich countries, even when adjusted for population size or density. Robin Bienstock calculated Spain has about twice as many cell sites as the U.S. and I believe several other Europeans are similar. China Mobile has already deployed 500,000 4G cell sites compared to the 20,000 to 50,000 of a Verizon or AT&T. This isn't isn't a precise comparison. Verizon and AT&T refuse to release the number of sites they have. No one I know breaks out small cells and other factors. The U.S. population is 315 million while China has about four times as many people. China Mobile is certainly building a far denser network than the Americans, especially because they adding hundreds of thousands of sites over the next year and a half. The U.S. could be called a world leader from about 2009 to 2012. Verizon's CDMA network had no graceful 3.5G upgrade. They decided to take a risk on the then new LTE technology and it worked well. Since then, several Europeans have leapfrogged Verizon's network. Several have much faster networks according to Akamai and other testing. Barack Obama recently claimed the U.S. had the world's best wireless network, a politician's truth. He should get staff factcheckers who aren't three years out of date.
  • In Q2 2015,, several telcos have deployed "carrier aggregation," using two 20 MHz channels for peak speeds up to 300 megabits. Others are starting to deploy 4x4 MIMO, using extra antennas to double speed again. No engineer doubts a gigabit (shared) is coming. Most expect 5G and even faster speeds starting 2020-2022.
  • Average speeds measured by Akamai are typically 10-20% of peak speeds.  Britain, for example, has an average speed of 20 megabits against a peak speed of 91 megabits. France averages 8 megabits against a peak of 48. The U.S. averages 4 megabits while peaking at 18.
        Probably half the phones in England still are the slower 3G units.  People too far from the cell site get slower speeds. The prices are set up so that many customers don't buy the higher speeds. Many throttle down customers even when there's plenty of capacity (90+% of the time) because they aren't set up yet just to do so when there's real congestion. Actual congestion is much less than people think and concentrated in a few cities for a few hours a day.
  • I'm confident that most large networks will be able to deliver 50-100 megabits download most of the time to most customers. Again, folks who happen to be far from the cell tower won't get that and there may be many other limits. But with 600-900 megabits (shared) on most cell towers, peak speeds well into the hundreds of megabits will be common. When you see the data, you discover that sharing works much better than anyone but an expert would guess.   
  • In competitive markets, consumer prices will be flat or down. LTE is much cheaper to operate than older technologies, perhaps 30-70%. $2-3/month more will easily cover the added transit and operating cost into the hundreds of megabits.
  • Less developed countries will see remarkable improvements, with 50-90% LTE coverage. Smartphones will be cheap enough that literally billions more will buy them, some second hand. I can't be sure about the number of people who will pay for mobile broadband but it should be well into the billions.
  • Consumer prices will be flat or down where competition is working. LTE is much cheaper to operate than older technologies, perhaps 30-70%. $2-3/month more will easily cover the added transit and operating cost into the hundreds of megabits.  Most markets are only weakly competitive and politics/competition will determine pricing. The U.S. will probably continue to be much more expensive than Europe because our regulator is so weak.
  • There will be 7-8 billion smartphones on Earth, including an active resale system. An amazing percentage of poor people will own one. Much of Africa has 40-80% mobile voice today and LTE smartphones are now down to $50-75 in telco volume.
  • Spectrum will always be a factor in wireless but probably less than today. Speeds will definitely go up, as will network capacity. Whether most people have a cap high enough for much video watching is a political decision. 

Wired

  • 70-90% of the homes in the developed world most of the time will be able to get 100 megabits down, 10 megabits up in five years.
  • Many landlines will offer half a gigabit, including nearly all cable. 200-700 megabit G.fast will deploy to tens of millions in Britain, AT&T, Swisscom and Belgacom according to the companies. In apartment buildings under 20 stories, many will get the higher speeds and maybe a gigabit. Deutsche Telekom is less ambitious, using 35b DSL for 100-150 megabit downloads and praying cable won't use their gigabit to win customers. France and Spain are going heavily into fiber designed for a gigabit.
  • Cisco's VNI projects "Broadband speeds will double by 2019. By 2019, global fixed broadband speeds will reach 43 Mbps, up from 20 Mbps in 2014." Not all homes will pay for the highest speeds, but 50 meg will generally be the minimum except for the cheapest offerings. 
  • 5-30% of homes will be effed, especially rural areas and countries with weak competition. 
  • These figures are highly likely, based on company plans (I've researched most of North America and Western Europe) and the presumption that carriers have smart leaders and want to make a profit. (The latter is certain, the former?) They also correspond to five year global equipment sales forecasts, likely growing at about the rate of customer additions.
  • Gigabit cable and 100 megabit LTE networks will cost the same or less to build than 10 megabit networks cost in 2014. The upgrade cost will often be modest and fit in the current capital spending budget. The cable upgrade will cost 2-5% of revenue each year and fiber more. That's cheap enough even monopoly-like markets can be profitably upgraded. 
  • In most competitive markets, consumer prices will be flat or down. LTE is much cheaper to operate than older technologies, perhaps 30-70%. $2-3/month more will easily cover the added transit and operating cost into the hundreds of megabits. Most markets are only weakly competitive and politics/competition will determine pricing. Most likely, the U.S., Britain, and Germany will see prices creep up on wireline. Indian and African prices are already so low for most services they may go up.

------------------------

Gigabit cable is deploying at Videotron and Suddenlink. Half-gigabit DSL (G.fast) is beginning a 4,000 home trial in Britain and likely to deploy to most of Britain and much of AT&T in the next 5 years. Fiber is taking off, with 100M homes passed in China today and probably twice that in five years. France and Spain will be mostly fibered. Italy may be an extreme laggard. As I write in 2015, Telecom Italia is planning only 40-70 megabit non vectored DSL and a very slow deployment. They can do that because they have no cable competition.

The most important change will be the billion+ Africans and Indians likely to connect via smartphones. Africa will have more Internet users than the U.S. in 2017, I calculated with the help of Cisco. For the developed world, the main change will be a welcome increase in speed and capacity.