A global analysis from 2020 to 2025, with estimates that cut through the hype.
“I will tell you, there’s a lot of units coming.” Liam Griffin, CEO Skyworks
5G is not going to change the world, no matter what you hear from Xi Jinping, Donald Trump, Boris Johnson, or Angela Merkel. None of them have any idea what they are talking about. The hype level is far beyond anything else I’ve seen in 22 years of reporting.
~63 million people had 5G phones at the end of June 2020. Demand is exploding in Asia, mostly China, where decent 5G phones sell for US$214-270. Yearend 5G will be between 200 million and 240 million.
Advanced wireless delivers useful improvements for us in the industry. Hans Vestberg at Verizon estimates his costs are down 90%. But performance generally is disappointing. Latency is typically 25-40 millisecond, not close to the promised 1 ms. With few exceptions, speeds are 50 Mbps (some low-band) to 400 Mbps. Most of Europe is 100-200 Mbps; Korea claims much higher speeds.
There are no exciting new apps that don’t work fine in 4G. After 18 months, none. I hope some thrilling new apps will appear over time. Some “use cases” are false narratives. Better and cheaper wireless is good for all of us, but not life-changing.
Open-RAN is exciting and working in 4G. Rakuten has 5,000 radios in service. CTO Tareq Amin has had to spend hundreds of millions on custom chips because Open-RAN is not quite ready. It remains the logical choice for new networks and new areas, but Vodafone CTO Scott Petty doesn’t expect primary deployments of Open-RAN until 2023-2025.
This will soon be part of my book, 5G. I have drafts of sections on applications (few,) strategies (take advantage of 5G capacity and low costs,) companies (watch Jio, Rakuten, and the new software vendors, Mavenir, Parallel Wireless, Radisys/Jio, & Altiostar,) and informed regulation, Ask me for 0.7 versions if these are important topics for you.
Future updates @analysisbranch
5G H1 figures
August 6 LG expects 5G phones below $200 by yearend. T-Mobile US turned on a few features and put out pr about 5G SA core. No data on improved speed or latency, just “up to” bs. It appears to be a “lite” 5G core, implementing a few features but not ready with most. (Limited data)
The biggest change is Nokia and Ericsson fixed a bug that reduced reach of 600 MHz spectrum by about 30%. 600 MHz NR is still probably slower than 4G.
Verizon is making a big push for 4G fixed wireless.
Inseego has a slew of orders for FWA routers. The 5G version has an improved antenna that Verizon is very hopeful about.
Pompeo announces the Great Firewall of Washington + 30 other countries is a huge policy story that will affect many things but not 5G standards from 3GPP. China is too strong. Ericsson and others have no choice but to follow because China is 70% of 5G. Result: on the technical level, 5G will have one world standard with China central.
August 3 NTT DOCOMO reached 149K 5G in June and 240,000 by the end of July ** Far EasTone in Taiwan “tens of thousands.” ** Ooredoo Qatar 200,000, more than anyone claims in Europe. ** Realme launched V5 5G, a phone with 4 cameras and good specs, for US$214
July 28 Added 4 million 5G fixed wireless routers and pucks, as suggested by Daryl Schoolar of Omdia. The figure is mine, and really a wild guess. I also added the July reports.
Big hubbub about “Deterministic Networking.” This essentially extends QoS/network slicing to connections across networks. the Internet should support quality of service and guaranteed low latency transmission, including when traversing multiple networks. For example, a German security agency in Hamburg served by DT might want a guaranteed connection to NATO forces in Italy served by Vodafone. Niantic might want a guaranteed low latency connection to the US and Nigeria for multi-player Pokemon Go.
The telcos think they will be able to charge tens of billions more for guaranteed low latency. They are probably fooling themselves, but meanwhile are exerting massive pressure to redesign the Internet in line with their hopes of controlling everything.
Delivering cross-network guaranteed service has been approached often over the years, but it hasn’t worked well enough to become a large-volume mainstream service. See the comments of Columbia Prof Henning Schulzrinne, who understands it much better than I do.
They are exerting enormous pressure to make this happen as quickly as possible, although it’s at least 2-3 years away. Supporters include Deutsche Telekom, British Telecom, Verizon, China Mobile, China Telecom, and suppliers including Ericsson, Cisco, and Huawei. There is also strong support from the Chinese government and I believe from the security agencies of many other countries. Of course they want tighter control of the Internet.
De facto, the 5G SA core already provides for QoS/network slicing on a network. Deterministic networking is an attempt to ensure performance across multiple networks.
It’s not clear who will pay very much a low latency product, other than government surveillance. Verizon thinks game services will pay, but that’s limited by the likelihood that your game opponent is not on the same network.
The other tool for low latency is Edge servers, but few in the West are committed. Verizon has turned on 1, with 10 more by the end of 2020. It will need thousands to deliver most of the applications they tout. DT is Germany is a little ahead. No one in the West has firm plans for Edge in volume despite all the brave talk.
Update July 10 The results from independent 5G tests are dismal, some included below. It appears that mid-band has a much bigger problem with walls and windows than expected, but I don’t have enough data. If so, much of my strategy suggestions are off-base because they assume a larger increase in capacity.
Qualcomm announced an even cheaper 5G chip for later this year. At those prices, most Android phone buyers will choose 5G.
17.5 million 5G phones shipped in China in June, which I believe is more than the six month total in the entire rest of the world. Almost all western telcos have such low counts they refuse to release figures. The 6 month total is 63.6 million. 10-14 million were sold in 2019, so the total 5G phones in China are ~75 million. That’s at least three times the total in the rest of the world. However, it is much less than the carrier figure of > 100 million “contracts,” which includes many still using 4G phones.
60% of phones sold in China in June were 5G. As peices come down elsewhere, I expect similar trends elsewhere.
MTN in South Africa has upgraded hundreds of base stations. End update.
Verizon claimed latency is 30 ms, although it sometimes measures lower. That’s 30% lower than 4G averages, although 4G latency is also falling. T-Mobile’s tested average 5G speed is 49 Mbps and AT&T’s 61 Mbps. The Open Signal chart at left summarizes over 10,000 tests. Compare it to the 4G results from Canada, 69 Mbps at Bell and 75 Mbps at Telus.
Mid-band spectrum allows 5G (and 4G) to reach 100’s of megabits, as confirmed by the British 5G companies and the Korean data from Open Signal.
Ask yourself, what can you do at 250 Mbps that you can’t do at 75 Mbps? 4K video typically is 15 Mbps; few of us watch more than three 4K videos at once. Huge downloads are faster, but how often do you download huge files? All claims of economic benefits depend on superior applications, which aren’t on the horizon. Nearly all the “studies” that find economic returns describe applications like IoT and connected cars that work well in 4G. The claims are unsupportable
I’ve interviewed over 100 senior people in the industry, including CTOs of world-class carriers and many of the inventors. Almost all are horrified by how the marketing people, the politicians, and some dreamers abuse the facts.
So what is 5G?
5G is a marketing term for two dozen important improvements that began in 2010. Networks will be (modestly) faster, have more capacity, and cost far less per bit delivered.
You will be able to do more, including top quality video on your phone. About 200 million of us will get 5G phones in 2020 as decent 6.5 inch phones with three cameras now cost US$228-$285 in a competitive market. Connections should rise to 500 million in 2021. By 2022 or 2023, prices will be so low that more than half the phones sold will be 5G.
The 10X speed improvements, 1-millisecond latency, and gigabits for everyone are pipe dreams or will take years to come out of the lab. T-Mobile USA CTO Neville Ray gave sworn testimony in the Sprint merger that 5G speeds will go up 20% to 50% at best.
Typical 5G speeds will be 100-300 Mbps, while decent 4G is now usually 50-150 Mbps. Verizon specifies 30 ms latency, not that much better than the 35-50 ms latency typical of 4G. [cm_simple_footnote id=”500″]
You will not spend your life in a virtual reality headset. Remote surgery will not become common. Autonomous cars do not require 5G and the carmakers have no intent of tying you to a network that won’t cover most roads for years. You can’t pull over and stop the car when you are out of range of wireless range.
Much will improve in five or ten years, but that’s too far ahead for me to project.
This publication is distributed under a Creative Commons Attribution-NonCommercial License CC BY-NC Cory Doctorow has proven it’s possible to give away your work and make it up on related sales. I welcome consulting, do speeches, and expect to make money on the book. I have several Zoom presentations that can bring your team up to speed. There’s a lot here that is not common wisdom, but I believe my research is solid. My charges are reasonable. email@example.com. 347-603-6442 in New York.
As I write in July, 2020, over 80 million 5G phones are connected. Here is test data from earlier in the year.
As you can see, the speed of actual 5G averages between 47 Mbps & 240 Mbps, except Verizon. Numerous other tests show similar results to Open Signal.
Verizon’s millimeter wave 5G does have great capacity, typically three times more. But it is only available to 0.4% of the U.S.
As mid-band spectrum is released in the U.S., speeds will likely increase to the 100-300 Mbps range. Only occasionally will 5G speeds be much higher than that for years. It would break the laws of physics.
100-300 Mbps is fine but is only a modest increase on 4G manufactured today. PC Mag has tested a 4G site in Manhattan at over 500 Mbps using LAA. [Footnote:] LAA and 4-carrier aggregation do not yet work in 5G, so the actual 5G speeds often are less than 4G on the same network. Really. [end footnote]
Four and five year old LTE is likely 30-70 Mbps, but LTE equipment today is normally 75-150 Mbps. The average 4G speed on Telus already is 75 Mbps. The same is true on the upgraded parts of AT&T and Verizon.
[Most reports like this are mostly used for numbers, which everyone knows may not be accurate. 3 or 4 years out in new technology is almost impossible to get right. I have the best numbers I can offer, but my intent here is to explain what’s going on so the reader can judge the accuracy and adapt when things change.]
Sidebar here: How the industry in 2018 bastardized the meaning of 4G. In frequencies below 2 GHz, 4G software with “5G NR” software is now called “5G.” Unfortunately, 5G NR software does almost nothing for performance below 2 GHz. The actual gain if capacity mostly come from Massive MIMO and the newly available mid-band spectrum.
Why will 200 million people buy it in 2020?
[This section is too long and most will be moved down in the report. I’ll add a sidebar here explaining the most important technologies were Massive MIMO, carrier aggregation, and systems that could use mid-band and millimeter wave spectrum. Skip a few pages if al you want are the numbers rather than the history and the gossip]
5G phones have now come down so much in price many sensible people will go for 5G. Decent 5G phones now cost as little as US$228 in China for the Xiaomi Redmi 30. That’s a 6.7′ phone, with a quad camera and a 120 MHz screen refresh. The price was dropped to US$197 for the 618 sale day. 5 other decent phones are under $285.
13 phonemakers are struggling for a market with room for only 4 or 5 to make a profit. The price war has dropped prices by about $100 from where everyone expected them to be by this time. Qualcomm predicts phones will be under US$150 by the end of the year.
5 chipmakers are now delivering 5G chips: Qualcomm, Mediatek, Samsung, Huawei, and UNISOC/Spreadtrum. When Mediatek started shipping, Qualcomm is rumored to have dropped prices by $20-30 to hold on to market share. 4 multinational vendors offer the important radio frequency front end. BOE now has enormous capacity for screens, winning share from the Koreans.
The Chinese prices will inevitably come to the West. It costs less than $2 to airfreight a phone from China to Europe. I’m already talking to telcos planning a strategy to win market share with low priced phones.
The Chinese decided in the spring of 2019 to massively “accelerate” the 5G deployment.
Under orders from Miao Wei at MIIT, the telcos moved their plans from 2020 to 2019. The government had decided to leave the U.S. in the dust and directed the three carriers to build at a furious, almost impossible rate.
[footnote reading the Chinese press in Google translation]
The Chinese telcos often do what the West considers impossible. In four years. they connected over 300 million apartments to FTTH. They are now over 400 million FTTH, more than twice the rest of the world.
By August 2019, it had become clear the Chinese were going to meet their 2019 targets. I wrote Almost all 5G estimates for 2019-2020 need to be doubled. Almost all estimates for 2019 were under 10 million subscribers. The result at the end of the year was 19 million, almost all in China and Korea.
The U.S. had declared a “race for 5G.” It was a brilliant lobbying move by the telcos’ lead in D.C. Meredith Baker, aiming to get a series of concessions from the government like lower municipal fees. She proved a worthy successor to Jim Cicconi of AT&T, the now-retired head of the Washington 2+2=5 squad.
The FCC had to give the telcos major concessions because they needed to install 400,000 small cells for 5G. They convinced the President that would have billions and trillions of benefits for the U.S. economy.
By the time the FCC went along, it was obvious the small cells wouldn’t be needed for years. The carriers had decided to mostly use mid and low-band spectrum rather than millimeter wave, so they needed far fewer small cells.
Crown Castle, which builds about half the small cells, a year and a half later told Wall Street they hadn’t increased the small cell build. Meredith did such a good job many in DC still believe the FCC gift to the telcos resulted in far more 5G.
The Chinese were already ahead of the US and the Europeans and decided to “accelerate” to embarass the Americans. Ironically, it was the U.S. plan the inspired the Chinese speedup. The telcos had to do some slick budget maneuvers to comply but of course they did. The CEOs would have been fired if they didn’t meet quotas. (That’s an improvement over the Mao days, when they might have been executed.
Huawei received about 40% of the orders for radios and phones, $10’s of billion in 2019 and 2020. It’s easy to see why the Chinese think, that a benefit.
December 2019, Xiaomi dropped the 5G phone price to a startling $285.
I decided that would drive demand and produced a 210 million estimate for 2020, again twice almost everyone else in the West
January and February were slow in both China and Korea. Then the pandemic hit. I reduced my estimate to 195-205 million.
But Xi Jinping and the Politburo decided that 5G would be the heart of the massive “New Infrastructure” stimulus. The telcos responded. They now have 300,000 base stations and are on track for 600,000 year end. 15 million 5G phones were sold in April, before the biggest price cuts hit. China Unicom has lowered the monthly fee from about US$18 to about US$13.
I want to see june and maybe july figures, but I expect I’ll be raising the low estimate to 190 million when they come in.
Note: “Low-band 5G” is currently slower than decent 4G. It will never be much faster than 4G. It is not included here. It will be large in the U.S. and now Germany. If you don’t believe me, ask any engineer building a network anywhere in the world. Software can’t yield much more speed in low-band without breaking the laws of physics.
Unfortunately, that means any analysis of 5G in the U.S. or Europe from now on will be somewhere between confusing and wildly misleading. Including mine, although I’ll do my best not to include low-band in my numbers.
Update June 14
MIIT has confirmed the Chinese telcos are exaggerating their number of 5G users. The real number is closer to 50 million rather than the 75 million claimed as “contracts” See No 5G Phone? China May Count You as 5G Anyway. I have not lowered my 2020 estimate of worldwide 5G subs. MIIT also raised planned 2020 5G actual users to 180 million. Telco CEOs get fired if they don’t meet quotas so they will find a way. The 30% price cut and low phone prices should do the trick.
5G phone prices are down to US$228 in China (Xiaomi Redmi 10X) and $266 (Huawei Honor X10.)
5G has few practical advantages over decent 4G, which runs at an average of 75 Mbps in Canada already. What fails at 75 Mbps and succeeds at 200 Mbps? Latency is typically 25-40 ms, not that much better than 4G 30-55 ms. After a year, not one major new use case is even close.
Applied Materials, the largest producer of chipmaking equipment, is considering moving production from Texas to Singapore to get around U.S. China sanctions.
Update May 24
Japan is finally moving, with NTT setting a goal of 2.5 million within a year. Japan: Soon Millions of 5G Users Rakuten’s software-based network isn’t working yet for 5G; when it does, expect rapid growth. KDDI made a point of charging the same price for 5G as for 4G. Almost no carrier is finding demand for 5G at higher prices, except maybe in Korea.
U.S. 3.4 million 5G phones sold in Q1. That’s a remarkable figure, because none of the U.S. 5G networks except Sprint were any good.
U.S. buyers of $1,000+ S20 phones are thinking ahead. 93% of 3.4 million 5G phones sold in the first three months were expensive Samsung S20’s, the first three bars in the illustration. Other phones were only 7%. T-Mobile is now offering an OnePlus and an LG for $700, so will not continue that dominant. Apple 5G won’t be out until late fall.
I’m guessing that Verizon pushed premium buyers to the Samsung 5G and they were willing to spend the extra. One reason my estimate for 5G in 2020 is probably the highest in the world is I believe that many will choose 5G phones as the price gap narrows.
Phones: In Europe, 5G phone prices are down to 400 euros. In China, many are under US$300 and falling. The difference between 4G and 5G phones may be as little as $50-75. 90% of Korea is covered with 5G. 50% of China will be covered yearend and most of the U.S. and probably Japan in 2021. In those countries, 5G phones are or soon will be the right choice for all but the poor. Most people keep phones 3-5 years.
The market isn’t large enough to support the first ten vendors but Meizu, Sony, and Lenovo are jumping in. Despite low prices, phone sales are continuing to fall even after reopening. China’s April sales were encouraging, however.
India: Bharti and Vodafone are using politics to hold off 5G, although Jio is ready to go and scale massively. Phone prices are already low enough for many Indians to choose 5G phones. Because of the 1.4 billion Indians, the 5G numbers in India will rapidly pass most other countries as deployment widens.
Europe is so far behind that no carrier is willing to release customer figures. Xavier Niel’s Eir has covered 25% of Ireland. No other company has reported even 10% coverage, except for the second-rate “low-band” 5G. Xavi is ready to go with 5G in France but Martin Bouygues is begging ARCEP to continue delaying.
Telefonica had been intending to use mmWave to grow fixed wireless in Germany. They aren’t going forward yet, possibly because the deal to share Vodafone’s cable network is cheaper.
Telefonica and Vodafone, challengers to the incumbents, had a natural business strategy of moving aggressively with mmWave and coverage. The technical people were strongly in favor. Management has refused, presumably because they fear the incumbent Empires will strike back. Both are also keeping capex down to keep the stock price and their bonuses up.
Monthly prices: KDDI and others are finding customers are resisting higher prices for 5G. Even proud Verizon is not collecting a premium, as T-Mobile continues to win customers. Korea’s government is pressuring the telcos to bring down prices.
5G capex and cost is about the same as 4G and capacity is much greater. In a competitive market, price per bit would be falling rapidly.
IoT: 5G was never going to be important for IoT because it has few improvements over 4G IoT. 5G can support a million IoT connections per cell but so far no applications have needed more than 4G can support. 90%+ of IoT will continue to be Wi-Fi and Bluetooth, including the 11 devices in our house.
Cellular IoT has an advantage where the receivers are widely dispersed (meter reading) or outside (cars.) Enrique Blanco, Telefonica CTO, points out that all current IoT requirements can be met with 4G and the NSA core. (He’s optimistic about demand for the SA core in a few years.)
Autonomous cars: Although extraordinarily exciting. autonomous cars will not be an important market for 5G for at least 5-10 years. All cars will be connected, for Waze-like apps and much more. The traffic could be substantial but not more than 4G can handle.
5G SA core: Suddenly, telcos are moving to the 5G SA core. It could be they think there’s a strong market for QoS/network slicing. National security agencies are interested, but few others seem close to orders. (Think real-time facial recognition on a crowd the size of New Year’s in Times Square.) More likely, the CTOs have become converts to the cloud-native, microservice designs possible in the 5G core. That significantly eases the transition to network functions virtualization. Rakuten’s early success with a software-defined, fully cloud-native system is convincing evidence the new software architecture can work.
Ericsson is delivering a US$400 million SA core to China Mobile, as are Huawei and ZTE. It’s expected by year-end.
“Fake 5G”: Ericsson and Nokia are demoing software to use small amounts of low-band 4G spectrum for 5G NR. The performance is nearly identical to 4G. Although the companies at 3GPP agreed to call anything with NR software “5G,” I consider it faking. Verizon, Swisscom, and now Deutsche Telekom are advertising 5G coverage in areas they can only deliver 4G speeds.
June 2020 H1 5G subscribers 86M (84M-92M)
For latest update on H1, http://analysisbranch.com/2020/07/24/june-2020-h1-5g-subscribers-83m-78m-88m/
For latest update on everything telecom, @analysisbranch
July 28 Added 4 million Fixed wireless routers and pucks, as Daryl Schoolar of Omdia made the sensible suggestion 5G routers should be included. Unfortunately, I can find no primary data. The 4 million is a guess. I have not tried to divide them by country. Data extremely welcome.
July 27 China is ~65 million, Korea 7.5 million, the U.S. 5-6 million, and the rest of the world perhaps 4 million. Only China and Korea provide official figures. On July 27, Ken Hyers of Strategy Analytics emailed me that Q2 5G phone sales in the US were ~2.7 million, down from 3.4 million in Q1. The total of 6.1 million phone set aq cap on the U.S. 5G, with a certain number in store inventory and transit. With firm figures for the three largest deployments, the uncertainty in the figures has become less.
I have to estimate/guess the numbers in the rest of the world. None of the European or southeast Asian carriers has released figures. I infer the number is quite low.
With decent phones selling for US$230 and the price going down, China will almost certainly meet the yearend 150 million plan. Verizon and others expect a big boost from the iPhone 5G, so I’m confident in my 210 million year-end figure. That estimate and 65 pages more of analysis is at http://analysisbranch.com/2020/06/19/5g-the-facts-and-the-future/
I believe it’s important to “show your work” That’s crucial in 5G in 2020 because the most widely reported figures for Q1 almost certainly are 40-50% too high. The Chinese carriers are reporting about 115 million “5G contracts” but only about 70 million 5G phones have shipped in China.
5G performance is highly disappointing. While Korea claims over 90% coverage, Open Signal only connected to 5G 1 in 6 tries. 5G #fail. 85% no 5G in “90% covered” Korea I believe, but haven’t confirmed, that the problem is that mid-band struggles getting indoors. Low-band speeds are often slower than 4G, especially at lower frequencies. See Finally, Data: US 5G slower than Canada’s 4G. Believe it
Latency is about the same as decent 4G. Verizon claims 30 ms. 1 ms latency is a fantasy outside the lab. Essentially none of the highly touted “use cases” and “new apps” have appreciable takeups.
5G is selling far above almost all predictions except mine because the phone price in China is little more than 4G. Decent phones go for US$230-260 in China, with prices falling there and everywhere else. Tens of millions of people have decided to pay the small premium for a phone that won’t be obsolete as soon. I would.
I have only indirect data on most of the world. If you want to be accurate, please think of the range of 84 million to 92 million rather than the 86 million headline figure. In the next few weeks, I expect the phone-makers to report counts, which will allow me to firm up these estimates.
An analyst firm put out a 63 million figure for Q1, almost certainly a mistake but frequently repeated. The highest plausible estimate of 5G phone sales in 2019 and the first quarter of 2020 is 45 million and it is probably a lower than that. (Strategy Analytics reports 24 million for Q1 2020)
I’ve urged them to put out a correction and am not naming them here.
“I make many mistakes,” the Butler said. I’m sure I have some, although I’ve done a great deal of research, firstname.lastname@example.org I’ll issue a correction ASAP.
China: ~65 million.
China’s telcos are reporting ~115 million “5G contracts” but an authoritative government source (CAICT) reports only 64 million 5G phones shipped. Since 4G and 5G contracts are the same price, I assume the telcos are persuading many 4G customers to sign up for a “5G contract.” China doesn’t need to overstate the numbers; even the lower figure is three times as many as the rest of the world.
Over 410,000 base stations have been upgraded and 15,000 more are being done each week. China Mobile expects a total of 600,000 5G cells yearend, covering about 700,000,000 people. All is mid-band.
17 million 5G phones shipped in June, many selling for US$230-260. 30-gigabyte service costs $13-18/month. China is on track to easily meet the 150 million year-end target. Counterpoint reports 60% of 5G phones in June were Huawei, which has shipped over 20 million 5G phones in China and probably over 30 million worldwide.
Unofficial sources claim July is far ahead of June.
Korea: ~7.5 million
Korea reported 6.9 million 5G subs at the end of May. The previous few months were between 500,000 and 600,000, so 7.5 million is a sensible figure for the end of June. All are mid-band, mostly 100-400 Mbps down. Open Signal data implies the indoor coverage is terrible. See 5G #fail. 85% no 5G in “90% covered” Korea
U.S.A. 5-6 million.
US 5G coverage is awful, so I was surprised when Strategy Analytics reported Samsung sold over 3 million expensive 5G phones in Q1. Most probably were sold by Verizon, despite Verizon customers only connecting to 5G 0.4% of the time. I infer that high-end Samsung buyers are spending more for a phone that will not be obsolete in a year or 2. CEO Hans Vestberg has said people are buying 5G phones even where Verizon does not have 5G coverage.
So far, almost all AT&T & T-Mobile has been the ridiculous “low-band 5G,” actually slower than much 4G. See Finally, Data: US 5G slower than Canada’s 4G. Believe it The companies are making it nearly impossible to separate the 5G at 4G speeds” from other 5G. I will exclude them if I can. Any reporter or analyst who doesn’t try to make the distinction should point out that much “5G” is slower than much “4G.”
Europe ?2 million
No European carrier has enough 5G customers to release a figure. I infer from that and the limited availability that there are few actual subscribers. More data welcome.
Gulf ? 1 million
The UAE, Qatar, and Saudi Arabia have some of the most extensive deployments of 5G. There is little or no public data on the number of subscribers.
Japan ? 0.3 million
3 carriers are deploying, with NTT DOCOMO shooting for 1 million early next year.
South and Southeast Asia ?0.3 million
Viettel and almost all the Southeast Asian countries are just starting to deploy. Jio in India is ready to move rapidly when the government approves. Look for very rapid growth in India in 2022 and possibly earlier. The projections of 18 million in 2024 are far too low.
Australia ?0.2 million
Lots of pr, little data
Latin America ?0.1 million
Almost all talk so far. I’m talking with a Brazilian with ambitious plans for a year or two from now.
Africa ? 0.1 million
MTN in South Africa has recently deployed mid-band, but few subscribers so far.
Canada ? 0.1 million
Just getting started
Russia, most of Latin America, and almost all of Africa have little more than pr.
For estimates of year-end 2020 and through 2025, http://analysisbranch.com/2020/06/19/5g-the-facts-and-the-future/
2020: Surprise free 210M, Low 175M, High 230M
Surprise-free 210 million 5G 2020 scenario
5G in 2020 is an Asian story, which will generate over 70% of connections. In 2019, Korea added 4.67 million connections, almost 10% of the country.. 93% of the population is covered. The goal for 2020 is 15 million. Apri: Korean subs fell from 800K in August 2019 to half that in January 2020. The companies combined to eliminate subsidies.
China plans 150 million 5G connections in 2020, directed by MIIT Minister Miao Wei with support from the highest levels. The three telcos know they must deliver. Until Corona, reaching the goal was highly likely. Corona will reduce the Q1 figures by 10-15 million.
Xi Jinping and the Politburo have declared 5G a crucial part of the $5 trillion stimulus. The 150 million connections are likely to be met. 550,000 new cells will be added to the existing 132,000. About 50% of the country will be covered.
No other country has provided estimates for 2020 5G. Many of the companies have revealed vague deployment plans, from which I infer a figure of 45 million 5G connections outside of China and Korea.. This is a soft number, one reason I offer a lower scenario below.
Apple will introduce the 5G iPhone in the fall. Verizon anticipates substantial demand and is rushing to have a network ready. More than half the premium phones in the West are iPhones. If many of them upgrade to 5G late in the year, the 45 million figure is reasonable. Apple expects a blockbuster year and has raised chip orders 50%. However, it is already seeing delays as the supply chain struggles. There are rumors that difficulties with the antenna may push deliveries back to the very end of the year.
Japan is the only country with a large population with the potential for rapid takeup of 5G. With the fourth carrier, Rakuten, now deploying, the other telcos have an incentive to grow quickly. In May, NTT set a goal of 2.5 million 5G subscribers in the first half of 2021.
India’s Reliance Jio, with 370,000,000 4G customers, is equipped to start selling 5G “90 days after the government approves.” Approval is now held up by politics and unlikely to be a large factor in 2020. When Jio upgrades, it could grow very rapidly.
- Much lower phone prices. The Xiaomi $285 Redmi 5G went on sale in January. This is killer. By May, Xiaomi and Huawei were offering decent phones at US$230 and discounting under $200. Xiaomi founder, Le Jun, explains, “Today’s China is an era of overcapacity.”
- Apple is raising chip orders as much as 50%, per Digitimes. The $399 iPhone SE, close to market, should be a very hot product. When iPhone 5G ships in the fall, it will probably inspire a surge in demand.
- 5G’s low cost per bit allows improved offerings. Better services, not far-future new apps, are the way for telcos to monetize 5G. Verizon sees 90% savings, DT 70%*. Once the network is built, the marginal cost is very low because of enormous capacity. That’s a big incentive to move customers to 5G and sell more to fill the network.
- Surprisingly low network costs allow better coverage sooner. Korea has 93% coverage with 190K radios. China in April had about 10% coverage with 132K radios. Europe is far behind. Deutsche Telekom and Telecom Italia expect only 20% coverage entering 2022. The extra cost to reach 60-80% in 2021 is 1-2% of sales per year. Few European carriers are willing to make the investment, which I think is a mistake. Telefonica Deutschland is raising capex almost 20%. If well-executed, it will pull ahead in the German market.
If 5G costs were high, carriers offering 5G would have to increase capital spending. The opposite is happening, although all major carriers are building 5G. The “very high cost of 5G” is a myth. Some, including Verizon, NTT, and Orange, actually are able to reduce capex.
Mid-band 5G, rather than millimeter wave, has reasonable costs and good reach. mmWave would require many new small cells and backhaul. No one except Verizon is building much mmWave in 2020 or 2021. Using existing cells is much cheaper than building new ones.
Radios have plummeted in price. Very large carriers are paying < $15,000 if analysts have it right. Indoor small cells are coming to market for $1,000 to $3,000. China plans to deploy millions.
Other savings include the use of wireless backhaul rather than fiber where fiber would be too expensive. Even Verizon is planning 20% wireless backhaul. mmWave will need fiber, but mid-band works fine with the 5 gigabit and faster microwave that’s fallen in cost. Inband backhaul, using the same 5G frequencies, looks promising and will save cost.
- Huge sales in China of 150M+ phones and 300,000 or more radios, give Huawei an important boost. That’s strategically important as Huawei is under attack by the U.S. Chinese demand has also pushed ZTE to a record high market value.
- The 5G hype is overwhelming in the West so far. Result: Many people are enthusiastic and natural buyers when phone prices come down. Verizon’s Hans Vestberg reports people buying 5G phones even where it doesn’t offer 5G. I expect that most mid-range & premium customers soon will buy 5G as coverage builds past about ?10% and Chinese prices spread worldwide. 4G phones will be obsolete by a year or two after that. A worry is that customers will discover 5G does very little that’s important today and the demand will fall, a key question.
- Possibly emerging competition. T-Mobile has 160 MHz of golden spectrum at 2500. It committed to cover 80% of the U.S. with 100 Mbps in four years to take over Sprint. It also committed to high capex levels. Its logical strategic move is to emulate Korea and get to 80% in 12-18 months. If so, the other carriers will have to speed up. New entrants Rakutan in Japan, 1&1 Drillisch in Germany, and China Broadcast could have a similar impact, although that’s more probable in 2021.
- Vendor financing, often at attractive terms, allows cash-constrained telcos to move forward. Finland is financing Nokia sales (AT&T,) Korea is financing Samsung ($billions for Reliance Jio,) and China both Huawei and ZTE. Almost any interest telco can call one of the vendors, say “build it,” and have the bill paid by vendor governments. Carlos Slim’s American Movil just announced 5G in carriers across Latin America. My guess is they will go slowly, but the Chinese government would almost certainly finance Movil in Mexico and Brazil if asked.
- National pride, economic, and trade goals inspire Korea and China. The two plan 165 million in 2020. That could spread to Japan and South Asia. Europe and the US mostly talk as we enter 2020. To meet government goals, Chinese telcos price 5G at $12-18 after discounts (30 gigabytes.) The Chinese see a major economic boost from 5G. That’s widely believed worldwide even if it is nonsense. The “studies” claiming a large economic boost are bogus; 4G can serve nearly all the applications discussed, as Telefonica CTO Enrique Blanco tells anyone who will listen.
Analysis that straightens the argument
- The 5G transition is a “once in a decade” opportunity to win market share. The quote is from LG Uplus, explaining to investors why a jump in marketing costs was correct. It added 4 points of market share despite retaliatory actions by the others. Xiaomi said almost exactly the same thing to justify the $285 phone price. Both companies have been especially aggressive.
- That’s the natural but unlikely move by Telefonica/O2 and Vodafone, among others. Early enthusiasm is mostly due to hype, but customers want 5G The much higher than expected response in Korea and China is convincing. Polls and market research consistently show at least half of buyers want 5G phones.
- 5G phone prices are coming down rapidly. At Xiaomi, the difference between 4G and 5G is $55. In 2020, the difference in manufacturing cost will be $15-35, driving price convergence. It’s becoming clear that 4G phones will be out-of-date and uncool in a year or two in many countries.
- 5G Coverage will expand at differing rates in different countries. But in two years, coverage will be wide enough 5G phones will often/usually find a signal. 4G phones will be perfectly functional for years, but who wants to take an obsolete phone out in front of friends?
- Qualcomm, Qorvo, and Skyworks have raised 5G 2020 estimates to over 200M. They supply essentially every major phone maker except Huawei and know what their customers intend to build.
The forecast would be more certain knowing:
- When/whether the “collaborative cartels” break Europe is held back primarily by cartel-like behavior. If competition were strong, at least one carrier in each market would be aggressively building. The others would follow. A smart regulator will make sure one carrier rapidly builds 5G. 1&1 Drillisch in Germany, Free in Italy, and 3UK need customers. Could Merkel persuade the European Investment Bank to help Drillish build 5G? Update Feb 16 Telefonica Deutschland raised capex ~20%, an interesting decision.
- When will the mid-band spectrum be released in all countries? Large chunks of spectrum between 2.5 GHz and 4.2 GHz are extremely efficient due to Massive MIMO. The ideal is 100 MHz per company. Lower frequencies need larger antennas for Massive MIMO, making it impractical below about 1800 MHz
- “Low-band 5G” is a joke and pr stunt, but mid-band typically delivers 100-300 megabits. mmWave delivers 3X that capacity but will be very rare for at least several years.
- Will carriers price aggressively because 5G is much less expensive per bit? Almost all are now charging the previous 4G price (or more) for 5G. making it very profitable when scale is reached. Verizon and BT want a premium but are meeting customer resistance. In a strongly competitive market, prices will fall towards the lower costs over time. Most countries have inadequate competition for the market to work its miracles.
Threats to the surprise-free estimate
Estimates from analysts like Dell’oro, Point-Topic, and Matt Davis generally are on target because they usually begin with several years of data and add key factors. Is the market approaching saturation? Will new technology or lower prices have a large impact? Have any of the companies announced plans?
When I first wrote this in December 2019, I working with six months of data from Korea and two months from China. The numbers are very uncertain. In my update in June 2020, the Chinese 150 million figure looks solid and perhaps low.
People may realize the 5G emperor has no clothes.
In reality, 5G has little practical advantage for most consumers. So far, most believe the hype. 4G 5 / 17 5G 2020 210 million projection is 75-125 Mbps in more and more places. Other than major downloads, how often do you need more than 50 Mbps on your phone? That’s twelve HD TV channels or three 4K TV channels.
All 5G chips come from two manufacturing sources
TSMC in Taiwan produces most chips. Samsung produces chips for Samsung and Oppo phones. Qualcomm will shift some production of the X60 to Samsung. Both are expanding capacity for the state-of-the-art 7nm and 5 nm foundries. They are buying $200M+ EUV machines as fast as ASML, the only manufacturer, can build them. Demand from Apple, Huawei, AMD, Qualcomm, and Nvidia is far more than expected. TSMC is confident it can add capacity fast enough. It won’t be easy. An earthquake in Taiwan could be devastating.
The US-China trade war
Huawei has won, with sales continuing to climb despite a ferocious attack by a superpower. It now makes phones and radios with no US component. Huawei’s $17B annual research budget means thousands of engineers are inventing substitutes. HiSilicon is state-of-the-art in 5G chips, AI, & network processors. It had almost no RF chips for the front end of phones a year ago. It now produces very sophisticated ones. The Google cutoff is reducing phone sales in Europe, the only major US weapon that is having an impact.
I wrote in an earlier draft “I don’t think the US could force TSMC to cut off all chips for Chinese phone makers, but it’s not impossible.” Since then it has made that threat and Taiwan was apparently ready to accede to American demands. It’s now unclear.
Something I haven’t thought of
Black Swans happen. Coruna has cut Q1 by 10-15 million. Many more if it isn’t soon contained. However, the impact could be reversed by China stimulus
In 20 years of reporting, I’ve never heard as many errors and lies. I’ve been calling b_______ on many for almost two years, but too many can’t hear their mistakes. Confirmation bias is everywhere. Everyone wants to believe 5G is wonderful, that it will transform economies, and will produce big profits for carriers and vendors. All are extremely unlikely.
Speeds are mostly 100-300 Mbps down, not much different than new 4G. (Except mmWave.)
Latencies are ~30 ms. , not much better than most 4G. Edge networks can bring either 4G or 5G latencies to 10-20 ms, but will be very limited outside Asia for years.
Nearly all new apps needing 5G will be small for years. Nearly everything will work fine on 4G. Autonomous cars are far away and won’t need 5G. Remote surgery
Low band 5G is slower than good 4G, at least for years. It doesn’t belong in 5G numbers. Speeds are typically 50-125 Mbps.
Claim: 5G networks are very expensive to build
Actual: 5G network deployments in the real world are not expensive compared to telco capex
Verizon and NTT DOCOMO are building extensive 5G while cutting capex. Orange and Telefonica expect the same. Nearly all 5G is mid-band, at about 4G cost to deploy. It has decent reach and requires few new towers. Deutsche Telekom and Sprint confirm that.
The high estimates are based on untrue assumptions except for mmWave, less than 5% of 5G deployments.
April: I now have data. $23,000 for 5G Base Stations, $56,000/cell Complete. Quantity 200,000 Prices go up rapidly in smaller quantities,
Claim: A telco is building rapidly when it targets less than 40%
Actual: 50% coverage is practical in two years across the developed world. Korea actually reached 93% in about 18 months. China almost certainly will reach 50% in about 18 months. The plans of 20% in two years from DT, Telecom Italia, and other Europeans are far less than the carriers could build.
Mid-band 5G – 95% of radios – is a gigabit
Actual: 5G networks generally deliver 100-300 Mbps. Gigabits are for the lab or mmWave. Above 400-500 Mbps will be the exception except for mmWave.
Today’s 5G networks have very low latency, such as 1 ms
Actual: Most 5G networks have ~30 ms latency, as announced by Verizon. 4G typically is 35-50 ms, but new equipment can be less than 30 ms. The major improvement, short TTI, is in both 4G and 5G today.
Edge networks can bring that down to 10-20 ms but are few. URLLC may reduce that by an additional 3-5 ms, but URLLC deployment is not imminent. Less than 10 ms will mostly be a lab phenomenon for years.
Autonomous cars require 5G
Actual: Thousands of autonomous cars are on the road today without a 4G connection. There will be millions in a few years. They have to be able to work when there is no signal.
As far as I know, none of the autonomous cars close to production will require 5G. The automakers don’t want to be dependent on the telcos – and don’t want to share the income.
All cars will be connected for entertainment and much more. But connected cars do not need extreme speeds or very low latency. They work fine with 4G.
Claim: 5G will grow remote surgery
Actual: 5G is almost irrelevant for remote surgery. Fiber is faster. Any facility that can afford a million-dollar remote surgery machine is likely to be connected by fiber. So are most surgeon’s hospitals and offices. Unless surgeons operate from the beach, 5G doesn’t matter.
Claim: Ultra Reliable Low Latency Communications (URLLC) delivers as named
Actual: Almost all security experts deprecate claims 5G URLLC will be ultrareliable. The 3GPP standards make that claim because some known problems have been fixed. New ones are already developing. The general expectation is that 5G networks will be more vulnerable than 4G because they have so many points of attack. Notoriously insecure IoT will be a major part of the system.
Claim: 5G will have a major effect on jobs or the economy
Actual: Use common sense, please. Very few applications need better than the 50 Mbps down and 40 ms latency common in 4G. Most of the capacity of 5G will be dedicated to video and games. Does watching more TV really improve the economy? All the projections start with unsound assumptions. 5G is not needed for autonomous cars and trucks. Nearly all the high-impact use cases work well with 4G. I’ve looked at the primary sources, such as Qualcomm’s IHS study. It claims “In 2035, 5G will enable $13.2 trillion of global economic output.” A strong majority of the applications cited work fine in 4G.
Even if there were natural uses for 5G, it’s absurd to estimate 15 years ahead on something that has no track record. Unsupportable assumptions from Qualcomm imply that 5G IoT uses less power than 4G IoT and that 5G has better reach than 4G. Neither is true.
Claim: It’s impractical to deploy faster than we are, according to many telcos
Actual: There are no technical or economic reasons 5G couldn’t cover 80% of the developed world in 2 or 3 years. Korea reached over 93% in 18 months. China is on track to reach 80% in 30 months. The engineers in Germany are just as capable as the Koreans. Because Germany and the UK are less dense than Korea. 93% in 18 months may not be practical but 80% surely is. BT is actively training technicians for fiber installs and could do so if tower workers were in short supply. In the US, service companies are clamoring for more work. Verizon & AT&T have cut capex. Simply holding capex steady would allow much faster 5G builds. Deutsche Telekom profits are up significantly but the money is not being spent on the network.
Company problems, limited competition, and timid regulators are leaving Europe far behind. Europe is dominated by de facto cartels. “If we build more 5G, so will our competitors. The result: we’re both spending more money but stay in the same position.” That attitude is doing more to hold back 5G in Europe than costs, regulators, or spectrum.
It’s likely a mistaken analysis. Networks take a long time to plan and build. You can’t be a “fast follower.” The first mover can expect an important gain in market share and reputation, as demonstrated in Korea. Vodafone was explicit. “We will match our competitors in Germany.” (Quote from memory.)
No one had to meet in obscure airport motels. Collaboration is established by open statements, typically on financial calls. “Rational” behavior, companies like this agree, is for all to limit investment. Nearly everyone already has a smartphone, so better networks will not attract more customers. The same income with lower capex = higher profits.
The tech people at major carriers agree with me. Those at Verizon and AT&T argued strenuously against the investment cutbacks, but only in private. I’m hearing similar from top technical people in Europe.
Claim: Releasing more mmWave spectrum will lead to a faster buildout
Actual: Except for Verizon in limited areas, carriers who have high band spectrum do not expect to use it extensively for years except in very limited locations. Few carriers will build out that spectrum until they have reached near-capacity in mid-band, which requires fewer sites and is much cheaper.
Claim: Reducing the barriers and cost for adding small cells will lead to much faster rollout of 5G.
Actual: America did just that but the actual small cell build is slightly less than predicted before the new FCC rules. Almost all small cells are going to areas where the capacity is needed, typically business districts. The new cells are needed for 4G as well.
Making it easier to deploy small cells saves carriers money, but they would have found a way to build most of the cells because they are necessary.
The Low Scenario: Perhaps Only 175 Million in 2020
June, 2020, I raised the low estimate to 175 million as China is clearly going to reach 150 million. I think it will go up again when I have full Q2 figures.
GSMA, one of the very best analyst groups, still has a 2020 estimate much lower than mine. So do several others. Its Global Mobile Trends 2020 is very thoughtful and well worth a download. When people I respect have a different opinion, I go back and look closely. I’ve outlined here what’s behind my numbers.
China’s plan to sell 150 million 5G phones in 2020 has strong government backing. The country is unlikely to fall very short. Korea expects 10 million more and 93% of the country already is covered by 5G. The U.S., Australia, Canada, and Japan expect growth.
MIIT, which controls the three telcos, declared in summer 2019 that China would add 150 million 5G users in 2020. The country is highly unlikely to fall well beneath the goal. Authoritarian governments are not my choice, but in China it has proven effective.
The Chinese carriers nearly always find ways to deliver what the government demands. MIIT can fire managers who are unsatisfactory and often does. The budget and conditions are in place to meet that figure. The typical technical publication produces several articles each week about how great 5G is, knowing the state wants them to create demand. The general press, like People’s Daily, often adds to the hoopla.
More pragmatically, the price of 5G phones is so close to the price of 4G phones, it’s the sensible choice of many consumers. Xiaomi charges $285 for a decent 5G phone. Chinese buy about
400 million smartphones a year. June update: 350 million after Corona as prices drop to US$228. Given the price drop, it would not be surprising if 40% of the phones sold in 2020 were 5G.
What surprises could drive 2020 sales lower?
Corona is reducing 5G sales in Q1 by 10-15 million.
China expects to bring it under control fairly rapidly and restore production with a huge stimulus. The consensus of experts is that fairly rapid control is likely but none are certain. If Corona lingers, not just China will be affected. The world is likely to go into recession and fewer people buy new phones. April: The world has gone into recession and people are buying fewer phones. If China hadn’t renewed its 150 million plan, I’d have to lower estimates considerably. June update: China is now clearly on track.
There are other plausible factors that could hold back 5G:
If people realize that 5G has few if any practical advantages, they could hold off buying.
If TSMC can’t raise production sufficiently, that would be a major holdback. All 5G chips except Samsung and Oppo are produced at TSMC, which is already asking six month’s lead time on new orders. TSMC & Samsung are adding capacity as rapidly as ASML can deliver $200 million EUV machines. Both are optimistic they can meet the very high demand from Apple, AMD, Nvidia, and many others. An earthquake in Taiwan, a defect in chemicals, or anything else that holds back production could have a large impact.
Almost all US and European 5G builds are so limited they are essentially pr for now. While many of the top European tech people agree with me accelerated builds are a smart move, their CEOs still are holding back. My guess is the Europeans will respond when they realize the demand, but none are moving yet. Currently, they are counting on a gentleman’s agreement not to build.
Apple’s 5G iPhone is expected to create a surge in demand in the fall. If it’s late, many purchases will be delayed into 2021. It’s rumored to be held up as Apple tries to shrink the antennas and keep the phone thin.
and surely other factors I haven’t anticipated.
The high scenario: 230 million in 2020
The industry is ready to produce enough radios, switches, routers, optics, and for far more than 210 million users in 2020. The telcos remain extremely profitable even if growth is slow. A modest increase in capex – 2 or 3% of sales – could bring most developed countries to 80% coverage quickly. April: Reduced from 265 million to 230 million on global economic problems.
Consumers and buyers tend to move slowly at the beginning unless acted on by an external force. That’s happened in Korea and China. The government decided to make it so. They are currently over 90% of 5G subscriptions and will remain over 70% in 2020. In Korea, the telcos profits are now higher.
If governments or dedicated carriers chose to move hard, things will be very different. But most regulators, including all in Europe, are inclined to defer to the companies. Most carriers would rather give more to shareholders rather than invest.
The likely result will be slow growth in most of Europe.
The price of phones comes down worldwide and drives sales
One reason demand is exploding in China but slow-moving in the West is that 5G phones in China cost about half the price. Decent 5G phones cost US$285. China Mobile expects the price to come down below $200 by the end of the year. June update: Prices are down to US$228 with sales under $200.
Those low prices for most 5G phones make sense because the parts for a 5G phone in 2020 will only cost $15-35 more than 4G in 2019. Nine phone makers are fighting for a market where only 5 or 5 can make a profit. Apple, Google, and Lenovo/Motorola will join them. In China, prices down far more rapidly than anyone expected.
It only costs $1 to airfreight a phone. Sprint is already promising phones from $300-$500, although T-Mobile will probably be more cautious.
Any large telco can make three phone calls and get bids 60-70% below most European and US current phone prices. If Free in France or Italy makes those calls, the other telcos will have little choice but to follow. Xavi – please.
I’ve already had a query about whether you could make money by importing phones to Europe. It’s probably a good business, although small incompatibilities could be a problem.
5G is a once in a decade opportunity for companies to win.
Europe might go beyond pr
The massive hype around 5G in the West is almost totally hot air, incompetent regulators, and companies unwilling to give customers what other countries offer.
Korea covered half the country by the middle of 2019 and 93% at the end of the year. Germany and England are less dense, so 80-85% in two years is probably comparable. Any honorable regulator will demand that telcos across the developed world at least cover 80% in two years.
I do not know any Western companies planning more than 10% at the end of 2019. Germany and Italy are talking only 20% in 2021. If 5G is important, that’s ridiculous. April: Xavier Niel’s Free has covered more than a quarter of Ireland.
Alternately, an aggressive challenger could force the incumbents to respond. I’ve calculated that a 10-15% capex increase at Vodafone or Telefonica would pull them well in front and win customers. Because it would take a while for others to catch up, the added customers almost certainly will more than make up the cost.
It’s the right business move for companies like those, although the CFOs don’t see it yet.
Apple prices 5G aggressively and heavily promotes it
Tim Cook will sell the iPhone SE 2 for $399 according to press reports. It has also been cutting prices in China and India. Steve Jobs never did that, but it’s a sensible move for Apple to also offer phones below super-premium prices. Apple is making tens of billions on services. The 30% cut of game revenues is enormous. The more iPhones, the more service income.
Apple is a year behind on 5G. It could sell a 5G phone based on the SE 2 at $499 and actually have a higher profit margin over the 4G version.
I’m betting Tim won’t do it in 2020, but it’s certainly possible.
T-Mobile U.S. builds fast on the 2.5 GHz golden spectrum
Once the Germans say “Go!,” Neville Ray can and should move at extraordinary speed. That was Sprint’s plan; it still has more 5G coverage than anyone else in the U.S. despite shutting down upgrades months ago pending the merger. April: The build is on and Verizon raised capex $billion to keep up.
T-Mobile promised to invest as much as both companies combined for three years. It also promised the FCC to offer 5G to over 90% in three years and 100 megabits to over 80% in three years. The 160 MHz of mid-band spectrum is among the best in the world, with the lowest 5G cost to deploy.
AT&T and Verizon would have to respond by speeding up deployment. Currently, they are reducing capex, not good for the quality of networks in the U.S.
Neville Ray and John Saw have plans ready to go. They may use most of 2019 to merge the companies, however.
When T-Mobile does go, just watch.
Japan comes in strong
The Japanese have turned on 5G. NTT expects 2.5 million 5G users in the first half of 2021. My surprise-free estimate has only a modest contribution from Japan. Will they go slow? Rakuten, the new 4th telco, has promised to change the market.
Seizo Onoe, the CTO at NTT DOCOMO, and Tariq Amin, CTO at new fourth carrier Rakuten, are among the most respected in the business. Given the capex, they could match what the Koreans have done, 50-80% coverage the first year.
The public wants to buy 5G; priced right, 10% of Japanese, ~12 million, might sign on quickly. The Korean telcos had to subsidize the early, expensive phones to get a high take rate. The Japanese can offer reasonably priced phones from the beginning because phone prices have fallen so much.
At Rakuten, Tariq Amin is building one of the most advanced networks in the world, completely software-defined and cloud native. The network has gone live with 4G and is designed to launch 5G soon. Rakuten intends to follow the Jio model: Efficiency for very low costs that allow low prices to win customers. The 370 million 4G customers Jio has won in 4 years are more than the population of the US. They will change the Internet.
The surprise-free assumption is that 2020 5G in Japan will be limited, but it’s easy to envisage a scenario of much higher growth.
China “accelerates” even more
Minister Miao Wei believes 5G will accelerate the economy and in the spring told the carriers to “accelerate.” The results are remarkable; over 100 million people can get 5G today. The plan for 2020 is 150 million subscribers and over 600 million servable. That is realistic given that the Chinese buy ~400 million phones a year. If the stimulus after Corona focuses on 5G, it could go even higher. April: More than 150 million looks unlikely.
“Acceleration” is important for political reasons as well. Huawei sells close to half of the phones in China. The increased sales of radios and phones in China are replacing the sales lost in Europe because of the US boycott.
To further support Huawei and for other reasons, the government could tell the telcos to sell even more 5G than the current plan. That might require device subsidies and other promotion, but not beyond what the companies can afford.
Other surprises are possible
The technology and cost factors could make 5G in Africa and Latin America profitable. The companies now are planning very little in 2020, but national pride or competitive demands could change that.
“Unknown unknowns” and black swans are always possible, especially in something growing as fast as 5G.
Low-band is not included in my 5G figures.
“5G low-band” runs at about the speed of 4G. I think it’s a scam to pretend it’s something special. It’s just 4G with some software, “5G NR.” In low bands, the software does almost nothing.
Don’t believe the hype.
Further data from our reporting
The January price of $285 startled everyone. $300 was expected by the summer but Xiaomi decided to cut prices 6 months earlier.
Xiaomi Redmi 5G: $285. The explosion is here. (From December)
Xiaomi’s first 2020 phone is US$200 cheaper than anything else in the world, half the price of many, and less than a third of the prices in the West. Huawei and China Mobile expected the price to fall under $300 in the summer of 2020. Shipping in January, Xiaomi moved 6 months before anyone anticipated.
Lu Weibing of Xiaomi is seizing “a historic opportunity” to win market share for the next decade. Huawei has jumped to over 40% market share in China. The low price is a weapon against the Huawei Honor brand, Lu is
“Very confident that K30 will completely crush the V30 of friends. This product must be the turning point between the two brands. We will form a crushing trend.”
The Redmi K30 5G uses the new Qualcomm Snapdragon 765G, priced lower than the 865G in prior Qualcomm 5G phones. While chip prices are confidential, Qualcomm has reduced the price differential between the 765G and 4G chips. The 4G version only $50 less than the 5G.
Looking forward from 2020 to 2025
I have no experience writing fiction, but 5G estimates even a year out are speculative. Four and five year forecasts are guaranteed inaccurate because 5G is new and changing rapidly.
Yet people building networks need to think years ahead. Reporters love the headlines. The year I spent researching 5G gives me some insight. Originally, I wanted to leave anything beyond 2 or 3 years to my brother, a Hugo-nominated science fiction writer. But inquiring minds want to know.
The surprise-free, most likely estimates are:
- 2020 195-205,000,000 revised in April from 210 million
- 2021 500,000,000 (The key to this figure is an assumption half of China’s 400 million phones will be 5G. Some 5G phones will be priced under $200.)
- 2022 950,000,000 (300M China. Western Europe, Japan, South Asia and the U.S. meaningful.)
- 2023 1,550,000,000 (India and others should become significant. China slows down as saturation is in sight.)
- 2024 2,250,000,000 (4G & 5G phones will be similar in cost so probably half the phones will be 5G)
- 2025 3,050,000,000
Low estimates would be
- 2020 175,000,000 Revised from 165 million in April as China is deeply committed to 150 million and T-Mobile is driving the U.S.
- 2021 410,000,000 (Europe and the U.S. hold back. In China as well, customers are disappointed with 5G performance, which has little practical value. European carriers wink and nod and collaborate on holding back investment.)
- 2022 800,000,000
- 2023 1,300,000,000
- 2024 1,900,000,000 (Closer to established estimates such as Ericsson.)
- 2025 2,600,000,000
High estimates would be
- 2020 230,000,000 Revised April from 265 million on low phone sales and economic struggles. (Xiaomi is selling a decent phone for $285 in China. People choose 5G phons because they cost little more and won’t be obsolete as soon. iPhone 5G should be a blockbuster, late 2020 and into 2021)
- 2021 650,000,000 (Phone prices will be down to $200 5G becomes the obvious choice. U.S., Japan, and Canada deploy large networks. Europeans go beyond pr stunts and build networks.)
- 2022 1,200,000,000 (India should come in. Reliance Jio already has equipment in place but politics is holding things up.)
- 2023 2,100,000,000 (Brazil, Indonesia, Thailand, South Africa, & Vietnam contribute.)
- 2024 3,000,000,000 (5G phones will be under $100 and the obvious choice for most.)
- 2025 4,000,000,000 (Africa’s billion people embrace 5G.)
My key assumptions
- 5G phones are already down to $285 (Xiaomi) and Oppo predicts $150 later in 2020. The gap between 4G & 5G prices is narrowing, making 5G the obvious choice for most in any country with extensive coverage.
- Customers want to buy 5G. The Korean and Chinese results are convincing. Surveys elsewhere correspond. At some point, many people will see through the 5G hype and realize there is little practical advantage. My assumption is that will only have a modest effect on sales. April update: Korean sales fell after the carriers agreed to end subsidies.
- 5G networks have enormous capacity, able to deliver far more than most carriers can expect to sell. Once the 5G networks are in place, carriers will have a strong incentive to promote 5G.
- 5G networks are not expensive to build by telco standards.1 They often cost less in total than 4G and cost much less per bit. NTT, Verizon, AT&T, and Orange are building 5G nets and cutting capex. Countries like Germany and Italy could very rapidly reach 70% or better coverage by pulling up 1-2% of sales for added capex. April: I now have data. See $23,000 for 5G Base Stations, $56,000/cell Complete. Quantity 200,000
- China is committed to ~50% coverage by the end of 2020, confirmed by Xi Jinping despite Corona. Even with problems, most of China will be covered by 2021-2022. (China will be 70% of 5G adds in 2019, 2020, & possibly 2021.) See Xi Jinping: Accelerate 5G to Restore the Economy.
- 1/2 or more of the ~400 million phones Chinese buy in 2021 will be 5G. In 2022, it is reasonable to estimate 3/4ths of the phones sold in China will be 5G. April: Chinese phone sales were way down in Q1 2020. May: Chinese sales in April rebounded.
- The iPhone 5G will kickstart sales in the U.S. and Europe, starting in late 2020 or early 2021.
- India will be a major factor starting in 2022-2023. 400 million Indians went 4G in the last 4 years. 5G will be slow getting started but once it’s built, the carriers will promote it strongly to ease congestion in lower frequencies.
- By 2023 & 2024, 5G costs will be low enough operators will deploy fairly widely in countries like Indonesia, Pakistan, and Nigeria. 5G capacity, especially from the use of 3300-4200 MHz spectrum, will be crucial to delivering a decent Internet experience. A respected source estimates only 3% of Africans will use 5G in 2025. My estimates are much higher.
- 5G isn’t close to delivering the hype. It usually is little better than decent 4G. If buyers realize that, sales could slow.
- There are no new “use cases” likely to reach volume in the next two or three years and possibly longer. 5G will allow carriers to support much more of what we do today, including better video. Will that be enough to drive sales?
- European carriers are approaching 5G as a cartel would. They are going much slower than makes sense based on the technology and economics. Korea is at 93% coverage. China will be ~50% by the end of 2020. But Telecom Italia and Deutsche Telekom plan only 20% entering 2022. I do not believe anyone is meeting in secret. Defacto agreement can be achieved by public signaling. May update: DT is joining the Americans in selling “5G” in low bands running at 4G speeds. This will pump up figures, The right thing to do is not to include the second-rate “5G” but it will be hard to do that.
- Prices of phones in Europe are twice the price in China. That should change in a few months, but again a cartel would keep prices high. April: $400-500 phones are now reaching Europe.
- Most carriers in the West are holding down capex in favor of higher dividends and stock buybacks. The stock price is more important than profits in future years.
- Apple is having some problems and the iPhone 5G may be very limited in 2020. That would reduce sales in 2020 although 2021 should recover them. April: Latest rumors have Apple back on track.
Why these estimates are higher than most
Analysts I respect have significantly lower estimates than I do. I could be wrong, of course. The data on 5G is extremely limited. I believe I’m giving more weight to:
- The declining price of phones. A decent phone (Xiaomi) can be bought today in China for $285. Carriers and vendors expect the price to fall to $200 and then to below $150. Except for the very cheapest tiers, more and more people will choose 5G as the cost difference continues to decline.
I’ve looked closely at the components of a 5G phone that could raise the price. The chip is more complicated, but Moore’s Law will continue to bring down the cost. It’s slowing down but isn’t dead yet. The RF (Radio Frequency) parts including filters and amplifiers are much more complex than 4G but new methods and materials are coming very quickly.
- I believe China will meet its plan for 150 million 5G connections in 2020. Xi Jinping has listed 5G as one of the key priorities of the stimulus coming. The CEOs of China Mobile & China Unicom have pledged to install over 550,000 radios in 2020. They will find a way, if only to keep their jobs.
In the past, the Chinese carriers have delivered what the West thought impossible, including 300 million FTTH connections in less than 4 years. That’s homes connected, not just homes passed.
- India added 400 million 4G LTE connections in the last four years, an astounding story that is changing the Internet. That was driven by Mukesh Ambani’s Reliance Jio, which is ready to lead into 5G. Much of the equipment is in place, including fiber backhaul. It has promised to start selling 5G within 90 days of getting a license, which is currently tied up in politics.
Jio is well-funded, profitable, and extremely capable. It may not reach volume until 2022 or 2023, but will eventually deliver hundreds of millions of 5G connections.
- I have a great deal of respect for African companies such as Liquid Telecom and Safaricom. They have the skills and the financing to accomplish great things in 5G. There are very few landlines in Africa; the quality of the Internet is held back by the limits of wireless. May: Little has happened but talk is strong. One opinion is that Africa will move on 4G Massive MIMO to hold down costs while upping capacity.
These numbers do not include IoT. I believe that’s industry-standard practice and sensible. 5G IoT has one important difference from 4G NB-IoT: it can handle a million units in a single cell. As I write in June, 2020, no real-world examples seem to need more connections than 4G supports.
Low-band is slower today than 4G and will never be much faster. I would prefer to exclude that but the companies are refusing to break out the data.
If you see a mistake or have new data, please email me email@example.com. Thank you.
The key factors
Most 5G phones will be sold in China and possibly the U.S. for the first few years. Korea and possibly Japan will also deploy rapidly, but their populations are much lower.
- Number and price of phones sold
Chinese smartphone sales have been ~400 million per year. 25% of the sales in January were 5G. Prices in China are as low as $285 and falling. June update: Prices down to US$228.
It’s reasonable to expect 50% and then 75% of the phones in China to be 5G fairly soon. Indians buy less than half as many smartphones as the Chinese. India does have a substantial middle class, but 5G volume in India will be limited until prices come down to $100-200.
10% of Koreans bought 5G phones in the first year. The carriers expect 30% (~15 million) to choose 5G by the end of 2020. May: Well over 200,000 cells have been upgraded and the carriers are likely to meet the new target of 500,000. Korea subscriptions are disappointing, however.
European coverage is often less than 5% and most carriers are moving very slowly. Of course demand is low.
Strong competition is usually the best way to get a large corporation to change plans. When the CEO fears catastrophe, they move. Verizon built the first large fiber network in the world because cable modems were killing them. CEO Ivan Seidenberg said, “W have to get the cable modems out of the house.” He was scared.
Already in 5G we are seeing the power of fear. LG Uplus was the first Korean carrier to aggressively promote 5G. It rapidly gained market share. SK & KT decided they had no choice but to speed up and spend more. May update: The SK & KT empires struck back, matching the LG offers. After a few months, the carriers found a way to all pull back on subsidies and monthly customer adds have been down since December.
In China, the carriers are making sure to keep up with each other. In the U.S., AT&T thought it had to match Verizon before the “low-band 5G” meme confused everything. Sprint’s 160 MHz of golden spectrum at 2500 MHz is about to destabilize things. Verizon is scrambling to keep up. May update: Verizon has added $billion to capex.
I have a thought experiment below: What would happen if Telefonica Deutschland jumped ahead with a fast deployment? The cost would be reasonable, perhaps 2% of annual sales.
With textbook competition, that would be the right move and highly profitable. In the real world of German telecom, Telefonica has to think about what would happen if The Empire Strikes Back. (DT) For now, TD feels safer not angering the giant and is holding back. Germany looks to be years behind Korea and Japan.
Which is not good for the country. Does the government have the power to do anything about the cartel-like behavior? May update: At Vodafone and Telefonica, senior technical people want to accelerate. Top management refuses to spend.
The competition in 5G phones is working wonders. No one expected phones below $300 for another 6 months. But 10 companies are chasing a market where only four or five can make a profit. Xiaomi decided to forward price at $285 for a decent phone. My research found that a 5G phone soon will only cost $15-35 more to manufacture. Competition is working well. May update: The U.S. big 3 sold over 3 million Samsung S20 5G phones in the first quarter. Buyers of phones over $1,000 are choosing 5G.
- Business plans of the companies
Early in 2019, it may have been sensible to go slowly on 5G until we had more data from the field. By the middle of the year, it became obvious 5G mid-band was working. Reach is good and deployment costs are reasonable. Demand looked strong. When the price of phones came down late in 2019, investing more should have been the right move – unless you were scared of retaliation.
Last summer, two CTOs I respect told me they agreed but their bosses weren’t ready to go along. 5G cost per bit is so much lower it is the right choice from the technical side. Most of the European carriers are falling years behind the Asians. Verizon is risking its future by cutting capex. Why? Under conventional supply and demand economics, most would be building faster.
The business side sees it differently. A colleague close to telco strategists pointed out to me what your competitor will do often is more important than any technical considerations. Will the Empire Strike Back? With only 3 or 4 companies, that’s highly likely. The common result is cartel-like behavior without any illegal meetings taking place.
In addition, the stock price is top of mind for most CEOs. The price is highly related to buybacks, dividends, and cash flow. To protect the cash flow, many companies are freezing and cutting capex when despite a likely high cost in future years.
Deutsche Telekom, Verizon, and AT&T are clearly making short-term decisions in order to boost the stock price. AT&T CEO Randall Stephenson is leaving the end of the year. A high stock price is worth tens of millions to him. Short term decisions make sense and Randall has cut capex by $2 billion. Vestberg & Höttges are also scared of a price drop if they invested any more.
- Government promotion
Korea and China are far ahead in 5G because the government thinks it important. They told the companies to build and they did. In China, Minister Miao Wei said “Accelerate!” in mid-2019. The plan had been to hold off until 2020 but instead the telcos upgraded 132,000 sites by year-end 2019. That’s probably twice as many as the entire Western world.
Governments in the West are also enthusiastic but almost nothing they have done is making much difference. (Except making mid-band spectrum available.) The U.K. required duct sharing. The U.S. reduced the fees telcos pay for rights of way and pole attachments by hundreds of millions or more. D.C. is still claiming that resulted in a faster small cell build but actually the U.S. is adding fewer small cells than had been planned.
In D.C., the telcos spend hundreds of millions of dollars on influence every year. Comcast’s lead, David Cohen, made $18 million a year. That’s literally 100 times as much as the best public advocates make. The going offer for a connected lobbyist is at least $1 million. For that kind of money, they hire some of the best influencers in the world. I call them the 2+2=5 gang, for what they can persuade regulators to believe.
Incentives and nudges rarely work.
Review by countries
Some countries are growing ten times as fast as others of similar size and income. You can’t get a meaningful answer without going country by country, at least for the larger countries. For several years, Korea’s 5G penetration will be more than twice Japan’s despite a lower per capita income.
[There’s a lot to update here, some of which is above.]
High Population Countries
More than half of 5G users through 2022 will be in China and the U.S. India should become important in 2022 or 2023. China and India have populations of ~1.4 billion. Add the 331 million people of the U.S. and it’s easy to see the starting point of any forecast. These three countries are about 40% of the world population,
China is moving at breakneck speed and dominates the figures for 2019 & 2020. That is likely to continue for 2021. China Daily estimates 1.24 billion Internet users2 which is not adjusted for people with more than 1 sim card. China also has more than 400 million fixed broadband users, almost all on FTTH. ~400 million phones are sold yearly. How many will be 5G? 1/4th of phones sold in January 2020 were 5G. The government called the telcos to a conference and reiterated the “guidance” of 150 million 5G connections by the end of 2020. Over half the phones sold each month will be 5G by late 2020 or early 2021. A reasonable estimate is 150 million 2020, 200 million more in 2021, and continued growth after that.
At the end of May, China Mobile reported 55 million 5G “packages” sold and China Telecom 35 million. Unicom hasn’t released and figures and would take packages over 100 million. The number of 5G phones sold is considerably lower, a more accurate measure of 5G.
China already is doing early deployments of the standalone core, the real 5G core. It will be one to three years ahead of the rest of the world. China Mobile has given US$400 million contracts to Huawei, ZTE, and Ericsson for SA cores, with deployments in 2020.
China will rapidly connect over a billion to Edge networks, dwarfing any other Edge deployment. 2020 has been designated a year for testing, but gear is already being installed and a rapid ramp expected in 2021.
Minister Miao Wei in 2017 had set a target of 90% of China being within 25 ms of an Edge server in 2025. The target now is 10-15 ms and coverage well before 2025.
China Broadcast, which offers cable service as well, has the fourth 5G license and very desirable spectrum at 700 MHz. It is just testing now but expects to start signing customers before the end of the year. State Grid, the massive electric company, is working with China Broadcast.
The full details aren’t public, but some form of network sharing is planned between China Broadcast and China Mobile. China Telecom and China UNICOM are doing a joint build and sharing 200 MHz of mis-band spectrum, the largest allocation in the world.
Few outside India believed GSMA’s estimate of 88 million 5G customers in 2025. I think Ambani will grow 5G in India much faster than that.
India has less spectrum per capita and relatively few towers. The carriers will need to put 3300-4200 MHz spectrum to use, probably using 5G rather than 4G. Reliance Jio, which has grown to 370 million 4G customers in less than four years. It is ready to turn on 5G within 90 days of getting government approval.
India has a substantial middle class that can afford $200 5G phones. Except for the political issue, it would be easy to expect 20 million in 2022 and rapid rise from there. It could even go faster if Jio wanted to move customers to the new spectrum.
But I had no idea what Jio was planning.
Bharti and Vodafone are doing everything possible to delay 5G because they have very limited funds for investment. Vodafone has said it will go broke without a government bailout. I can’t predict how long politics will delay things.
158 million smartphones sold in India in 2019. 3 Most were 4G. As soon as a substantial fraction of India’s phone sales become 5G, it will affect the global totals.
3.4 million 5G Samsung Galaxy phones sold in Q1. That’s far more than I expected. Verizon expects explosive sales when the iPhone 5G ships. 30 million U.S. 5G phones in 2020 are possible, but I don’t have enough data.
DT CEO Timotheus Höttges just said he wants to be #1 in the U.S. If he gives Neville Ray the budget to build and Mike Sievert the marketing budget, his chances are good.
Verizon and AT&T in most of the country will be using low band spectrum. Open Signal found actual speeds of 62 Mbps, slower than most of AT&T’s own upgraded 4G cells. (Chart below.) That can quickly be deployed using dynamic spectrum sharing, which is reaching the field.
The 2.5 GHz spectrum has better reach than higher frequencies. Most of the country can be covered from existing towers. That’s fast and relatively cheap. TMO plans a major sales push in the late summer, hoping to get in front of Verizon & the iPhone 5G.
It’s ridiculous to call connections slower than 4G “5G.” But the marketing people in 2018 persuaded the world that adding “5G NR” software to 4G networks make them “5G” and wonderful.
NR software adds almost nothing to the performance below 2 GHz in low band spectrum. (That’s why everyone wants mid-band, 3.3-4.2 GHz. That will creep into the market over the next three years.)
So far, the regulators and most reporters have been bamboozled. It’s not clear whether users will care about the difference between 50 Mbps and 200 Mbps. It has little practical effect.
Verizon’s mmWave, the top line, is far faster than anything else. But it will be in a limited area, perhaps 10-15% of the country in 2020 and ~30% in two or three years. (Verizon refuses to provide details.)
T-Mobile has committed to 80% & 90% coverage as part of the merger deal. It also pledged to keep capex at the level of the combined companies, more than enough for a rapid deployment. It will have a much better network than Verizon or AT&T. That could inspire them to build faster and offer great promotions.
The politicians are doing an incredible job fooling themselves. First, they’ve told each other that 5G could transform the economy. The telcos D.C. front, the CTIA trade association, invented a fake “Race to 5G.” They refuse to believe their eyes when they read the sworn testimony of T-Mobile CTO Neville Ray that 5G at best would be 20%-50% faster.4 In practice, 4G is faster because 5G carrier aggregation and LAA are not yet developed.
Population by country, in millions (UN figures via Wikipedia)
Countries from 97 million to 300 million people
The next 12 nations are about 20% of the world population, Japan is the only country in this group with more than very modest plans for 2020. Russia may or may not build rapidly in 2021. Vietnam is included in this group because it has aggressive plans.
By 2023 & 2024, 5G costs will be low enough operators likely will deploy fairly widely in countries like Indonesia, Pakistan, and Nigeria. 5G capacity, especially from the use of 3300-4200 MHz spectrum, will deliver a better Internet. My estimates would be 5-25% 5G in 2025. With political will, 5G will spread even faster. People I respect strongly disagree.
Indonesia The 274 million Indonesians nominally have over 300 million mobile connections. More than 150 million are Internet-connected, nearly all of which are on mobile. Minister of Communication Johnny G Plate in January 2020 said, “Don’t rush to 5G.”5 5G will be modest until 2022, but I’d expect the very competitive Indonesian telcos will expand rapidly after that.
Pakistan Zong, owned by China Mobile, advertised its 5G trial until requested to stop by the government. Prime Minister Imran Khan has called for rapidly releasing 5G spectrum.6 It’s a very fluid situation; if Zong or another carrier starts building 5G, things could move fairly rapidly.
Brazil Telefonica Brazil has 11 million homes passed with fiber to the home, about three times as many as British Telecom and Deutsche Telekom combined. The economy is large enough that companies like Huawei are manufacturing domestically. Parent company Telefonica is one of the most technically capable in the industry. It is selling off the rest of its Latin American companies but intends to strengthen Brazil. American Movil/Claro will begin 5G in Brazil and other countries this year, probably modestly.
There’s little announced 5G and the country is in an economic crisis. But it is capable of remarkable advances.
Nigeria Nigeria is proud to be the largest country in Africa with politicians who regularly bemoan its Internet backwardness. Despite a billion dollars in fines, MTN of South Africa refuses to provide a decent Internet. It’s the largest telco in Nigeria and it should do better. I was shocked that CEO Rob Shuster said
This [4G] is the technology that would be used for very specific cases. It would not be a technology for everybody because most people don’t need it, your phone works fine on just 3G … What we are doing now is to learn from the technology and get our network ready for it but I think 3G is much more relevant in most of our markets, [mfn]https://www.benjamindada.com/5g-network-nigeria/ [/mfn]
Shuster has been very successful but that is nonsense. 4G and 5G are about 90% less expensive per bit than 3G. The upgrades will more than pay for themselves in cost savings. Mukesh Ambani’s Reliance Jio has signed almost 400 million subscribers in less than 4 years, all 4G. He’s actually profitable because the 4G costs are so low.
In 2014, Stanford Professor Paulraj told me that MU-MIMO will be the cost-effective way to bring enough capacity for a decent Internet to countries like Nigeria. It’s since been proven that 4G or 5G in the mid-band spectrum, using 64 small antennas of Massive MIMO, is remarkably cost-effective. MTN is a huge international conglomerate with thousands of engineers, some very respected. Any one of them can confirm the advantages of 4G & 5G.
Both Rwanda and Myanmar have 90% 4G coverage despite per capita incomes much lower than Nigeria. I’d like to believe things will get better in Nigeria in a few years.
I don’t know whether the decisionmakers are uninformed or …
Bangladesh’s economy, I was surprised to discover, “Has grown 188 percent since 2009.” [mfn] https://www.weforum.org/agenda/2019/10/bangladesh-is-booming/ [/mfn] $30 billion of textile production has moved here from China. It has a huge pool of programmers ad engineers graduating from college every year and a rapidly growing Internet. April: Coronavirus is deeply hurting textile exports.
“Bangladesh’s telecom minister has again said that the country will roll out 5G services by the year 2021, and that the government will ensure that the services spread across the country by the same year.” [mfn] https://www.dhakatribune.com/opinion/op-ed/2019/10/26/are-we-ready-for-5g [/mfn]
I have to do more research but it may be a factor in 2023 and later. Huawei already has a large team in the country and is promoting 5G.
Vladimir Putin and Xi Jinping had a historic meeting in 2019. To firm up the alliance, Huawei and MTS, Russia’s largest carrier, agreed to build a 5G network, starting in Moscow. Huawei is investing heavily in Russia, including large research and training centers. I haven’t seen any significant 5G announcements for 2020 but expect substantial progress after that.
Carlos Slim’s American Movil is deploying 5G in all of its Latin American territories. I believe these will be limited and primarily PR. Historically, Movil has been slow to upgrade but markets like Brazil are becoming competitive.
Seizo Onoe, NTT DOCOMO CTO, has long been an international leader in 5G. His presentation, The Myths of 5G, was very influential in the tech community. In 2017, he presented data strongly suggesting that 5G deployments would not be more expensive, which is now proven correct.
Onoe and his peers at Softbank and KDDI can match the capability of any other telco. They could easily build whatever the companies choose to invest in and could have been world leaders. DOCOMO instead cut capex 15% and the carriers agreed to hold up 5G until just before the 2020 Olympics. I’ve found no data on how fast the carriers will build.
The wild card here is the new fourth carrier, Rakuten, which is building one of the most advanced networks in the world. Tareq Amin started winning awards as a pioneer even before things were really working. Cloud native RAN, SSN, NFV, huge Edge Cloud, and every other buzzword. The cost per bit should be among the lowest in the world, although it’s months behind schedule.
It’s just getting rolling in March 2020 with projected prices half that of the other three. It’s all 4G for now. The upgrade to 5G was originally set for June 2020 but will likely be late.
In Italy and Canada, the carriers upgraded to meet the new fourth carrier. I expected similar in Japan but capital spending has been modest. The companies aren’t providing estimates and I don’t want to make a wild guess.
May update: NTT DOCOMO released a plan for 2.5 million 5G subscribers by the first quarter of 2021 and a rapid coverage increase next year. Rakuten’s 5G, as expected, is not yet ready.
Telcos from many parts of the world are bidding billions for the rights to compete with the current state-controlled monopoly. Two are likely to win franchises. The prices of 5G gear has come down so rapidly it may be included. Ethiopians need a better Internet; Mid-band Massive MIMO, 4G or 5G, is the right technical solution.
China Telecom and local investors are building a new third carrier with strong government support. CT is partnering on 250,000 5G radios in 2020 in China. It easily can deploy in the Philippines, but I haven’t heard anything concrete.
Etisalat is rapidly deploying 5G in the Gulf. Vodafone offers 5G in all its European networks. Orange has begun 5G in Europe. I believe that the government should demand the companies also bring 5G to Egypt, but I find no evidence that’s about to happen.
Viettel has become an international giant, with 110 million subscribers across 11 countries:
- East Timor
An investment of over $1B has brought 4G to ~90% of Myanmar. It is controlled by the Vietnamese military and strongly supported by the government.
The BBC is skeptical about Viettel’s plan to build its own 5G equipment. [mfn] https://www.bbc.com/news/technology-51178369 [/mfn] The country has been growing at about 6% per year for the last six years and intends to become a technology powerhouse.
Viettel plans to begin 5G in June 2020 and deploy widely in 2021.
Not long ago, Telekom Germany boss Dirk Wössner did not consider low-band truly 5G. He projected mid-band coverage of only 20% of the country at the end of 2021. It has made no change in the network but now claims it will reach 50% of Germany in the summer of 2020.
Of course, what’s happening is the company has decided to promote “low-band 5G,” which probably is slower than 4G in some locations. That’s definitely true in the U.S. but so far DT has not released average or typical speeds for its 5G.
Consumers rarely know the difference and most reporters are too lazy to ask what the real speeds are. Regulators are turning a blind eye so they can claim progress on 5G.
Telefonica in 2018 had a definite plan to use millimeter wave to offer a fixed service to compete with DT. DT’s unbundling price was too high to the Spaniards to make a profit. That plan was killed after Vodafone offered a better sharing deal to Telefonica to get approval of the takeover of Liberty Global.
DT has a dozen or so servers inside its network, which could be called “Level 4 Edge servers.” They are designed for 20-25 ms latency. That’s great for multi-player Pokemon GO but not fast enough for VR.
In Western Europe, the telcos seem to be investing more in pr than in 5G deployments so far. The Gulf countries are doing much better. The UAE, Dubai, Qatar, and Saudi Arabia are building quickly. Only Saudi Arabia has enough people to affect worldwide 5G figures.
Latin American and African potential should not be underestimated. The key advantage of 5G is greater capacity, which is badly needed. African telcos, many African owned, have connected hundreds of millions. They have the skills and substantial capital.
Some smaller countries – Ireland, Taiwan, Australia, UAE, Scandinavia, Benelux – will have fast growth but not enough consumers to make a large difference in world totals.
Europe is 2 or 3 years behind Korea and China. With the exception of Xavier Niel’s Eir (25%), I believe no European carrier has revealed current coverage or 2020 coverage plans. No subscriber counts are available, presumably because the count is remarkably low.
For the moment, the outlook is very pessimistic. I hope that changes.
DT & Vodafone have made many assertions, but as far as I can determine have upgraded fewer than 3% of their sites. Telefonica Deutschland has interesting ideas and has just raised capital spending $200 million. That’s enough to put them ahead.
The Brits, Spaniards, Turks, and Italians are just getting started. None of them have indicated they will reach even 10% coverage in 2020, although I’m hopeful about BT. The French, in cooperation with the government, haven’t even begun.
Vodafone and Telefonica, as challengers in several countries, would normally be much more aggressive.
This has to be considered both a market and a regulatory failure.
Turkey, in both Europe and Asia, has a large, Internet savvy population and a great deal of potential.
American Movil of Mexico, with companies in most Latam countries, promised to begin 5G in 2020 across the network. Telefonica will do similar, at least in Brazil. None have announced substantial plans. It’s natural for Latam to grow rapidly between 2022-2024.
Uruguay months ago announced it had begun offering commercial 5G service, but I’ve seen few facts on the ground.
Put bluntly, the Internet sucks in most of Africa, with caps so low you can’t watch many videos. The only way to change that is by adding wireless capacity because landlines are few.
Despite that, Kenya and many other African countries have an exciting Internet culture.
Massive MIMO with plenty of spectrum (100 MHz or more) is the best technical response. There’s enormous amounts of spectrum available from 2.5 to 4.2 GHz if regulators are effective. That band was designed for 4G, which works well. Almost all new equipment is 5G, not 4G, which will be the common choice.
GSMA predicts only 3% of Africans in 2025 will be using 5G. That would be a serious failure of regulation.
The Gulf States are probably second only to Korea in 5G deployment. Starting in 2018, Dubai, Abu Dhabi, Qatar, and even Oman have competed for who can make the earliest and most extravagant claims. The actual build lagged the announcements but is substantial.
Saudi Arabia, an affluent country with good 5G coverage and 34,000,000 people, is large enough to matter for the figures.
North Africa, Jordan, Syria, and Iraq have little beyond announcements so far.
Thailand, Singapore, Malaysia, and Vietnam have ambitious plans. Even Myanmar has 90% 4G coverage despite extreme poverty. The civil qualities of the governments are unfortunate, but they have done a relatively good job delivering communications.
Number of phones sold
With similar populations, Indians bought ~158 million phones in 20197 while the Chinese bought more than twice as many.
1/4 of phones sold in China in January 2020 were 5G. If that trend continues, China would sell ~100 million 5G phones in 2020. The % of 5G phones will probably increase enough to reach 150 million.
India is unlikely to reach 25% 5G phones until 2023 or 2024. Even then, it will only amount to 40 or 50 million phones. That would be twice any likely European country but far beneath China.
Overall, smartphone sales have been flat to down the last two years. People are holding on to their phones longer because three and four year old phones work fine. That’s likely to continue, keeping annual sales around 1.5 billion.
April: Q1 2020 was far down and recovery looks very slow.
Price of phones
Decent 5G phones in China now cost $228 (Xiaomi.) It will fall rapidly as 11 companies are chasing a market where only 4 or 5 can make a profit. The parts cost of a 5G phone is only $20-45 more than a 4G phone. The gap is narrowing as less expensive chips reach the market,
5G phones in Europe and the United States mostly cost from $800-$1500. It only costs about $2 to airfreight a phone. Carriers that bring China prices to the West will see a surge of sales. Xavier Niel, I’m looking at you. April: Several $400-500 phones have been announced in Europe and the U.S.
The important band for 5G is from 2500-4200 MHz, both in China and around the world. If the telcos don’t step up, others will and customers will buy sim-only service.
The price coming down will be particularly important in India. It has a large middle class that can absorb tens of millions of 5G phones. When the price falls to $150 and less, hundreds of millions of Indians can buy 5G. That is likely in 3-5 years.
Price is also important in Latin America. Argentina, Mexico, and Southern Brazil have higher per capita incomes than China. American Movil will offer 5G in the major Latin American countries in 2020. Initially, coverage will be modest but in a few years Latin America should see a significant 5G take rate.
Africa is experimenting with 5G and will deploy modestly north of the Sahara and in South Africa. Kenya has an ambitious Internet startup culture. South Africa and Egypt have substantial middle classes. However, volume 5G will depend on the price trend.
Coverage & Competition
Except Ireland and maybe Switzerland, no major European country claims even 10% coverage of 5G. The claim is made that European demand is weak. Of course demand is weak. Why would people pay the very high prices of European or U.S. 5G phones when they virtually can’t use them anywhere?
Every survey and measure of demand, as well as the Asian experience, suggests many will choose 5G. I believe that the first carrier with decent coverage and reasonably priced phones will gain a major advantage. (People I respect strongly disagree.)
A thought experiment Say a carrier like Telefonica Deutschland covered 20% of Germany in the next 4-6 months and publicized 75% coverage next year. Say they offered decent phones at 25% higher than the China price, with many between $370 & $550. Say – as is likely – that the competition coverage is ~10% and growing slowly. Say the price per month of 5G was similar to the price of 4G. (It is.)
How many people would switch?
Consumer surveys find many would change providers. The early Korean experience was a large shift to LG Uplus until KT and SK caught up.
Telefonica Deutschland (or Vodafone Germany) could do that by raising capex $200-350 million. That’s a rough estimate for upgrading several thousand existing towers. Upgrading selected towers just isn’t that expensive.
Make your assumption about how many of the 82 million Germans will switch. Add something for the value of being perceived as the most advanced. Do the arithmetic.
It’s absolutely the right move if the competition does not change plans to match. I’m 90% sure it makes sense even if the other companies decide to catch up. Networks take time to build. It would probably take 6-12 months to plan and build, enough time to more than recover the costs.
Using Telefonica Deutschland as an example is deliberate. It recently raised capex ~$200 million. It’s public comments do not suggest a rapid expansion of 5G so far. Xavier Niel’s Eir has covered 25% of Ireland. Vodafone (and AT&T) are holding back against the strong advice of technical experts.
Korea reached 93% coverage in the first year. China installed 132,000 radios in 2019 and has 550,000 in the plans for 2020. Even Ireland is 25% covered.
70-80% by 2022 is practical in almost every developed country by 2022. At most, it would require pulling 1-2% of sales into capex a year or two.
It’s been clear since the middle of 2019 that the cost of 5G deployments is not higher than 4G. NTT DOCOMO CTO Seizo Onoe has been calling the high cost of 5G a “myth” since 2016. The experience of the first year, including Sprint in the US, proves he is right.
22 out of the first 24 5G networks are mid-band. Massive MIMO – 64 small antennas – delivers the reach and performance of 4G 1800 in 5G 3300-4200. The cost isn’t small, but telco capital budgets are huge.
If 5G is important, almost every developed country should be mostly covered quickly.
Government promotion is making a difference in Korea and China
From Angela Merkel to Donald Trump, western leaders have declared the importance of 5G. Except for releasing mid-band spectrum, the governments have done very little that has resulted in more deployment.
The telcos have used the desire for 5G to advance their agenda but haven’t built very much. In the U.S., the FCC reduced telco payments for pole attachment and municipal right of way because the telco people in DC said that would result in a massive increase in small cells. A year later, small cell deployments are below previous plans.
Nearly all “incentives” for telcos fail at their primary purpose. Instead, most become boons to the companies and the shareholders. Few officials can stand up to the company lobbying, especially the million dollar silver tongues in D.C. with budgets in the hundreds of millions. It’s called “regulatory capture” and is not quite ubiquitous.
My completely unscientific observation is that typically only about a quarter of the funds spent efficiently serve a purpose such as reaching the unserved; more go to waste and company profits. But the results vary widely. Nearly all of the U.S. Broadband Stimulus was wasted. The U.S. Lifeline $10/phone for the poor does have waste, but most of the money connects people.
On the other hand, Korea “urged” the companies to build, with specific targets. The Chinese gave a firm direction “Accelerate.” In both cases, the government set 5G as a priority and worked directly.
Similar could be achieved in other countries by attaching requirements to the decisions telcos need. Rarely is this done well, but Mathias Kurth successfully required filling in the “white spaces on the map” before LTE spectrum could be used in the more profitable cities.
It seems like every fourth story in the Chinese tech press trumpets 5G and how it will transform the economy. People’s Daily & Xinhua also are loaded with promotion. The telcos would have to spend billions advertising to achieve as much.
Xi himself has called 5G a crucial part of the recovery plan. In China, when the leader speaks, everyone falls in line, including almost all of the press.
The government owns the telcos. Since Minister Miao Wei in Spring of 2019 said “Accelerate 5G,” the response has
Spectrum is only a major coverage factor in countries that haven’t distributed mid-band. As I write, that includes the U.S. and India but both are moving forward. In countries where most telcos have 80 or more MHz between 2.5 GHz and 4.2 GHz, more spectrum is unlikely to have much effect.
80-100 MHz in mid-band has enormous capacity, more than most carriers are able to sell. Where carriers have mid-band, purchases of higher frequencies are for the future or to prevent competition.
More spectrum will reduce carrier costs in time, a good thing. It’s possible added But it will be years before a meaningful impact is likely.
3300-4200 MHz is the primary frequency for 5G. 5G is designed to use a 100 MHz band of spectrum, which is nearly never available below 2500. Massive MIMO, deploying in 4G since 2016, makes this band practical for mobile.
2500 MHz is also important. It’s in use at China Mobile and Sprint, both of whom are getting excellent results. 2500 MHz is golden spectrum, with good reach and low buildout costs.
Millimeter wave – 24 GHz and up – has gigabits of capacity, about three times as much as mid-band. The only carrier in the world with extensive plans for mmWave is Verizon, which will use lower frequencies for the majority of its network to keep the cost down.
Everyone agrees that millimeter wave will be necessary one day if demand keeps expanding, but most think that day is far off, probably a decade. AT&T, Telstra, the Japanese, and the Koreans have modest plans for mmWave but little deployment. Neville Ray at T-Mobile U.S. believes mmWave will be cost effective in certain high-traffic locations and has invested heavily in spectrum. This was a big win for the mmWave advocates because Neville had been one of the strongest critics.
I continue to recommend that most telcos do a trial or even a modest deployment of mmWave. Adding mmWave requires extensive staffing training and revised systems. Neither can be done quickly if you discover you need to expand your capacity.
Hans Vestberg of Verizon believes the greater capacity of mmWave will prove a strategic advantage. That’s possible, although Hans has never given specifics beyond fixed wireless in some locations. If Verizon proves to be on target, your competitor might follow and force you to do likewise.
Having multi-gigabit mmWave in your network can be great for image and pr. Everyone wants to be the best, but few have capitalized on the reputation as well as Verizon has. Verizon was first in the U.S. to fiber to the home and was first in the world in 4G. It was perceived to be much better and in many ways it was.
I’m confident that a (modest) effort in mmWave is good insurance against surprise moves from the competition. It should more than pay for itself in marketing.