US Q2 5G 4-5 million. Year-end 15-25 million

According to Ken Hyers at Strategy Analytics, 3.4 million 5G phones shipped in Q1 in the U.S. and 2.7 million in Q2. Stores were closed and promotion limited. Some of the 6.1 million shipped were in transit or store inventory, hence the total of 5-6 million plus (modest) carryover from 2019. Update August 10 M Science released a lower estimate of 5G phone sales and I have reduced my U.S. Q2 figure to 4-5 million.

Most were expensive Samsung phones. SA gets information directly from the makers and has proven reliable in the past.

I expect far more than 2.7 million in the next two quarters. The 5G iPhone is in limited production, with an announcement expected in September. Hans Vestberg at Verizon expects large sales of the iPhone 5G, although there are rumors of limited production this year.

Phone prices are down to US$230-270 in China and 400 euros for decent models. (6.5″, three cameras, …) Nearly none of those models are available in the U.S. The cheapest 5G phone at Amazon was the Nubia gaming phone at US$629, made by ZTE. Nothing else was less than US$800. It’s less than US$2 to airfreight a phone to the U.S., so inevitably less expensive units will reach us. They may not access Verizon’s mmWave service, but mmWave connected only 0,4% of the time per Open Signal.

Sasha Sagan at PCMAG is enthusiastic about the OnePlus Nord, becoming widely available in Europe for 400 euros. He tweets “The @OnePlus Nord is an amazing midrange phone which fills me with rage that we don’t get these kinds of options in the US.” I’m sure we’ll see similar by September, driving up 5G sales.

In addition, both real and faux 5G networks are deploying rapidly. AT&T just announced they cover about half of the country with low-band, with speeds lower than decent 4G. T-Mobile has done similar. Verizon promised low-band to half the country by June although it is behind schedule. Verizon continues to refuse data on 5G coverage, except for the meaningless count of cities.

More significant, T-Mobile is rapidly building 5G midband in the Sprint 2.5 GHz golden spectrum. Neville Ray says the average speed is over 300 Mbps down as the network is almost unloaded. Nothing’s official, but T-Mobile is likely to cover 20% of the U.S. with mid-band very soon and 50-80% next year.

Verizon is building 4G/5G very widely in the shared CBRS spectrum, which it will supplement with dedicated 3.5 GHz spectrum after the auction is concluded.

Analyzing Telecom

Hottges
5G is not raising Canadian capex

TRAI is demanding that all telecom gear, not just phones, move to Made in India.

Turkcell’s Superbox 4G fixed wireless added 91,000 subscribers Q2. With far more 4G capacity than they can sell, telcos including Verizon are promoting 4G fixed.


Canadian telcos claim 5G at 1.7 Gbps. Actually, in the spectrum they have, 300 Mbps would be extraordinary in a real test. In the lab, they can combine four 4G bands (1.2-1.4 Gbps) with a single 5G band to reach that figure. 4G is still faster than low-band 5G until many problems are solved.

The first substantial 5G testing results are unbelievably bad. (Open Signal.) The scam of “low-band 5G,” spreading in Germany, is slower than decent 4G. Indoor results, including mid-band, are dismal.:

  • Verizon mmWave 5G customers connected 0.4% of the time despite $billions spent on the network.
  • Korea has >90% outdoor coverage but only 15% 5G connection rate
  • T-Mobile 5G 49 Mbps, AT&T 61 Mbps. Canada 4G 69-75 Mbps
  • My analysis, that 5G capacity would be great, may be totally unsound.
  • Latency at Verizon is 30 ms; no 5G deployment has very low latency. 1-10 ms is a fantasy outside the labs.
  • 10-20 ms latency will be possible in the few places outside China building Edge networks.

17.6 million 5G phones in June in China is on track for 150 million year-end. Fine phones from US$230-260 and $13 service plans driving demand, although few practical uses have developed.

Craig Moffett warns about “the fallacy of marginal cost advantage. … It’s not true with telecom.” Remarkable technical advances have resulted in more capacity almost everywhere than the telcos can sell. Result: the low costs of Verizon’s mmWave or Rakuten’s new network are useless unless buyers can be found. Verizon’s Hans Vestberg is learning that the hard way.

5G slower than 4G in the US Finally, Data: US 5G slower than Canada’s 4G. Believe it and Germany.

A huge question for Edge inside telco networks is whether the web giants will take over. It’s now clear most Edge is a hybrid cloud and hybrid clouds need huge teams of engineers to manage. Many telcos are deciding to hand over to the web giants and take a (modest) percentage. Even the telcos building their own, like Verizon and DT, will find profits hard to come by, Pal Zarandy notes. “Buyers like Google, Amazon, Microsoft, Facebook with massive bargaining power would bargain away MNOs’ margins. Near zero marginal cost network economics => near zero margin to be made on wholesale access.” He’s talking about other services than Edge but the point holds. Will the telcos be able to cash in on their terminating monopoly? To be seen.

Keith Bradsher in the NY Times dramatically makes clear the effectiveness of industrial policy support in his article on the growth of Chinese medical supply manufacturers. Indian telecom production is booming under protectionism. Ericsson & Nokia benefit from over a $billion in EU R&D in 5G.
The US chip industry has come together to demand $30 billion. Senator Mark Warner wants $billion for 5G in the U.S. From cruises to autos to pharma, business leaders decry government support except for their own companies.
US policy for 40 years has been to stimulate the economy with tax cuts, although only a small fraction of the money goes to expanding investment. The money would be far more effectively used for investment in strategic industries. That works whether the government is capitalist or communist – so long as it’s not captured by industry or totally corrupt.

“The Moving Finger writes; and, having writ, moves on.”  If you’d like America’s top China trade rep to speak for you instead, Jeffrey Gerrish has abandoned Trump and returned to Skadden Arps. Million dollar retainers expected.

“What jumped out to me was how users’ current thoughts are mostly around smartphones, IoT/sensor data collection, display boards, CCTV and so on. Whereas MNOs are focused on automation, robots, AR/VR and the ‘sexier’ applications.” Dean Bubley writes, adding that all the likely apps work fine on 4G. Telefonica CTO Enrique Blanco also wonders whether 5G is actually needed for IoT. (Hint: not.)

Some people still think 5G is expensive to build. AT&T & Orange are cutting capex while building 5G. Deutsche Telecom is keeping it flat. 5G does not cost more than 4G.

Tim Hottges, CEO of DT, is the Great Welsher. TMO wants out of merger agreements to serve 93% of Californians with carefully tested 300 Mbps by 2024 and to cut 1,000 jobs.

Telcos’ greatest problem: demand growth is slowing while technology is increasing at a ferocious rate. Hans Vestberg of Verizon claims, and I can confirm, that cost per bit is going down at 40%/year. Traffic growth is now typically 25%-35%. Every investor implicitly acknowledges this: the number of subscribers is crucial to the stock price response after quarterly earnings.
That’s the reason why millimeter wave, the real 5G, finds so little demand. When the enormous capacity of mid-band spectrum became clear, carriers realized they would have more capacity than they could sell without mmWave.
The primary factor in telco profitability is how close the carriers come to cartel-like pricing. CEOs talk about “rational pricing,” a true description that should be a signal to antitrust enforcers. When carriers have unused capacity in most locations, the natural act is to break from the (unspoken) cartel.

For a good understanding of AI today, read Architects of Intelligence: The truth about AI from the people building it by Martin Ford. To understand China, the U.S. and what’s going on, Kai-Fu Lee’s AI Superpowers: China, Silicon Valley, and the New World Order is deeply informed but easy to read. AI/Machine Learning has limits. It is great for translation, video surveillance, and speech recognition. It’s been disappointing, so far, in tasks like network optimization. (That could change)

Multi-cloud and hybrid cloud are crucial trends in data centers, Equinix is demonstrating, per Nick Del Deo of Moffett.

Continue reading

5G 2020 Country Review

In August 2019, early data from Korea and China convinced me that predictions for 5G in 2019 were much too low. In September, I published 5G 2020 Predictions: Surprising and surprise free 195-205 million, Low 175M, High 230M, also much higher than others. China had announced 600,000 cells and 150 million subscribers for 2020 and I chose to believe them. I reviewed plans in two dozen countries and produced this “Country Review.” I then did 5G 2020-2025: Surprise Free, Low & High Estimates.

Share freely. All use the Creative Commons Attribution-NonCommercial License CC BY-NC

Update June 18

If it’s slower than 4G, is it really 5G? AT&T and Deutsche Telekom have each revealed they have covered about 1/3rd of their country with “low-band 5G.” They refuse to answer questions about speed or latency, but they almost surely are slower than decent 4G today. It’s 4G LTE with “5G NR software,” which does very little below 2 GHz. Don’t believe the hype!

Not long ago, Telekom Germany boss Dirk Wössner did not consider low-band truly 5G but the marketing people have won. 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.

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.

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.

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.

Intel has a 5G cloud-native core working, according to a well-informed European. So does Cisco.

Update June 10

I have restored the December 2019 “surprise-free” of 210 million. 15 million 5G phones were sold in China in May, almost half of total phone sales. 618 holiday sale prices for a 5G Xiaomi Redmi 10X are US$198. A Huawei Honor phone was promoted at US$148 but sold out almost immediately.

Apple will begin volume production of 5G iPhones in July (Digitimes.) Rumors persist sales will be limited until late in the year, however. Verizon and others expect very strong sales when the 5G iPhones is available.

The three Chinese carriers have lowered the basic price of 5G from ~US$18 to ~US$13.

Continue reading