Fast Net.news

Q2 worldwide: 18M new terrestrial connections

Point topic Q2 terrestrial 230 75% of the growth was in Asia, mostly China. It's the slowest quarter in several years, as many people are now going wireless-only for data. The world total is 1.1 billion, which doesn't include cellular 4G. Point Topic, a highly respected source, reports a continued shift from DSL to FTTH, led by China. Nearly all of China's 400 million connections are now FTTH.

FTTH was up 19% on the year; all copper, down 7%. The growth in fiber astounded me, but I have confirmed the expansion at about 20% from several sources. China is almost done, but France and Spain are going quickly. England, Germany, and Italy are now joining in. I believe the 50M planned by Reliance Jio in India will mostly be FTTH, not just to the basement.

As you see in the chart at left, Asia and Latin America are about 65% of connections. India, which has less than 35M landlines, has well over 400 million 4G connections. 4G and the better 3G are fast enough for HD TV and should be included in broadband in my opinion. Including all wireless, about three-quarters of Internet connections are in the Global South. The gap is widening.

Wrong, Elliott! AT&T has one of the best managements in world telecom

ATT comparison 5 year 230 AT&T's Randall Stephenson, John Stankey, John Donovan, Ralph de la Vega and Bill Smith were probably the most effective telecom management of the last decade. Paul Singer and Jesse Cohn of Elliott Management - or whomever they brought in - could never match the record of the current management they want to displace. The contentions in the letter (below) are often misleading and in some cases simply errors. The best pros, like Craig Moffett, don't see a potential for big gains in T.

For example, Singer writes

"AT&T today is in prime position to be the early market leader in 5G given its premier spectrum positioning, ... We believe AT&T will be able to quickly move forward while its main competitors remain either spectrum-disadvantaged or distracted as they integrate major transactions. However, while AT&T is well positioned, success in 5G will require meaningful investment and improved execution; anything less and AT&T risks missing this opportunity and falling behind again."

That's nonsense for two reasons. First, Verizon and Sprint have better 5G spectrum positions than AT&T. AT&T and Verizon have similar low band spectrum. Verizon has 800 MHz of 28 GHz mmWave spectrum, more and better than AT&T's mmWave. Sprint's 160 MHz at 2.6 GHz is providing exceptional reach. It currently has covered over 11% of the U.S., more than any one else in the U.S. and could rapidly expand - with or without the merger.

Adtran will build PON in Egypt

Adtran is entering a joint manufacturing deal with the Arab Organization for Industrialization (AOI,) a large manufacturer controlled by the Egyptian military. AOI assembles tanks and jet aircraft. It produces AK-47's. The company is diversifying into solar energy and communications. A GPON box is much simpler than a supersonic jet; Egypt should have no problem with the manufacture.

Egypt has a strong industrial base that was first developed by Muhammad Ali at the beginning of the 19th century. It was stifled under British colonization until 1952. It exports widely and has respected universities.

As Chinese wages have gone up, manufacturers have been moving to find lower labor costs. Vietnam has enjoyed a mini-boom. Egypt's nominal per capita income is about midway between Vietnam and China, low enough that wages will be competitive.

For the last two years, FTTH has increased by 20% or more each year.

From $10: Jio expects 35M FTTH

15 million Indians have put in reservations for Jio fibre. Starting Sept 5, Jio will be connecting millions of people. With 300,000 kilometres of fibre in place and 500,000 "trial" customers, deployments will begin in 1,500 cities. Jio will expand rapidly from there.

Prices begin at ~US$10 for 100 Mbps. Other offerings, including video services and much more, will be priced up to $150. Ambani promises that next year he will include Bollywood films on the same day they open in theatres.

Reliance Jio in three years has connected 340M Indians to 4G, more Internet subscribers than the US has people. Neither I nor many other people believe that possible. I'm no longer a doubter.

He's invested over $30 billion to build the network and reached (modest) profitability. Reliance sold 20% of its refining business to Saudi Aramco for $15 billion. Towers and fibre assets are likely to be sold for more cash. The funds are there to manage $billions in initial losses.

Jio bought several cable companies to have people on the ground for installations and an initial customer base. It has a massive retail presence that will help sales.

The Internet is no longer a Western lake.

Landline traffic growth 30%, >terabyte users becoming significant

James Wright bought a 4K TV and suddenly Comcast's terabyte cap hit him. I've written before that Comcast's terabyte cap was reasonable. Bandwidth is cheap but it isn't free. That is no longer true. The bandwidth needed has gone up ~30% per year. The cap hasn't risen despite the cost to deliver each bit has gone down dramatically. The cap is now hindering cordcutting.

OpenVault found the average bandwidth use in 2019 in the US was 271 gigabytes, up 25%. Five years ago, very few drew a terabyte per month. Now, the number is meaningful and rising, particularly among those who don't take cable TV. Internet-only households consumed 85% more bandwidth, 390 GB per month.

Many 4K TVs now cost US$300-700, including a remarkable new one from Huawei. A typical American home watching 4 hours of TV a day at 4K is close to the terabit. Add a kid or two, and overages will be hard to prevent.

The European average in 2019 is 170 GB, up 28%. Unless Comcast rapidly raises the cap, users are pressured into taking Comcast's TV.

(Thanks to Robert Conger of Adtran for pointing me to how 4K TVs may be an obstacle to fixed wireless in large volumes.)

Chip bottom: Sales down 7%

IHS forecasts a 7% drop in 2019 chip sales as prices plummet in memory, blockchain miners disappear, and mobile phone sales drop. This is consistent with many other data points, such as TSMC's quarterly earnings. Lower prices for memory chips and most others are holding down equipment pricing, including PC's and phones.

Four sectors stand out in the chip business.

5G wireless chips can only be produced on TSMC's 7-nanometer production line or by Samsung, mostly for its own use. Huawei and Qualcomm are producing 5G chips with reasonable yields at TSMC. Six months ago, 5G phones were not expected before the summer because Apple had bought out most of TSMC's 7 nm production. The decline of iPhone sales opened some capacity and Qualcomm is supplying chips to Xiaomi, Oppo, Vivo, and ZTE. China Unicom is offering phones from those four, Huawei, and I believe Samsung.

Qualcomm's prices are reasonable enough that Xiaomi announced 5G phones at US$600, which others will match. Intel is now out of the market. MediaTek and UNISOC are behind, so Qualcomm has little incentive to adjust prices. IHS emails me:

In the near term, our analysts don’t see much pressure on BOM pricing due to the strong Qualcomm market presence but in the longer term, BOM prices will be under pressure as soon as China Mobile comes online with their 5G SA network. It is also unlikely that semiconductor manufacturing (of application processors and modems) will drive the market into oversupply.

Mobile World Congress 10 gig - on fiber

A year ago, people refused to believe Xavier Niel's Salt was selling 10 gig for US$53 in Switzerland. In August, I reported Huawei can bring 10G prices within US$5 of GPON prices.

10 gig is now planned at Bell Canada, in Korea and I believe parts of China, and by smaller carriers in England and the U.S. The cost is so low I expect to see it many more places.

Ookla, developers of the popular speedtest.net, brought their new 10 gigabit test software to the Mobile Web Conference. There is no mobile offering that can confirm the speed. Instead, they brought two Mac Minis and hardwired them.

Samsung claimed 5G will go 40 gigabits. Others suggest 60 gigabits or more. Those are politicians' truths.

Wikipedia edits Cloud & Edge computing

I began the process of updating and clarifying the first paragraphs of these articles. Improvements welcome.

Cloud computing makes computer system resources, especially storage and computing power, available on demand without direct active management by the user. The term is generally used to describe data centers available to many users over the Internet. Large clouds, predominant today, often have functions distributed over multiple locations from central servers. If the connection to the user is relatively close, it may be designated an Edge server.

Clouds may be limited to a single organization (enterprise clouds,) be available to many organizations (public cloud,) or a combination of both (hybrid cloud.) The largest public cloud is Amazon AWS.

Edge computing brings memory and computing power closer to the location where it is needed. Definitions vary.

More Articles ...