Telefonica Brazil passes AT&T, Verizon with 16M FTTH homes passed

Enrique Blanco, Telefonica CTO, is telecom's strongest believer in fiber. His Brazilian operation has now passed 16 million homes, almost a quarter of the country. The goal is 24 million in a few years. Only 3 million have purchased so far, but fiber generally catches 40% of broadband after a few years. 

From the financial call:

FTTH revenues were up 56% year-over-year, while IPTV was up 26.9% resulting in a combined growth of 47.3%. Analyst Carlos Sequeira noted, "The FTTH operation is growing super fast. Net additions are impressive. So everything is going super well." He went on to ask if Vivo could even speed up the construction.

This is part of a worldwide trend of adding fiber. In the US, Frontier is building again and talks of 3 million as it exits bankruptcy. BT is passing ~2M homes/year, up from almost nothing three years ago. Orange has just announced it's "accelerating fiber." It has 22 million connections available, most shared with Free. That's over 60% of the country. 

Two years ago, almost no one expected this kind of growth. I didn't.

In six weeks, wireless could reach 30%-60% of students without a connection

The U.S. wireless networks could provide a decent Internet connection - 25 to 75+ Mbps - to tens of millions in weeks. The networks have enough spare capacity, even though Verizon and AT&T have been cutting investment for several years.

The obstacles are cowardice and ignorance in Washington. Buying millions of lines, the government should be paying a wholesale price of $8-15 per month per connection. Of course the companies would rather be paid a retail rate of $30-50 per month, and some in DC don't realize the actual costs or are afraid to take on the telcos.

Telecom is a high fixed cost, low variable cost industry. Verizon and T-Mobile have built a network to ~97% of the population. Once the network is built, the marginal cost of adding a subscriber is $3-7 per month. That's why companies compete to sell Lifeline service at $10/month. (Cable economics are similar, which is why Comcast can sell a $10 offering to the poor.)

20 million homes added at $15/month for 10 months of school is $600 million. That's about 7% of the Universal Service Fund. I believe there is also money allocated to connecting students in the CARES stimulus.

Read more ...

AT&T killing DSL (Dave in USA Today)

AT&T will "phase out outdated services like DSL and new orders for the service will no longer be supported after October 1." Thanks to Rob Pegoraro for including me in his story. AT&T will continue offering the DSL from their field terminals, calling it fiber to the node. It's the central office connections that are being closed, most under 6 Mbps.  

DSL is now often delivering a gigabit with G.fast. Adtran is actually seeing an uptick in G.fast in Germany, served from fiber to the basement. G.mgfast is still moving in standards, for 3 and 6 Gbps. But the faster DSL requires a terminal in the field, normally backhauled by fiber. That cost money the telcos didn't feel like spending.

Verizon and AT&T, around 2011, “basically gave up on fighting cable over a third of its territory,” I told Rob. Both decided they ultimately saved money with a single, wireless network. They continued to milk the territories for whatever they could get, and hemorrhaged subscribers to cable. Bruce Kushnick discovered that both companies are now down to under 700,000 pure DSL customers, losing more every quarter.

I estimated that DSL remains the only landline choice for 3 to 6% of the U.S. LTE is now routinely 50-125 Mbps. Elon Musk is launching hundreds of low orbit satellites with DSL like latencies. There's little need for slow DSL.

Cable is the best choice for the 2/3rds of the US where the telcos haven't built fiber home.

ASSIA Equipe Work-From-Home Manager

John Cioffi pretty much invented DSL and has spent most of the last 30 years making your connections better. ASSIA developed CloudCheck, which delivered optimum Wi-Fi experience. It announced this week EQUIPE, which manages employees' WI-Fi and connection reliability.

Read more ...

Realme 5G down to $145

The Realme V3 5G  has a 6.5-inch screen, three rear cameras, and a 5,000 mAh battery. The Mediatek Dimensity 720 is similar to the popular Qualcomm 765 in Antutu testing. It should be fine for most practical purposes, although the main camera has only 13 megapixels. 

CEO Xu Qi expects to sell 50 million phones this year and soon reach 100 million. Realme, Oppo, Vivo & OnePlus are brands of BBK. Added together, in some quarters they are #2 in worldwide phone sales. They spend $billions on research.

 I'm probably going to raise my 2020 yearend estimate of 210,000,000 5G subs because of the low prices and rumors about iPhone 5G.

Qualcomm 4 kilometer mmWave not close to Ted's 11 kilometers in 2016

Qualcomm and Ericsson claim a "groundbreaking milestone" delivering millimeter wave 5G almost 4 kilometers. NYU Professor Ted Rappaport did 11 kilometers four years ago. Qualcomm's actual achievement is making the antenna smaller than the one Ted used. They should at least have acknowledged his work. See Millimeter wave "5G will come sooner and reach longer distances" Rappaport from December, 2016

Put one radio high enough to provide line of sight and use directional antennas and of course mmWave can go a distance. It requires the perfect topography, although a radio on top of a high tower may have some practical use.

In the next few months, Verizon and others will be demonstrating new antennas that may solve the catastrophe that was early mmWave. Verizon in 2017 thought it would get 600-1000 meters of reach. In practice, it was often less than 200 meters. Verizon stopped taking orders although the publicity never stopped.

Ted pointed out:

Many people continue to propagate the incorrect myth that mmwave is severely limited in distance. This is not accurate. The fact is that the distances at mmwave will only be limited by rain and fog, not by the nature of mmwave. This is because the "lossiness" of mmwave, compared to lower frequencies, only occurs in the first meter of propagating distance, but this "higher loss" is canceled out by keeping the antennas the same physical size at all frequencies.

Read more ...

Marvell: 5 nm 20-40% better

Kevin O'Buckley of Marvell tells EE Times, “We’ve been able to achieve on average 40 percent lower power for a given design point and performance point. We’ve been able to achieve 40 percent greater integration, mostly measured as die area shrink, which can be used either to pack in more performance in a given die area or in some cases, lower costs.” Marvell finds, "TSMC’s 5nm technology which delivers approximately 20 percent faster speed or 40 percent power reduction compared to the previous 7nm generation."

Moore's Law is slowing down, but how much? Far too many claims are made based on inappropriate comparisons. O'Buckley answered me, "40% lower power was by taking some key designs in N7 and moving them to N5 at fixed performance (and lower voltage.)" (There is an immediate process, TSMC's N7+, which uses EUV.)

Read more ...

$400 TCL REVLL 5G at T-Mobile: Here comes 5G in the USA

5G in China now has about 100 million subscribers but the results in the West have been dismal. T-Mobile is bringing lower-priced phones to the U.S. while upgrading ~3,000 towers per month. On the lightly loaded network, speeds average over 300 Mbps. A promotion drives the cost down to $200.

1/3rd of the U.S. will be covered with mid-band 5G yearend, and They won't stop. TMO has passed AT&T to become #2 in the U.S. and is going after Verizon. For at least a year and possibly two or three, TMO will have a much better network.

Both Verizon and AT&T have been cutting capex and praying consumers won't care that most of their "5G" is slower than decent 4G. Verizon's mmWave Ultra is the fastest network in the world, but 5G customers connect to it less than 1% of the time.

Read more ...