Template Tools
Q2 U.S.: Comcast strong. AT&T and other telcos weak

DSL getting killed where not upgraded. AT&T lost 55K broadband customers. They want to abandon about 25% of their landline territory and go wireless only. The areas are profitable but rural wireless customers are extraordinarily profitable once the network is built. So they want to eliminate competition even from their own landlines. They stopped maintaining those lines years ago and many have 1999 technology 6 megabit DSLAMs. AT&T and Verizon are best thought of as companies with two parts. The upgraded areas, FiOS and U-Verse, are doing fine against cable. They are getting clobbered where they haven't upgraded but allow that/ They want to shut off most of those lines. Century/Qwest lost 2,000 customers and Windstream 17,000. 

     Comcast added 203M subscribers to 21,271,000. Some received the $10 rate for poor familes but I believe most at paying full rate. But Cablevision and Cable One actually lost customers. Telcos can compete just fine against cable; in Canada at Britain, telco DSL is beating cable. http://bit.ly/1t89qXa 

     Revisiting the debate on fiber versus DSL. AT&T lost 55K, Verizon added 46K. There have been several recent quarters where Verizon FiOS fiber home did noticeably better against the competition than AT&T's fiber/DSL. Verizon offered 25 and 50 megabits upstream, while DOCSIS is stuck at 1-5 megabits up. Fiber remains a magic word, with connotations of modernity and reliability. The glamour of Google's well-publicized gigabit fiber may be helping the (not so fast) Verizon variety. Different marketing and pricing strategies could explain the (relatively modest) difference in results. 

Read more...
 
Current Edge Tolls Actually Small

They could gouge if the precedent is set. I've called the current prices that the carriers are demanding from the video people "a hill of beans," that wouldn't be worth fighting over if they weren't likely to go up. Respected streaming media analyst Dan Rayburn pointed me to this comment from Chris Libertelli of Netflix confirming what I knew from off the record sources.  

“It’s fair to say that the interconnection fees that are being charged here are — and I think our CFO said this at the same conference that Jim alluded too — not as big for example as our content costs, so the price differentiation you’re referring too is likely to be so small that the costs of it would outweigh the benefits because the differences in price wouldn’t be so huge and customer affecting.”  http://bit.ly/1zdFUBg

    Rayburn I think goes too far suggesting that means edge tolls/sender pays are OK. We all know Ed Whitacre said "They are not going to use my pipes without paying."

Read more...
 
New York, Los Angeles Getting 400 Meg Cable

300-400 Meg (shared) now standard off the shelf. I've heard from an engineer that the gear for New York is being installed and tested. They've been losing customers to both Verizon FiOS and AT&T U-Verse in LA, Time Warner Cable  is upgrading download speeds. My 15 down, 1 up service ($63/month) is set to go to 50/5 for the same price.

Going from 160 shared to 300-400 shared in DOCSIS 3.0 is only a software change on most equipment. Extra backhaul costs are minimal. Now that the analog switch-off has freed spectrum channels, expect that change almost everywhere. Giant Liberty Global is upgrading across Europe and offering 200-250 meg service routinely.

Time Warner has also committed to a gigabit (shared) in 2016, per this strong LAT article by Paresh Dave http://lat.ms/1rEQZrY.  No one except the engineers believed John Chapman of Cisco in 2004 when he promised the gigabit in DOCSIS 3.0 but now equipment is starting to ship. Last year, TWC also offered a gig in 2016 to North Carolina.

Read more...
 
"This argument is utterly farcical."

Straight talk in a DC world of obfuscation. The automakers want to restrict users of WiFi despite having 75 MHz of monopoly spectrum of their own. That's enough to build two Verizon sized networks, far more than makes sense to monopolize in 2014. TIA, the manufacturers' group, is right to call it farcical. http://bit.ly/Autofarce  The car guys want to hobble the WiFi band near the spectrum they control above 5850 MHz. I'm tempted to petition the FCC to return 40 MHz of the auto spectrum to the public domain. 35 MHz is plenty for even safety-related car communication.

     Denise Coffey and Dileep Srihari of TIA were blunt, "The Automakers state that the FCC must protect the 'reasonable interests and expectations of its operators' – with which TIA, of course, concurs. However, what the automakers assert to be their “reasonable” expectations is entirely unreasonable, bordering on unbelievable."

     In five or ten years more people will understand that we already have too much monopoly spectrum. WiFi provides enormous benefits with much more to come.  Soon, adding WiFi to today's 100 megabit vectored DSL would allow nundreds of megabits to most homes without needing expensive construction. (The tech works although the business side needs to be developed.) Requiring car makers to add 50 cents or so of parts to make their radios work right is a very small price for major public benefits.

    'Use it or share it" - Mike Calabrese's compelling slogan, is the way to go on all spectrum.

     

 
DSL beating cable yet again

Winning in Canada and England.  DSL upgraded since 2005 goes 25-50 megabits (VDSL2) and competes well with cable. 1998 ADSL (3-6 megabits) gets clobbered by cable. The dismal results from areas with obsolete equipment, especially in the U.S., have convinced many that the race is over. It's a different story where the DSLAMs aren't 10 years out of date.

    British Telecom "added 104,000 retail broadband customers." Cable competitor Virgin Media actually lost 300 subscribers. Bell Canada added 18K and Telus 15K. Rogers and Shaw cable added only 14K, combined. In a recent quarter, AT&T actually beat cable in U-Verse homes, about 60% of their network.

Read more...
 
Triductor, the New Chinese VDSL & G.hn Chipmaker

Yaolong Tan "Our chips are 8% faster than Broadcom's" Yaolong Tan earned his doctorate at UCLA’s Henry Samueli School of Engineering in 2000. He worked in Silicon Valley for years, and now is back in China. He's CEO of Triductor, founded in 2006. He is shipping VDSL2 vectored cpe chips to Chinese manufacturers who in turn are distributing the boxes worldwide.

Tan received his UCLA degree directly from Henry Samueli. He has enormous respect for Broadcom. But he's not afraid to take on Broadcom's chips. China is deeply committed to replacing imported chips with Chinese designs. Tan said,"The technology of this chip used to be monopolized by America, so our country had to spend tens of millions of dollars importing from overseas. What my team and I want to do is realize the localization of this chip in the new developing wave of semi-conductor industry, to equip Chinese people with their own high-speed video networks."

Triductor, like HiSilicon, has also announced a G.hn chip. While G.fast is getting the publicity, thanks to an effective campaign by the ITU, it's two years or more away. G.hn, a much simpler system, is already being used to extend "fiber to the basement" to apartments at hundreds of megabits. China Telecom & Unicom, the monopoly landline providers, are fiercely resisting government demands they upgrade something like 100 million apartments from DSL to fiber. Fiber to the basement + G.hn might be an attractive alternative.

Read more...
 
Forbes: India has more net users than the United States

Their 243 million Internet connection estimate is high but the result is inevitable soon. I wouldn't count 2G smartphones with minimal data allowances who rarely if ever connect to the net, so my figure would be lower. However Indian 3G & 4G connections will soon pass the 315 million population of the U.S.. Fewer than 20M of these connections are fixed, mostly DSL. With fewer than 40M landlines in place, the Indian future is inevitably mobile.

    Newly elected Prime Minister Narendra Modi talks about the "broadband highway" and has promised almost $10B over five years for broadband, smart cities, wireless in Naxalite territory and more. Much of that money will be needed to prop up ailing government carriers BSNL & MTNL. There's an ongoing project to connect 250,000 villages that's two years behind schedule. Completion is now set for 2017, but the local loop is not included. 69% of Indians - 700M - still live in villages.

   New Minister Ravi Shankar Prasad is promising to meet the revised schedule. He may be the right man to overcome the inertia and corruption that has plagued so many Indian projects. Prasad prosecuted a corrupt Governor of Bihar state and put him in jail. 

Read more...
 
3.5 GHz Spectrum: Bells want it but WiFi clearly a better choice

China Mobile giving up after $3B, 4.3M small cells. 100 MHz of spectrum - enough to build 3 networks the size of Verizon - will become available in much of the U.S. The military has agreed to share the 3550-3650 MHz spectrum in much of the U.S. They primarily use the frequencies on the coasts so they don't need the MHz. The high frequencies have a short range so can't be used from towers but are great for WiFi and other small cells.

     Bill Smith, now of AT&T, taught me "The sooner the bit gets to a landline, the less it costs us." Every plan for the urban wireless future is based on WiFi, small cells or short range high frequency transmitters. You can usually have 10-30 small cells in the space covered by a tower, each reusing the same spectrum. The most efficient method turns out to be public WiFi rather than monopoly use, which is becoming obsolete.

Read more...
 
50 MHz of Spectrum Creates 500 megabit DSL

Today's 100 megabit DSL combined with today's gigabit WiFi, driving to a gigabit. 50 MHz of newly available spectrum may be diverted from sharing to private use, particularly LTE small cells the Bells probably won't build anyway. I'm headed to Colorado for the TPI conference where the 3.5 GHz spectrum will be a hot topic. I wrote this quickly because I wanted a dramatic example of what could be done with that 50 MHz.

How DSL + WiFi can get to hundreds of megabits and more. Nobody believes John Cioffi (yet) but it's easy to understand why it will work. Vectored DSL is delivering 50-100 megabits reliably in Europe. ~40M lines are on order at Deutsche Telekom and others. Vectoring is ready although not all the problems are solved.

Because DSL isn't shared, that means six apartments or nearby homes receive a total of 600 megabits (at 100 megabits/home.) Most places, 99% of the time the total demand will be less than 200 megabits. The result: 400-500 megabits will usually be available to share with neighbors. With WiFi now going into the gigabits (links), that can be shared between 5 - or 25 - families.

Read more...
 
5M U-Verse Homes on Hold

Biggest current U.S. broadband build virtually stopped  I haven't confirmed Paul de Sa's suggestion that the AT&T/DIRECTV deal is a cause. 

"Prior to the DTV deal, Project VIP (announced November 12, 2012) planned to expand AT&T's U-verse video footprint from 24.5m customer locations as of 4Q12 to 32.9m by 4Q15.6 At the end of 1Q14, we estimate the project was ~45% complete, with ~28.2m locations offered U-verse video." 

    I have double confirmation that the AT&T neighborhood DSLAM (FTTN build) is barely moving forward. I've asked AT&T if they'd release the actual deployment figures for the last three quarters to get a precise measure of what's going on. 

Read more...
 
Verizon ends unlimited service

Verizon until now hasn't imposed caps on some of their older customers, knowing many would leave if they did. Customers hate caps and limits are not required most places most of the time. They've changed that now, with what could be very modest restrictions that would have little impact on customers. There's not immoral - or neutrality breaking - about modest limits honestly related to actual congestion and costs.

    Fortunately, technology is improving so fast that today networks can be built for reasonable cost with truly minimal congestion. That should continue well into the 5G future and next decade despite steadily rising traffic. Very few traffic limits limits are necessary for network purposes today. The caps are for raising prices by collecting more from heavy users, not avoiding congestion.

    Verizon's actual proposal should be rejected by the FCC because it actually doesn't disclose what they are doing. How often would I be throttled? How much slower would I go? In nearly 4,000 words, Verizon doesn't answer those basic questions. Comcast's similar plan in times of congestion reduces some customer up to 30% but no more. 12 meg would fall to 7 meg when one of their engineers described it a while back. The fall would only be for 15 minutes and very rarely even several hours/month. That's so unobtrusive I don't think anyone has even noticed it in practice. I certainly would accept it, especially because the minimum speed on most Comcast networks has gone up.

    Nowhere in the 3,779 word document below does Verizon tell you how much your speed would be reduced or how often. 

Read more...
 
Huawei/HiSilicon Coming on Fast

Producing a vectored VDSL chip. Like Henry Ford steel & timber mills, Huawei seems to be following the path to "vertical integration." Their latest DSLAM board features vectored VDSL chips from their HiSilicon subsidiary. There is no announcement and I haven't been able to find any article about the chips in either English or (Google-translated) Chinese but I have multiple sources within the industry. HiSilicon also has a G.hn chip close to market.

    Huawei is the chosen supplier for vectoring to Telecom Italia & Fastweb/Swisscom and bids on nearly all other large contracts. The Italian deployment is struggling, however. TI & Fastweb agreed to link their vectored DSLAMs and unbundle local loops. Huawei promised they would be able to have separate terminals, possibly 50 meters apart, but still vector the local loops. As far as I can determine, they haven't delivered yet and the companies are scrambling.

    HiSilicon did $1.3B in chip sales in 2013. Growth is well into double digits although 90% of chips still go to the parent company. Digitimes believes HiSilicon, like Samsung, is actively looking for outside customers. R & D budget is well into the $hundreds of millions. HiSilicon chips are in Hewlett-Packard terminals. They have an 8 core cell phone application processor (ARM Big-Little) that is among the leaders. They are among the first with 300 megabit CAT 6 LTE, possibly beating Qualcomm to market. These are fabricated at 28 nanometers and soon below. TSMC, with massive orders from Apple, is essentially sold out at advanced nodes for the rest of the year. HiSilicon will probably be capacity-constrained.

    TSMC 16nm FinFET chips will be TSMC's best in 2015 and HiSilicon is the first vendor to commit.    

Read more...
 
Broadband Growth in One Chart

Africa 5%, U.S. & West ~1%; China huge. This chart from Point-Topic makes wireline clear in 2014. The West is almost saturated, espcially the U.S. China, Latin America and East Europe are going two or three times faster. Africa, with nearly no wires, has the most to grow.

     This would look different if wireless smartphones were included. The countries with few wires - India, Indonesia and most of Africa - are seeing fantastic growth in mobile broadband. By around 2017, there will be more Africans than Americans on the net, I've calculated from Cisco data.

Read more...
 
<< Start < Prev 1 2 Next > End >>

Page 1 of 2