Fast Net News, DSL Prime, DOCSIS Report

Turnbull’s party ahead 58-42. Malcolm Turnbull wants to shift the NBN from fiber to the home to fiber/DSL and cable modems, as I reported a year ago. http://bit.ly/163TNcf He’s the likely incoming Australian Communication Minister and will have the final say.

Supporting existing VDSL chip, extending bonding. Imran Hajimusa of Lantiq had a great demo of 4 HD channels streaming over WiFi a while back. They are now sampling and about to ship this improved WiFi, 3x3 MIMO 802.11n, a technology they purchased from Metalink. They've also upgraded their bonding capacity, which now can use VDSL profiles 17 & 30 for two line download speeds of 150-200 megabits over short loops. Their previous VDSL bonding could only do up to profile 8, for speeds under 100 megabits.  

  Telcos are screaming for reliable TV grade WiFi. I've heard claims as many as 50% of service calls are due to WiFi rather than DSL problems. When you carry your iPad to the bedroom upstairs, you don't want to lose the football game stream. Dirk Wieberneit of Lantiq reports raw WiFi speeds normally are fine, but don't often reach far enough into the home. The 3 antennas of their MIMO chips allow further reach with beamforming 

After restructuring, progress. Lantiq wowed the industry several years with the first public demonstrations of vectoring, but Broadcom has beaten them to market with production chips. Similarly, their WiFi 3x3 MIMO 802.11n chips, acquired from Metalink, had impressive demonstrations delivering four streams of video but haven’t made it to market. Artusi tells me both are now ready to go.

   Artusi has been publicly silent since he took over from Christian Wolff last year, waiting to be ready with product. He came with a mandate from owners Golden Gate Capital to rein in costs and has eliminated both facilities and some staff positions.

  “Munich has cold winters compared to Arizona, but I’m liking the job” Artusi tells me. “It’s been quite a challenge to create a lean and mean company, but that’s essential to succeed as a fabless chipmaker today.” Lantiq spun out of Infineon, itself once a giant division of Siemens.  “Carveouts are always challenging because of baggage from the prior owner. It’s very costly to restructure, especially in Europe. Golden Gate has given me the resources to reduce net debt to zero.”

  Artusi is shifting the company focus from the chips themselves to becoming “a broadband company.” Lantiq has a large portfolio of packet processing software and firmwire Dan is leveraging into other markets. They have a close relationship with Altair Semi for LTE gateways for fixed wireless, likely soon a massive growth market in Africa, India and Indonesia. They are also working with Qualcomm on 802.11ac, the rapidly emerging next WiFI standard.

  Lantiq is also on track with a chip for the potentially popular “fiber to the distribution point,” a way to raise DSL speeds well over 200 megabits. Locally powered nodes with reach of perhaps 100 meters are tempting to carriers like Swisscom, although the market remains unproven.

  Lantiq's a private company, not required to release financial details. The below Standard and Poor's release has data I otherwise haven't been able to obtain. I missed it back in December and Artusi has added more details above.

Overview

-- German semiconductor designer Lantiq Beteiligungs-GmbH & Co. KG and its lenders have recently signed an amendment to the credit agreement for its outstanding loans.

-- Furthermore, Lantiq's owner, private equity company Golden Gate Capital, has supported the group with an additional $81 million in capital in the current quarter.

OFCOM needs to sort out competition issues. BT is passing 19M premises with upgraded connections, nearly all fiber/VDSL. Josh Taylor now reports they will begin vectoring later this year. BT’s Mike Galvin tells him “"We see vectoring as one of the cornerstone technologies going forward.” http://zd.net/16ZOA2T

    BT uses ECI and Huawei gear, both of whom made early commitments to vector. ECI told me in 2012 their equipment was “vector-ready” although customers hadn’t deployed the service yet. Huawei has won orders for several million vectored lines from Swisscom and Fastweb, also not yet in production.

   Britain has some of the most active retail competition in the world, something Ed Richards at OFCOM doesn’t want to compromise. Vectoring has been delayed although they recently received presentations on the subject. ASSIA presented the possibility of “multi-tenant” management of a vectored deployment, a constructive step if full unbundling proves impractical. “Bitstream unbundling” is technically easy and the cheapest way to go, but the incumbents have been imposing upside-down pricing that defeats the point of a high speed network.

   DSL local gear is generally non-blocking, which means the cost to deliver the full speed of the line is literally the same as the cost of a throttled connection. So a cost based unbundled price would be the same (? $20) whether the customer takes 7 meg or 70 meg service. Backhaul and transit add only a modest additional cost to competitors with enough scale to have their own fiber. Sky and TalkTalk have their own fiber to exchanges serving over 70% of the British population. Either could offer full speed for say $35, undercutting BT’s plan to charge much more for the higher speeds.

Verizon hikes ~20%, 55% higher than 2007. Pressed into a corner by Connecticut Senator Blumenthal, Jules promised to work with the Senator on these non-competitive rate increases. JG may be offering just a “politician’s promise,” of course, but Blumenthal as Connecticut Attorney General proved he can make things happen.

   Karl Bode at DSL Reports broke the story of a $5 Verizon increase on existing DSL customers. http://bit.ly/Z0NbB9, most of whom were paying $20-35. That’s 15% to 25%. New customers at Verizon are in even worse shape. After the introductory period, Verizon now wants ~$50 + local taxes for one meg, adding a requirement you take a phone package. It’s at least $60/month for anything faster than one meg. For FiOS, the cheapest service they offer is an amazing $84.95 - and in some FiOS areas they won't sell DSL any more. Bode also reports AT&T increases that appear to be 10-15% but details vary. http://bit.ly/ZB8sSR Neither Cecilia Kang at WashPo nor any other DC reporter has yet picked up the AT&T story, so DSLR was almost certainly the inspiration for the question.

  Here in New York, minimum broadband costs $50 to $85 from Verizon and $55 from Cablevision, although Time Warner does have a 1 meg data service for $30 in some areas. There are some great deals for the first 12 months, but after that the $130-160/month triple play is the main offering. That compares to $37 for 5-15 megabit down + a phone line at Sky in the UK or less than $50 for triple play at any of the five largest French carriers.

  The Senator challenged JG. “AT&T announced a rate hike on their DSL service. This comes just a week after Verizon announced an almost identical rate hike. Both of them trouble me. We ought to pay closer attention.” JG tried to change the subject, saying “We’ve seen much better trends in mobile competition over the last couple of years than we’ve seen before.” With the Senator glaring, he added “I’m troubled by rate increases that aren’t based on competitive factors.” Then Jules said he’d work with Blumenthal investigating prices.

   Julius claims his greatest achievement is “broadband, broadband, broadband.” Deployment has almost stopped on his watch, adding only one percentage point. When Obama took over at the beginning of 2009 telco coverage was 84% and cable 96%, by FCC figures. http://fcc.us/ZBfd7h The latest figures are 85% and 97%, despite $billions in broadband stimulus and major merger promises by several carriers. http://bit.ly/WIHJsz Prices have gone up 10-20% on the cable side, even more on the telco side.

   Legacy?

Belgium intends to add vectoring piecemeal while leaving many older lines in place. The vectoring standard was designed to make that possible but effective management tools need to be developed. Belgacom was the first to receive regulator permission to vector because they didn’t have to resolve the problems of unbundling. There are very few unbundled lines in the country. They intend to spend 2013 continuing trials and making sure everything will run smoothly for 2014.

Alcatel had production equipment for the field in early 2012. The early trials at Swisscom, Belgacom and in the labs at nearly all the major telcos were very encouraging. By September 2012, the DSL Tsunami was apparent http://bit.ly/U79yWR. Deutsche Telecom and many others were convinced vectored DSL was the way to go because it was so much cheaper than fiber.

The delay may have been inevitable but is very disappointing.

Wall Street Journal discovers Vermont is the fastest state. http://on.wsj.com/1848tWZ Michel Guite at Vermontel charges only $35 for a true gigabit, upstream and down. That’s not a typo: funding from the stimulus means VTel can sell a gig of active Ethernet for a DSL price. All customers get the full gigabit as their basic service. Almost 1,000 homes are connected, with another 15,000 in the coming months. I’m proud to play a modest consulting role.

   “Is there any network anywhere in the world doing something better we should learn from? Maybe in Hong Kong or Singapore?” Michel asked me two years ago. He didn’t quite believe me when I replied “no,” but judge for yourself:

• Each home has dedicated fiber running active Ethernet at a gigabit both upstream and down. If requested, VTel has the capability of upgrading that to 10 gigabits or even 100 gigabits. No orders like that are likely for many years, however.

• In the home, VTel uses a model of Actiontec router that actualy can deliver the gigabit. They discovered in testing that most home “gigabit” routers actually topped out at $300.

• The Calix gear terminating the fiber in each exchange has a switching fabric designed to be non-blocking under essentially every likely traffic load.

• VTel connects to major exchange points in both Boston and New York for redundancy. If the Boston bombings had shut down the net there, VTel could instantly switch to New York.

    The cost to build the network in a mostly rural territory is $3,000 to $5,000 per home. The promise by VTel to pass most of the subsidy to consumers in low prices was surely one reason they were awarded the funds.

   Unlike most U.S. networks, all customers get the maximum speed of the network. The increased bandwidth demand is real but not excessive; no home has installed 20 HD TVs just because they have a connection fast enough to feed them. Fiber operating costs are predictably lower than copper while bandwidth is now extremely cheap. The result is the operating costs of the network offering everyone a gigabit may well be lower than a 10 megabit DSL or cable network.

   “I’m moving to Vermont,” was a comment on one of the news stories.

Many lines do not significantly improve. As predicted, ARCEP is allowing French networks to upgrade to VDSL2. Free.fr has been using VDSL chips in DSLAMs and the Freebox for a while, so instantly could go into testing. One lucky customer tested  “Gross rate: 68420 kbp/s - 17852 kbp/s.” (Freenews) A second customer at 1500 meters actually saw a (small) drop in speed. By ARCEP estimate, only a modest minority of customers will benefit. Update 5/22 FT's Yves Parfait confirms it will be FTTH to 60% of France by 2020. Via CommDay

Builds custom chassis to show off 100 meg on every port. “Customers probably will delay volume deployments until late in 2014. Our chip is production ready and will be available well before that,” Kourosh Amiri of Ikanos tells me. It’s an impressive chip, designed for node scale vectoring of 384 ports. It does full cancellation of noise in all tones, which Ikanos claims is substantially more effective than the competition’s “partial cancellation.” A ten gigabit serdes is built in to support the high speed interconnect needed for this kind of performance.  

    (The picture shows Ikanos' reference chassis supporting 384 ports at 100 megabits.)

    Contrary to Deutsche Telekom’s claims that it’s “impossible” to unbundle vectored VDSL, Ikanos is perfectly comfortable with two ISPs sharing the binder. “As long as the two DSLAMs are within about 50 meters, we can communicate between them and cancel noise on all lines,” adds Amiri. Telecom Italia and Fastweb have an official memorandum of understanding they will do just that. http://bit.ly/WR6tug Each is passing several million homes with fiber/DSL, often in the same node. Huawei, #2 DSLAM vendor to Alcatel, is understood to have the contract and presumably will choose Ikanos chips.

   Ikanos is serious about interoperability. “If you manufacture VDSL chips, bring a board to our labs and we’ll be glad to test with you.” They claim they are working closely with their lead competitor and expect few problems in interop.

   Promises, promises I’ve been hearing from everyone in VDSL for a decade, so I’ll remain skeptical of everyone’s claims until they are proven in the field.

 Q4 Cable +536K Telcos +72K. In the slow growing U.S. market, cable now has ~58%, a small change from the 55% five years ago. Cable is far ahead in net adds, clobbering telcos in ~30M homes with DSL that hasn't been upgraded. Telco price increases also hurt, with Verizon raising the minimum price of DSL to ~$50 and FiOS to $85. Upgraded DSL - 30M U-Verse, 7M CenturyQwest - is doing just fine so far, with few consumers willing to pay the high prices for more than 10-15 megabits. The U.S. market has three sectors. Fiber, mostly Verizon FiOS, doing well, partly aided by new customer deals advertised as lower as $85. Cablevision is losing to FiOS. 30-40M homes of older DSL, generally 3-6 megabits, is getting killed. About 40M homes - upgraded DSL/FTTN - is actually doing fine, something DC policymakers generally don't realize. Upgraded DSL, at least so far, is doing fine against cable in most market. Telus and Bell Canada most quarters beat Rogers and Shaw; British Telecom is generally beating Virginmedia. 

    Actual figures for the year include cable +2.4M to 46.8M. Telcos added a measly 300K to 34.6M. Comcast added 1.2M for the year, with over 100,000 poor families benefiting from the $10 offer. AT&T actually went -37K. Batya Levi of UBS actually predicted this decline. In Q4, Comcast added 342K, TWC 89K and  Charter 61K. AT&T was +2K as they began to resume U-Verse build and Verizon +72K. Windstream was - 2K despite new areas covered with stimulus money. Windstream and Frontier have generally not upgraded DSL and poor broadband sales will likely drag on the stock going forward. 

   All these figures are from Leichtman Research, whose data I have checked in the past with company financial filings and found reliable. 5-10% of U.S. broadband is with smaller companies, not included in these figures. There's no reason to think that would change any trends. 

From $2.1 billion sales in 2000 to just $135 million in 2012 Once a great power in voice and DSL modems, Conexant “is fighting for survival.” (Junko Yoshida) In 2003, Conexant bought Globespan for $860M and aspired to dominate the DSL chip market. Late products, a massive royalty suit won by Texas Instruments, and generally declining prices for DSL chips led to a near immediate sales shortfall of $100M and Dwight Decker took over from Armando Geday as CEO. The remains of their DSL business were sold to Ikanos for $54M in 2009 and continues.

George Soros’ Quantum Fund has put up $15M in DIP financing to give the company a chance at recovery. Quantum was the sole secured creditor and took a writeoff on part of their $197M credit facility. Andy Rappaport of August Capital and John Knoll of Golden Gate bought Conexant for $200M less than two years ago. They now signed the bankruptcy filling http://bit.ly/WjERzN alongside CEO Sailesh Chittipeddi Kodak and Silterra took losses over $1M. Details of the prepackaged bankruptcy weren’t disclosed but presumably Quantum owns nearly all the surviving equity.

Small chipmakers across the world are struggling as the cost of developing a chip has risen well into the tens of millions. Delayed entry or failing to find a market becomes incredibly costly. Venture and Private Equity for chip companies in the U.S. has virtually dried up.

Junko Yoshida’s report on the history and prospects of the company is worth a click. http://bit.ly/YoVwPk. One comment there is

“Conexant had the best engineers, advanced products, and strong financial lineup. Dwight Decker,Dan Artusi, Scott Mercer, Christian Scherp, Sailesh Chittipeddi and its cronies effectively wiped out everything with their financial expertise. Chapter 11 is the best solution for Conexant and put it out of misery. CNXT shareholders and its employees (ex and current) are the only victims, gone with their saving, trust, and retirement. RIP”

Not just Alcatel’s game anymore. 80% of Swiss homes will be offered 80 megabits and up from fiber to their homes or street cabinets by 2020. The street cabinets will come from Huawei, the first big announced win for Huawei in vectoring. Huawei, still #2 to Alcatel in DSL, has put major R & D into vectored DSL and scored a number of firsts in lab tests. This is the first confirmed customer for the equipment.
     Swisscom is already working with several cities and/or power companies to deliver fiber to the home. It’s a nearly unique deployment, with four strands going to each apartment for possible future competition. Outside those cities, they’ve now chosen to run fiber within 200 meters and then DSL. That’s closer to homes than the similar builds in Germany and Britain. While I think of DT’s build as 70+ megabits to most, Swisscom’s shorter distances will deliver 100 megabits to many more. They are calling it FTTS, fiber to the street, instead of FTTC, fiber to the curb or cabinet, because they are deploying deeper. Swisscom is in trials in Charrat (Valais), Grandfontaine (Jura) and Flerden (Grisons) but the major rollout is not scheduled until late 2013 or possibly 2014.
     Swisscom’s results have been among the best among European telcos, holding profits and dividends comparatively stable. They’ve actually been winning customers from UPC cable. Prices are somewhat high - in the 30’s - but the Swisscom service has been excellent and customers have responded. They are increasing investment for higher speeds to keep customers happy.
    They’ve had to take a billion writeoff on the investment in Italy’s Fastweb but they are now upgrading Fastweb as well. They’ll be doing 3.5M lines of fiber/DSL very rapidly, inspiring Italy to propose 7M lines vectored as well. Huawei has the inside track for that contract, especially because Fastweb and Telecom Italia are committed to working together to minimize interference.
    Alcatel continues to maintain such competitive deployments are impossible, echoing the desires of their hoped for client Deutsche Telekom to knock out competition. Huawei instead is working to overcome technical obstacles through “node-scale” vectoring and that presumably was a major advantage winning this contract.
    Swisscom promises some homes will see DSL speeds up to 400 megabits in future years and are actively involved in testing G.Fast.

Free.fr creates crisis by blocking Google ads. Stéphane Richard, CEO of France Telecom, broke the story "There are financial flows between Google and Orange, as well as other operators, in particular Deutsche Telekom. ... there is an overall balance of power. There are areas where Google can not do without us, for example in Africa" (Quotes from Les Echoes in Google translation.) Numerama previously reported that FT and Google had a deal without naming a source.

Presumably, Google’s position was “we don’t pay for termination but let’s see what we can work out.” Google needs fiber throughout the world and France Telecom owns a cable to Africa. Google needs space and facilities in peering points controlled by both Deutsche Telekom and France Telecom. Deutsche Telekom is moving heavily into server hosting.

This is "sender pays," a termination charge, that was opposed strongly by the U.S. and Europe at WCIT as deeply destructive of the Internet. It's a clear violation of net neutrality as defined by those who created the concept. The French government is unsure of how to handle this. On one hand, net neutrality is government policy. On the other hand, they support French telcos who are fighting to collect from mostly foreign content companies such as Google. The issue are fought publicly on policy principles. Behind the scenes, the discussion is "content" is American and "carriers" are European. The Europeans are happy to collect from America, usually ignoring the fact most of the money would go to a few giant telcos.

It’s likely that Google and FT made a deal with Google paying FT for something other than transit. FT got money, Google didn’t set a precedent, and people are happy for now.

Free, trying to get some, went nuclear. Xavi blocked Google (and other) Ads An immediate rebellion by websites afraid of going broke without Google ads got the government involved. Free.fr has been trying to collect from video sites such as Google and Dailymotion for years. Not long ago, he refused ordinary peering with Google. He demanded to be paid. As YouTube traffic predictably grew, congestion grew. As much as half the time, Free’s users couldn’t get their videos.

This was Google’s “nuclear option,” effectively cutting service to Free’s customers if Xavi didn’t give in. It was starting to hurt, especially after consumer publication Que Choisir did a survey that made clear the problem was extreme and asked the government to do something about it. Prominent Internet analyst Taylor Reynolds tweeted he was moving his broadband away from Free.