“11n is cheaper than 11g but can run three times as fast,” claims Bomin Wang, President of TrendChip, which has just launched an ADSL gateway design with 11n. "In a competitive market where the three-network convergence policy has been put in place, China's telecom operators are considering 11n as an essential specification in their new tenders to allow users to enjoy the convenience of high bandwidth and IPTV anywhere, anytime.”
11n chipmaker Quantenna has just received another $15M from Swisscom and VCs to develop their 4x4 MIMO chip. The prototypes are delivering 100 megabits over 100 feet in their testing, enough for wireless HD TV connections. Netgear has announced a product based on the chip. DSL folks know Quantenna's founder, Behrooz Rezvani, who previously was at Ikanos. The Lantiq-Metalink 11n chips have a nice demo of 4 HD signals carried across the demo room without any dropouts.
Jennie hates wires around our home, a problem these chips hope to solve.
Customers hate bandwidth caps, Verizon's market research shows, so "No Bandwidth Caps. Period!" is the highlight of their latest DSL ad campaign. There's no technical or cost reason a cap is needed on any large, wireline network; it's a way to block competitive video and efficiently raise prices.
Internet transit is down to $2/megabit at major peering points and costs of deploying broadband continue to drop steadily. Bandwidth growth continues, which Bill Smith of AT&T tells an FCC workshop has slightly raised his cost per customer. That's important information for DSL Prime readers whose job is to manage network costs, but I'd estimate any increase in the last year is less than 1% of the price of the service. The usual industry figure is the total bandwidth cost is about $1/month, 2-4% of the price charged.
Congressman Eric Massa became a hero to voters by fighting a Time Warner bandwidth cap. Fighting caps – and unreasonable prices – is a natural move for politicians and regulators worldwide who want public support. There's nothing immoral or even fattening about a cap that's reasonably related to costs. Iheard from my consumer-favoring friends for writing about a reasonable one, http://fastnetnews.com/docsisreport/163-c/53-comcasts-fair-250-gig-bandwidth-cap . But it turns out so few people are affected by the Comcast cap it saves very little money.
Update March 17: Mostly on target. Microsoft/Dell and the carriers have held off on their discount plans. Original The U.S. broadband plan accomplishes very little for affordability, quality, speed, or availability of broadband in the U.S., although it has other important achievements I describe below. In particular The “100 megabits to 100 million homes” is right on target for what will be achieved by 2015 without any broadband plan. (FCC/Columbia CITI November 2009). Based on cable company's official statements, I reported in August 2009 102 million homes would have 100 megabit capable DOCSIS by around 2013. http://bit.ly/c7jMuJ Fewer than 4% of U.S. homes that can only get satellite (“unserved”) will be reached because of the plan. It's more likely only 1-2% of homes will be upgraded. Broadband prices are more likely to increase than decrease because of the plan, especially if a multi-billion dollar Internet tax is included. There's nothing wrong with taxing the Internet like anything else, but this “fee” goes to the shareholders and bondholders of phone companies, not re-opening closed hospitals. Only a small fraction of the poor will get substantial help according to the best information I can find. In particular, the much-touted cable A+ plan provides “back of the bus broadband” throttled to a tenth the normal speed, available to less than one in five of the poor, and actually more expensive than Verizon's recent promotion. AT&T has offered similar, but I don't know if it's included. Since nearly all mobile phones will include broadband in a few years and far more than 90% of families have a mobile phone, the 90% take rate in 2020 would almost certainly be achieved without the plan, probably several years earlier.
Mike Gulett of Ikanos told me they make the best DSL chips when I visited. A few hours later, Imran Hajimusa of Lantiq (formerly part of Infineon) said something very similar. I'm sure I would have heard the same if I had stopped by Broadcom as well. Not being an engineer with a test lab, I'm not qualified to judge. Any chipmaker who's survived the tough DSL market obviously has an outstanding product as far as I'm concerned. Both had some impressive features to demonstrate.
Ikanos showed me how their rate-adjustment on the fly minimizes the dropouts and retrains that are the bane of IPTV carriers. Tom, Gavin, and the rest did a remarkable job back around 1993 defining the DSL standards and it's amazing how little problem we have with interference. Back in 2000, some competent engineers believed DSL would crash when the networks became loaded, but that clearly hasn't happened.
But the occasional interference problem is annoying for IPTV, especially if the dropout happens at a crucial moment in a football game.
34% of West Virginia homes can't get DSL, one of the lowest deployments in the developed world, according to the new FCC stats. Jay Rockefeller should have called in Verizon's lobbyists years ago and told them they'd never get a bill through the Senate if they didn't bring his state up to standards. Mountainous Wales is 99% covered and most rural areas of Europe are over 90%, so the argument this is because of impractically high cost comes out of the rear end of a male cow. The smallest rural carriers in the U.S. reach all but 8%, proving what's practical in rural areas. (All states below). 37% of New Hampshire, 31% of Virginia, 28% of Vermont, 27% of Maine, 26% of Michigan, 24% of Maryland and Mississippi, 22% of Arkansas, 21% of Alaska (despite massive subsidies) and New York all can't get DSL. Cable generally is better, but the FCC data is limited and doesn't include homes not passed by cable for TV. (Phone lines pass all homes but cable only about 96%.)
This data makes clear the largest problem in DSL deployment is where Verizon virtually stopped all upgrades around 2002 when they decided to sell
Congressman Jose Serrano is bringing FCC Chair Genachowski to Per Scholas, a community group in the Bronx 1 p.m. Monday. I've been considering what's the right question to focus on some real issues. Update after the event: Despite saying there would be a Q & A, they "ran out of time" and I didn't get to ask Julius the questions. Congressman Serrano did give me a few minutes, and that resulted in my item http://fastnetnews.com/docsisreport/163-c/2541-congressman-serrano-low-speed-lifeline-qabsolutely-unacceptableq End update “Is Back of the Bus broadband acceptable for the poor?” I'll ask the Congressman. There's a good chance the “lifeline broadband” will look like the NCTA Adoption Plus suggestion that was strongly backed by Jim Cicconi of AT&T. That's carefully designed to offer a fig leaf to the broadband plan while not serously threatening the cash cow video business. Only 1/6th of the poor would be eligible for a short term Internet connection at speeds too low for standard video. They want a pile of government money as part of the deal. Nate Anderson of Ars - who constantly outreports the Washington Post – noticed the price they asked was higher than the current promotions at Verizon and AT&T. For Genachowski, I'll probably ask “As we move to objective, data-driven policy, just what are the results we are seeing. In the first Obama year, how much did the basic price of broadband and telephone service go up or down? How many of the “unserved” broadband homes were newly offered service?” Obama's prime FCC goal to to bring broadband to everyone and make it affordable, but I haven't seen anything like an objective measures of results. My best information is that the last year has been the worst since 1998 in extending broadband to the “unserved”. Several large carriers have raised, not lowered their prices. AT&T California just raised basic phone prices 22% (LA Times). If I get a second question, it will be “what percent of the broadband lifeline you've supported will directly help the poor and what percent will go to the bottom line of the carriers.” Wall Street's Craig Moffett estimates that 10 megabit broadband's marginal cost is about $8 and contribution margins 80%. Free Press believes paying more than $10 for lifeline is a carrier subsidy; I'd use a higher figure ($15 or so, 50% margins) but am horrified by the general assumption that subsidies will be much higher and the speed probably crippled. I wrote http://fastnetnews.com/stim/179-s/2188-save-half-on-broadband-subsidies-dont-pay-retail-for-a-million-lines on the topic, which resonated with several in D.C. There's nothing wrong with making a profit helping the poor, but the main benefits of a program for the poor should not go to Brian Roberts and Randall Stephenson's shareholders. Cui Bono, Julius of the Supreme Court? Julius is a great guy who's done a remarkable job improving morale at the FCC but the real test is results.
Seidenberg's on the way out and Verizon is changing. They have now canceled planned FiOS deployments for all new territories such as Alexandria, Virginia. According to Bryant Ruiz Switzky in the Washington Business Journal, Verizon is "suspending Fios franchise expansion nationwide." They are "indefinitely postponing" building Alexandria after telling the city they would begin construction several months ago. Alexandria is one of the richest suburbs in the world and a natural part of the network with a lower than average likely construction cost. Verizon "will now focus on installing its network and gaining market share within the areas where it already has agreements." Bostonians and 10M other Verizon customers are apparently screwed.
Verizon has buildout commitments to New York and other cities that will keep some crews working, but had already suggested they might cut FiOS builds by 2/3rds in 2011. This is now a further cutback, canceling areas that for years they had been promising to serve. Verizon's Harry Mitchell sends their perspective. "The bottom line is that Verizon said in 2004 we’d build to pass about 18 million homes by year-end 2010, and we’re on track to do that with the franchises we currently have. Of course, we will also meet any buildout commitments we made in individual jurisdictions beyond 2010."
Ivan in an investor call suggested one reason they may be cutting their investment: the broadband plan and stimulus are reducing company spending. So Seidenberg suggested he might ask for government money, and the broadband plan has many "incentives" for him to spend less company money. Blair should take this as a signal to yank any offers to pay telcos to upgrade broadband where it already is available from cablecos or others. Smelling government money, they are cutting back their own investment and then demanding the government pay them instead. 2009 was almost certainly the worst year in a decade for expanding broadband in the U.S. Comany after company cancelled firm plans waiting for the government to pay them for what they intended to do without subsidy.
Over the last few years it's become apparent that Seidenberg's personal desire to beat the competition has been the primary reason the U.S. is not further behind. FiOS is the largest new network built in the Western world. Cable's DOCSIS 3.0 was developed as a response to FiOS. Brian Roberts of Comcast tells the story of lookng Ivan in the eyes, deciding he was going ahead, and then giving orders to his team and Cablelabs to go full steam ahead on DOCSIS.
Euille said Alexandria citizens are “clamoring” for Fios and often don’t understand why it isn’t available. “I’m sure the citizens are just as disheartened by this outcome as I am,” he said, adding that the city will look at alternatives.(Switzky) Alexandria can't afford for the neighboring towns to have better Internet service, so building their own network is the obvious step.
AT&T's customers have long hated the unfortunately named B-52 cabinets and now British Telecom is facing civil disobedience over the ugly 6 foot by 4 foot neighborhood boxes they are deploying. St. Albans said they are totally unacceptable in a "conservation district" and Sandridge parish councillor Chris Hackett has refused to let them dig foundations. Herts 24 now reports the police hear from a local man "I told them I'm not having this. They are trying to obstruct me from getting in and out of my drive. And anyway I don't intend to look at that eyesore from my house. They can either move it somewhere else or I will." http://bit.ly/cIhPct, including picture
Fortunately, new generation field cabinets can be half the size as chip densities go up. A nice unit from Ericsson does away with the fan, and quiet fans are available for any enclosure. It's time for BT to find a smaller, quieter box. It was a brit, after all, who promised "we shall defend our Island, whatever the cost may be, we shall fight on the beaches, we shall fight on the landing grounds, we shall fight in the fields and in the streets, we shall fight in the hills; we shall never surrender."
35M or so homes - about 100M people - don't take broadband in the U.S. You could subsidize 15-30M of them for 1/5th of the money spent each year on USF/ICC. Each family enrolled would pay say $7/month. The companies would make a normal profit, say 40% EBITDA. It requires no political tricks, and the economics are derived from top Wall Street analysts. It would be a very good thing, far more important than anything I've heard from the broadband plan. It is probably politically impossible. The FCC is hard to change.
I begin with the real economics of broadband. The marginal cost of adding a customer to any large broadband network is about $8 by wall street and my estimates. $8 provides 3 to 10 megabit service; reducing the speed to "back of the bus" speeds saves well under $1 (large carrier).
$8: The marginal cost/month of broadband in 85+% of U.S., essentially all large carriers. The $8 figure comes from leading Wall Street analyst Craig Moffett, presumably direct from internal numbers at the big cablecos. My own research confirms it, although I speak of a range of $5-12 across the developed world. I have factchecked it with AT&T as well as a senior cabler and a large RLEC, as well as off the record with many CTO types, etc. I come to that figure by adding up the cost of bandwidth, customer support, modems, and the other main inputs required to add a customer to an existing network.
$15 A reasonable price for the government to pay when buying millions of lines for lifeline service. That provides teh companies with a reasonable profit, perhaps 40% EBITDA. It's ridiculous to pay retain for millions of lines. We know $15 is reasonable because both Verizon and AT&T charged $15 price for basic broadband until recently and always said they. were profitable. Many European broadband prices are at that level when bought as a bundle. The cost and profit figures are on target for all the larger carriers.
$7 Customer pays (a bargain)
$8 Subsidy/month to get to the $15 total.
< $100 Subsidy per family per year
10,000 families served per million of subsidt. 10,000,000 per $B.
20-30B homes for $3B/year. That's a lot of money to you or me, but less than 20% of the current USF/ICC total. Verizon or AT&T annual cash flow, etc. It's practical to identify $billions in waste in USF/ICC that can cover it over time.
Important note: 5-10% of the U.S. is rural or otherwise has higher costs, which is why elsewhere I point to reducing high rural backhaul costs etc. as critical.
Dell'oro, one of the best analyst firms, reports a 12% decline in access equipment sales in 2009, which they think will only be partially reversed in 2010. They see the future as determined by upgrades, not new deployments, with fiber in the lead. They see the market split about evenly between GPON and EPON, cable CMTS upgrades as everyhing goes digital, and disappointing sales for DSL.
My take going forward sees China continuing as the dominant market. Their 12M/year DSL growth has been the driver for several years, but PON will take a large share of the Chinese future. There are 20M lines of fiber on order in China, with GPON making inroads at China Mobile and perhaps China Telecom. The norm is rapidly becoming fiber at least to the basement. Japan is 85% fiber already, limiting room for growth, and India will be predominatnly wireless because their wireline network is miniscule in (35M line) for a country of over a billion people.
Brazil has some grand plans, with Lula speaking of a possible $8B investment in fiber. But the developed world - where 60-80% of homes are already connected - has little room for growth.
For the last decade, every year someone quotable says Moore's law is dead. It just ain't so. Chip innovation is continuing, with Intel/Micron's new 25 nm flash memory pointing the way not just to smaller iPod nanos but predictable price/performance improvements in telecom gear. Most telecom gear is produced at 65 or 85 nanometers, with only a few chips, mostly for wireless, at 45 nanometers. The latest 25 nanometer process from Intel and Micron will over the next few years extend from NAND memory chips to comm chips, cutting typical space and power demands by 50-80%. It's hard to imagine a feature size so small that over 30 billion transistors can fit in a half inch square, but that's what we talking about here. About 50 atoms across by another measure. Altera promies even more demanding chips, FPGA's in 28nm this summer. Digitimes says they will be manufactured at TSMC, where many of the communications chips are also made,
Underestimating the power of Moore's Law leads to many of the stupid things you hear around D.C., like that the exaflood will destroy networks. In practice, the cost of carrying bits has gone down since 2002 at about the rate demand for bandwidth has gone up. That's highly unlikely to change for the next five years even if video causes 35% increases in traffic every year - as it probably will. An authoritative source - Bill Smith of AT&T - told an FCC workshop that lately the growth in demand has slightly outpaced the cost decline. That's important to DSL Prime readers whose job is managiing the cost of a huge network. But a modest change - say 5% - has minimal impact on the overall cost of a $20-50/month service. The working number for bandwidth costs is about $1/month/customer at a large wired carrier.