Article Info

AT&T Project Lightspeed and the Jedi Mind Trick

The Death Star is looking vulnerable these days.

Obi Wan MindTrickAT&T (T ) COO Randall Stephenson, speaking yesterday at Bank of America’s 2006 Media, Entertainment and Telecommunications conference attempted to exercise his marketing skills with a poor attempt at a Jedi Mind Trick.

From IP Democracy:

In terms of Lightspeed’s ability to push through hundreds of video channels, including high-def video “we’re not constrained by bandwidth. You’re not constrained by the size of the pipe anymore,” Stephenson said, referring to the switched-video capacity of the network which delivers only one service to a single customer at a time.

“In the foreseeable future, having a 15 Mbps Internet capability is irrelevant because the backbone doesn’t transport at those speeds” he told the conference attendees. Stephenson said that AT&T’s field tests have shown ‘no discernable difference’ between AT&T’s 1.5 Mbps service and Comcast’s 6 Mbps because the problem is not in the last mile but in the backbone.

This is total nonsense. Verizon delivers 15mbs consistently to my home. This is the latest of many odd and obtuse statements from AT&T which, to me, are begining to sound more and more like excuses rather than logical explanations. All of these statements clearly indicate to me that Project Lightspeed is not going very well.

Why this man is being forced to make a fool of himself?

AT&T (SBC originated the program) made a big gamble with Project Lightspeed when it decided not to build fiber directly to the home. Instead, AT&T brings fiber ‘almost’ to your home, then spans the last 1000 meters or less using conventional DSL technology to deliver next generation video and data services. The only major benefit to this approach, as opposed to using FTTH, is it eliminates the need to install new cabling to each and every home, saving install costs. AT&T’s approach is 1/3 to 1/2 the cost of putting fiber directly into the home.

The downsides could fill an entire page:

  • Inconsistent performance of DSL makes it hard to predict actual bandwidth after installation
  • DSL technology limited to 20-30Mbs
  • Requires a set top box for each TV in the home. Cable ready TV’s can’t tune IPTV
  • More expensive and complex IPTV head end hardware and software. Verizon FiOS and NTT’s approach copy Cable’s logical architecture exactly.
  • Requires more electronics in the field, more operational expenses
  • Requires use of MPEG-4 video compression, which doesn’t look as good as MPEG-2
  • Limited number of video channels can be viewed in the home at one time. Only one HD signal can be viewed in household at any given point

The theory is SBC has more underground wiring, where Verizon (VZ) is more aerial. Installing new fiber aerially is much cheaper than trenching- so some say Verizon had the luxury of using FTTH given the nature their infrastructure. AT&T is using FTTH for greenfield installs as it is now common knowledge that installing FTTH in new home developments is cheaper than copper when you take into consideration future operational expenses.

In short, AT&T is deploying a very complex architecture with major limitations in the interest of saving money. The major problem with this approach is it offers nothing better than what the incumbent cablecos can provide. Cable’s broadband is faster. Cable customers don’t need to worry about how many channels a household is watching simultaneously. Verizon’s approach delivers a user experience equivalent to cable with the ability to radically surpass it by deploying new bandwidth hungry applications as they emerge.

The only defense offered by AT&T is “IPTV is better”, and yes they are right – it is. IPTV gives you wide flexibility in content distrubution. But AT&T is using IPTV because they have to, not because they want to. The drinking straw bandwidth provided by DSL forces them to broadcast a few specific channels at a time. IPTV isn’t a technological strength, it is a technology deployed in order to make up for the inherent weakness of their DSL based approach. Nothing precludes building an IPTV infrastructure over fiber, or even DOCSIS 3.0 based cable modems.

AT&T is now preparing to trial the service. Assuming they can successfully navigate all of the above pitfalls, they will emerge with a product that is marginally better than the one offered by cable, and have done little to address the long term requirement of putting high bandwidth infrastructure into the home.

Verizon is making a major investment in FTTH, but I would bet the fiber they are installing will be in use for 100 years – they can just upgrade the equipment at each end. The complex hardware and electronics AT&T are deploying ‘near’ your home will junked within 10 years when they too are forced to put fiber through the last mile. Sure, they are saving money now – but the investment will have a much shorter timeline.

Hopefully, AT&T will stop the charade that their short-sighted investment is superior to FTTH and stop wasting capital on what is clearly a stopgap measure. When the Jedi Mind Tricks eventually wear off in the investment community the markets will extract their pound of flesh.

Nothing is so fatiguing as the eternal hanging on of an uncompleted task. – William James


Comments are disallowed for this post.

  1. Right on target. FTTN is really DSL and a press release, so it’s amazing how many “policy people” think SBC is doing a fiber build.

    Currently, SBC capex is only 70% of depreciation.


    Posted by Dave burstein | March 31, 2006, 1:17 AM
  2. MPEG-2 offers better quality than MPEG-4? Wow…. please read more about MPEG-4/H.264 before writing your next essay.



    Posted by john | March 31, 2006, 10:20 AM
  3. What about WiMax? S. Korea gets ~30 Mbps almost everywhere in the country using an older standard. The US needs to wake up to this and the cost savings (to business and personal users). If not our already weakened competitive position will be further weakened.


    Posted by Al | March 31, 2006, 12:49 PM
  4. John,

    MPEG 4 has many enhancements to improved coding that will deal with compression better. The higher compression makes MPEG4 less quality than MPEG2 just like MPEG2has less of the content artifacts than Analog.

    I think the big issue here is that MPEG2 HD and MPEG4 HD are so very much un forgiving in the low speeds of 10 to 30 mbps pipes.


    Posted by Kelly | March 31, 2006, 12:56 PM
  5. * Sooo much marketing spin on FTTN (ADSL2 or VDSL2) vs FTTH
    * AT&T and VZ both are deploying FTTH where they can and FTTN where they have to
    * Also … why get so bent on a comment from a COO on how his network works. It is not like this is the CTO talking, all the other execs are just spin.

    Posted by Iain | March 31, 2006, 2:21 PM
  6. My comment on MPEG4 was about visual quality. It was not about the compression ratio. I realize that MPEG4 provides a much better compression but this is currently at the expense of visual quality. I’m not a coding expert, but people I talk to and respect say that MPEG4 has a long way to go in terms of quality.

    If you look at what I think the biggest near term driver of IPTV narrowcasting will be – Sports – and combine it with HDTV, MPEG-4 is not very good. So the area where IPTV is most likely to be successful is the one area AT&T will struggle.

    As for Korean WiMAX (WiBro, to be exact) I think you are radically under the WiMax hype-spell. It isn’t widely deployed in Korea today. And it certainly isn’t 30Mbs sustained outside of a lab.

    Finally, as an investor, if an exec decides to speak about technology, he should know what he is talking about, or simply say – “I don’t know”.

    Posted by Andrew Schmitt | April 3, 2006, 9:04 AM
  7. *execs on technology – It would be great if execs said “I don’t know”, but as a realist I like “Kedrosky’s” quote a few months ago where he said something like “If you assume you misheard the CFO when he was talking strategy you’d be right more often than not”. sad but true.
    * Sports – (and news) are the only things that really require real time delivery. The remaining video content is better just downloaded (using your terms … this is Macro … many people like to own DVDs and will like to own films and TV shows)
    * BTW — Love your site. I like the tech detail.

    Posted by Iain | April 3, 2006, 1:03 PM
  8. The interesting measure here is about how successful Verizon has been in its deployment of FIOS. In areas where they have been let into the market (against strong opposition from the cable monopolies) they have proven to be able to deliver reliable connections at higher speeds for similar or less cost. I know there is a particularly strong fight going on in New Jersey where the cablecos are trying to prevent Verizon from being able to offer FIOS, but I sincerely hope the cablecos lose. It will be the best thing for competition and the consumers there.

    Posted by pkp646 | April 5, 2006, 1:59 PM
  9. I disagree with the thesis od the article.

    “DSL technology limited to 20-30Mbs”

    So ?
    Besides, it is not true with VDSL.

    “Requires use of MPEG-4 video compression, which doesn’t look as good as MPEG-2”
    I would challenge anyone to tell me if they are watching a MPEG2 or MPEG4 encoded movie.

    “Requires a set top box for each TV in the home. Cable ready TV’s can’t tune IPTV”.
    This issue will disappear in the future AFAIK.

    “The complex hardware and electronics AT&T are deploying ‘near’ your home will junked within 10 years”.
    You will computer will be junked within 3 years. Should you not have one ? Verizon will have the same issue in their labs. It is part of the game …

    Will ATT have to ‘upgrade’ to FTTH in the future, maybe ? But at this point in time, I think ATT is smarter in their deployment than Verizon.

    Posted by bizzbozz | April 5, 2006, 3:34 PM
  10. VDSL can’t reach speeds beyond 30Mbs without introducing significant spectral interference on adjacent ADSL lines, and only over distances that would make it useless as a FTTN brownfield deployment. Pair bonding maybe, VDSL – no way.

    AT&T’s economics don’t work if the equipment gets junked in three years, particularly if it doesn’t aggressively capture share. The biggest cost they face is the opportunity cost of keeping their existing subscribers by providing a compelling experience.

    As for MPEG-4, clearly you’ve never been in marketing. Tell someone the picture isn’t as good, give them a corner case example, and they’re sunk. It’s a marketing slam dunk for the cable guys – people are going to be falling all over each other to write the copy.

    Posted by Andrew Schmitt | April 5, 2006, 4:16 PM
  11. Regarding “Requires use of MPEG-4 video compression, which doesn’t look as good as MPEG-2…” and “I realize that MPEG4 provides a much better compression but this is currently at the expense of visual quality”:

    You’re not comparing the right things. It’s the combination of bitrate, coding technology, and the specific vendor’s encoder that define video quality (neglecting other ancillary factors such as transmission line quality, etc.) I could show you an MPEG-1 stream, or even a motion-JPEG stream, that looks better than MPEG-2 or MPEG-4, if you give me enough bits to code it with.

    The current state of MPEG-2 technology has “broadcast quality” SD streams at an average of 3 Mbps, and HD streams at an average of 14 Mbps. (Actual milage may vary). Because Lightspeed needs to put 3 of these down a DSL line (to cover the case of 3 TVs in the home), MPEG-2 HD will break the bank. MPEG-4 as a rule of thumb achieves the same level of quality at about half the bit rate. (Again, actual mileage may vary).

    The rest of your article is very nice.


    Posted by V | April 7, 2006, 4:15 PM
  12. I always laughed at how they could brazenly call this “Project Light Speed”
    “Project Copper Speed” or “Project Shannon’s Law” would have been more

    Posted by dw | April 19, 2006, 12:47 PM
  13. In the vast Majority of cases, Stephenson is correct — users won’t notice the difference between a 1Mb, 6Mb, or 15Mb connection. This is partially because the speed of the local loop has almost no impact on time required (or, more accurately, the perception of time required) to download a Google search page, e-mail or most text-based blogs. It is also because the speed in the local loop is less important than the extent to which the backbone bandwidth has been oversubscribed.

    Assuming Verizon has provided more bandwidth in the Internet connection (from the CO to the Internet, not your house to the CO), your 15Mb service may be much better than a FTTN or cable connection for applications sensitive to large, sustained bandwidth (and at least eliminates the local loop as a constraint). This would allow you to stream HD (MPEG2 or 4) in real-time, if such content was available to stream…

    Over time, this distinction will become important. Today, it rarely is.

    The more interesting question is whether the FTTH connection operates at lower latency than DSL or cable (which is likely given the simpler network). That would have an immediate, noticeable impact on the perception of response times for web surfing.

    Posted by Bruce | August 18, 2006, 12:13 PM
  14. Hello Andrew,
    Do you know how much AT&T is paying Alcatel for a VDSL2 port? it is interesting to know how much they really invest and what will be their ROI, or at least what they perceive it to be…

    Posted by Themnule | March 18, 2007, 6:53 PM
  15. I do not know exactly. I do know Huawei DSL ports were going for $15/port, though that may not have been ADSL2 . That is just for the SLIC card, not including Chassis. Think of it as the variable cost.

    I would bet it is less than $30 but more than $15. Other readers may guess better.

    Posted by Andrew Schmitt | March 18, 2007, 7:09 PM
  16. I didn’t know that the fiber optic cable does not go all the way to the home. What was behind that thinking other than cost savings, if there would be any cost savings. If you’re going to deploy fiber optic, you might as well take full advantage of it’s speed. At the same time they need to upgrade the backbone to match the speed to homes. There shouldn’t be any reason why a fiber optic deployment wouldn’t offer at least 1Gbps.

    Posted by Mark | April 6, 2007, 6:12 PM
  17. You are implying that MPEG-4 video will generally be inferior to an MPEG-2 video at the same bitrate.

    This is only likely under the most unfair comparison conditions — e.g., using the worst MPEG4 encoder/codec/player against the best MPEG2 encoder/codec/player.

    Generally, MPEG-4 streams will produce a much better picture at the same bitrate or maintain the same quality at a significantly lower bitrate than MPEG-2 streams.

    Posted by Josef | July 7, 2007, 5:12 PM
  18. Two thoughts:
    1)I like the Obi-Wan reference, however he used the “force” while AT&T is using the “farce”.

    2)I think AT&T is missing the boat. Most people in my neighborhood who have signed up for DSL wind up having to have new phone cable burried anyway. Whay doesn’t AT&T just burry fiber to their door rather than wire, and spread the cost out over more years?

    Posted by Brian | July 12, 2007, 3:03 PM
  19. Please do more research on infrastructure before writing more spin.

    ATT FTTN units will eventually be pushing 100 Mbps services short distances (>1 Kft) over coper drop wires. For concern regarding the copper from the FTTN to the premises the drop wires are 22 AWG vs 26 AWG for typical Cat5 or better cabling. It also contains a bonded/grounded shield (which Cat5 does not). Those speeds are attainable with reliability as long as everything is bonded/grounded. Most people do not seem to have a problem attaining and sustaining 100 Mbps or better transfer speeds in the LAN environment; whats the difference in running a long Cat5 drop or running a long heavier gauge OSP equivalent?

    FTTH is EXPENSIVE to deploy. How many times have you or someone you know had a drop wire damaged? Copper is cheap to replace, ATT doesn’t charge for drop wire (usually), what happens when you cut a fiber drop while landscaping? Who pays for that one? How do you power it? Power pairs? (copper pairs carrying up to -180 VDC). What about power outages? Where do you place the UPS? Who pays for it? Ever need to make a call when there is no power?

    The other problem is easements, a lot of municipalities will not give (or sell) any more space for ATT to bury ie Verizon in NYC and ATT in Miami. Manholes are usually full so now what? Pull out that working 1,800 pulp (paper) cable and put the customers out of service while they pull fiber and quickly do a site turn up? Not unless their is a natural disaster like in New Orleans. And if you did most of those cable runs are not in conduit so you don’t just slide a new fiber in place.

    For those who do not feel these speeds are attainable over the old PSTN facilities how is it that ATT and the former BellSouth has been deploying up to OC3-like speeds over copper reliably? aka Metro Ethernet. Most installs are fiber but when cost prohibitive they use their old copper plant using a DSL technology. Generally holding a 12-15 Mbps synchronous connection on each pair reliably. That many times is on 20 year old 18-26 AWG copper plant.

    Yes ATT has a LOT of work to do upgrading/repairing their network, and often they make costly mistakes, but you are comparing one of the largest telcos in the world with enough market capital to buy and sustain many small countries to the likes of some cable tv companies? Who though in their small monopolized markets can be fiercely competitive, they generally just do not stack up on the large scale.

    Sorry to defend the companies that were and now are ATT but after bring the world the Telephone, Data networking, the transistor, the cell phone, the solar cell, the laser, digital transmission and switching, the communications satellite, the first single chip digital signal processor, UNIX, the C programing language and there is probably a few things I have forgotten. Some one over their might know what they are doing. But why so much spin against ATT? Is it because inventing Long Distance communications allows people to call you that you do not want to hear from? (don’t forget to call Mom) Oh and that UNIX thing was rumored to of had something to do with that seldom used thing called the internet. (See the wiki for more info

    Just my two cents…Or was that a buck fifty?

    Posted by Chris | August 7, 2007, 10:09 PM
  20. Thanks for taking the time to leave an intelligent reply. I am not trying to spin, just write my opinion. It is great to see someone else taking the other side of the debate because you are right it is easy to dump on AT&T.

    The issue about fiber cuts is partially resolved with connectorized fiber. I grant you that if it si buried it is very hard to replace. But buried copper isn’t much easier.

    There is no need to apologize for defending AT&T for exactly the reasons you point out.

    I think it comes down to the fact that the infrastructure clearly will need replacing in the next 10-15 years, and the interest rate environment is (was?) very conducive to taking the leap now. Then again, someone from AT&T may invent a new UWB last mile delivery mechanism and make fiber obsolete….

    Great Comments, thank you.

    Posted by Andrew Schmitt | August 8, 2007, 8:04 AM
  21. Gentlemen,

    I’m sure the debate regarding DSL vs fiber will rage on, especially in this country. The truth is, worldwide there is little debate. The DSL forum’s most recent stats demonstrate DSL maintaining a 65% worldwide market share with no sign of slowing down. Fiber continues to hold

    Posted by George Hightower | September 3, 2007, 4:34 PM
  22. George:

    While I am inclined to agree with your conclusion, your method of argument is not appropriate.

    Market share isn’t the critical measurement here. One must look at the margin to see what is really happening – i.e. in the last year what % of broadband adds were FTTH. While still small you will see it is a more significant force.


    Posted by Andrew Schmitt | September 3, 2007, 7:52 PM
  23. Trackbacks / Pingbacks

  24. FTTHblog » Critique of AT&T’s Project Lightspeed | March 30, 2006, 1:13 PM
  25. Project Lightspeed at Broadband Reports - Nyquist Capital | April 3, 2006, 4:08 PM
  26. The Cablecos are Waking Up to Reality | August 17, 2006, 12:02 PM
  27. Lightreading Gives Ma Bell Both Barrels at Nyquist Capital | May 3, 2007, 3:03 PM