George Gilder published his technology report in June 1999
with the subject title of "Will At&T Survive Terayon". It
was Vol IV Number 6. Heading in to June, TERN was trading
at about $31 per share. Exiting June, TERN was around $50
per share.
While the report series is called a Technology Report, it is
not a *technical* report written in an objective style.
Rather, I find that it uses the tools of emotive words and
phrases, sweeping generalizations, belief statements, and
incomplete technical assessment or fact presentation. It is
a persuasive work, I think more to encourage faith than
provide due diligent technical analysis. I'm going to be a
little emotive myself at the lead in, but then will shift to
a more technical and fact critique.
Clearly, he loves CDMA and Qualcomm. Why? I don't know.
From my own experience with Sprint and a Sony CDMA phone, I
disliked the tonal quality of the voice calls and the
problem with base station fading that I opted to go with
AT&T and don't regret the switch. Service wise, both
companies were about equal, with the exception that Sprint
was still (and is still) rolling out it's digital service
and in my business travels, I couldn't even do analog
roam cities where I should've been able to. AT&T has/had
better roaming agreements and the one-rate plan. So as
Sprint turns up there system and rolls out marketing, of
course it's going to look like they are on the rise and
AT&T customers will try it. The question is, what will
their market shares be after some maturity?
Here's a belief statement example: on page 2 of the report,
bottom of first column into top of second. "The only
profitable voice services will be wireless. AT&T is very
proud of its cellular offerings. But here the company is
painting itself into a Time Division Multiplexing Access
(TDMA) corner, while the rest of the world is goes Code
Division Multiple Access (CMDA)." This style is putting
down AT&T for using TDMA (likewise inferring that TCI and
CableLabs DOCSIS which uses TDMA is put down also). The
report in the next paragraph goes on to argue domestic
shifts for taking over the U.S., but never comes back to
address the "rest of the world". U.S. must take over the
world, I guess.... Personally, I'm not sure about that in
view of the GSM presence and growth. We'll see what
happens. Clearly also, there is much more to providing a
service with a successful business plan than just the TDMA
vs CDMA technology opinions. Any way, this sort of spin
throughout the first several pages of the report just builds
CDMA up and puts TDMA and AT&T down.
Also, I never found any technical substance for explaining
why CDMA was better than TDMA for wireless other than the
take it on faith, it deals better with noise, if it works
for wireless, it'll work for wired, trust me, attitude.
Ergo CMDA, in his opinion, is the wireless winner,
therefore, since Terayon is doing S-CDMA in the wired world,
it must be the winner also. I can tell you that wired
environments are different than wireless environments.
He also puts down DSL that must "cram enough data through
those little wires".
After slamming everything but CDMA, Qualcomm, and cable, the
report turns to "Terayon Breaks Out". The side bar on page
6 is "Using Terayon's system, Shaw plans to become the first
operator with 100% cable modem coverage." From what we've
seen from Pat's research, I think that statement should've
said "system and stock growth".
Bottom of page 6, first column. "The cheap and effective
answer is Terayon's cable modem technology. Based on the
same spread spectrum, noise defeating principles as
Qualcomm's (QCOM) CDMA, Terayon's modems encode the data
signals and spread them across the available upstream
spectrum." First inaccuracy is the implication throughout
the report that only CDMA is "noise defeating". Technically
this is quite inaccurate as properly done TMDA approaches
are also noise defeating. S-CDMA does have advantages of
coping with certain types of narrow band interference only
better than TDMA on cable systems. However, the
S-CDMA uses vastly more complex algorithms, and hence,
probably about 2-3 times the number of ASIC gates to achieve
this gain. S-CDMA is able to work in noisier narrow band
environments than TDMA. However, it throws away spreading
codes (data carrying capacity) and raises the transmitting
power of the remaining codes. So in really noisy plants,
S-CDMA may work where TDMA fails. However, the available
bandwidth could be significantly reduced. This has the
financial effect of proportionately raising the
capitalization costs per cable modem support while
proportionately lowering the revenue per MHz allocated, in
addition to just plain throwing away spectrum as I mentioned
in a previous message. The second inaccuracy is "across the
available upstream spectrum". The Terayon system uses a
6MHz wide channel to spread over in both the downstream and
the upstream. It does not spread over the entire upstream
channel.
Another issue, the report makes sweeping generalizations of
the noise environment in the upstream by the absence of
talking about the difference between narrow band
interference and broadband interference. If the cable plant
is "dirty" noise wise, it's likely also letting in broadband
interference that will disturb any high speed data
modulation system. S-CDMA is not immune to broadband
interference. Therefore, in the presence of BB noise, the
operator will need to clean up the system, which usually has
the impact of lowering the whole noise floor. BB vs NB
noise is entirely plant specific and the noise "mileage"
differs substantially from plant to plant.
At the end of the roll over paragraph on the top of page 6,
second column. "in addition, the spread spectrum system
exploits all the bandwidth all the time, using the codes to
differentiate the signals sharing the conduit. Thus, like
wireless CDMA, it can gracefully accommodate bursts of data,
such as a rapidly downloaded film or webfile". This is an
example of an incomplete technical assessment for the
purposes of linking S-CDMA with CDMA. Technically, TDMA uses
all the bandwidth all the time, gracefully accommodating
bursts of data....." too. Also, noise bursts effect TDMA
as equally as S-CDMA. Errors are introduced and packets may
be dropped as a result. Higher layer protocols will
retransmit. S-CDMA may reduce it's spreading codes if the
noise persists to improve packet error performance. In this
case, S-CDMA may be spreading over the same RF spectral
bandwidth, but the data carrying capacity has been reduced
as previously discussed. The manner in which the
"bandwidth" is linked to "rapidly downloaded" is the design
of the statement however it is technically inaccurate when
inferring that the data carrying capacity will be the same
in all noise environments. It just ain't so.
The first paragraph on page 6 second column. "On May 25th,
Terayon announced that its S-CDMA (synchronous-CDMA)
technology now enables Terayon modems to operate in
previously unused downstream cable spectrum where signal
loss prevents the transmission of video channels." This is
the best example at the extremes both Terayon and this
report have gone to, in order to pump up Terayon's image by
the omission the complete facts to their benefit. The area
of "unused downstream spectrum" is called the "roll off
region". It's a subscriber subjective portion of the
downstream upper RF spectrum where analog TV signals begin
to appear visually too noisy - white specs in the picture -
due to RF propagation characteristics of *analog* signals.
It is technical fact, that digitally modulated signals can
work perfectly fine in the roll off region. Hybrid has been
doing this *long* before Terayon became a company. It is
also fact, that ITU J.83 Annex A & B 64 QAM modulation in 6
MHz channels (international standard) using existing
Broadcom chipsets were providing cable modem support in the
roll off region long before Terayon declared that only its
S-CDMA could do so. Motorola and Com21 were both exploiting
this in most if not all of their deployments because it's
the only spectrum available on 550 MHz cable systems because
all the lower spectrum is taken up with revenue bearing TV
channels. Claiming superiority by publishing half the
truth.
Second paragraph, page 6, second column. "In the early
cable modem market, Terayon's superiority went
unrewarded......Terayon's new DOCSIS compliant modem will
not be available until early next year. Cable providers
with already upgraded cable, or plans for upgrade, signed
long term supply agreements for hundreds of thousands of
modems with Nortel/LANCity and Motorola.". Terayon, like
Com21, wasn't in the early cable modem market. The first
round went to Hybrid, Zenith, Motorola, and LANCity. Com21
was late into the second generation market window. Terayon
at the end, but succeeded mostly because it had to buy into
the market with warrants. Based on this discussion and my
prior note, every Terayon deployment makes downstream data
carrying capacity unavailable by not using the maximum
bandwidth. The S-CDMA noise gains are technically small,
but faith-wise large. I can't see myself why it is vastly
superior. Also we all know that Terayon's "new DOCSIS
compliant modem" was really Toshiba's old compliant modem
done OEM. Also, note in the end of the paragraph that Com21
isn't mentioned at all. In fact, "Com21" does not appear
anywhere in the report. Terayon takes on the world, as this
report states, but it can't take on it's rival Com21?
Clearly the report did not want to draw anyone's attention
to Com21 which by that time was publicized as the
third largest supplier of cable modems in the U.S. and
the leading European cable modem supplier.
The next paragraph goes on to mention Shaw in depth and
plays up the superior technology pitch. Unfortunately, the
report doesn't mention a thing about the stock warrants.....
Last paragraph, page 6 going over to page 7. "In a break
through resembling Qualcomm's CDMA triumph in Third
Generation wireless, Terayon's technology finally prevailed
last November when Terayon was asked by CableLabs to join
with Broadcom (BRCM) in authoring the next generation DOCSIS
1.2 standard." We see where that has gone. Reading from
the CableLabs announcement on the postponement of the DOCSIS
1.2 was interesting. It mentioned fast tracking some TDMA
developments and broadening the look at HiPHY to others than
just S-CDMA. I know that some cable operators want just
DOCSIS 1.0/1.1 deployments with the feeling that the PHY is
fine. I know some other cable operators are interested in
Orthogonal Frequency Division Multiplexing (OFDM). We know
that some operators are interested in S-CDMA. Time will
tell. But it looks like CableLabs members were not
necessarily in favor of looking at just S-CDMA in DOCSIS 1.2.
Second paragraph, page 7, first column. "...Access
Communications, another of Canada's top ten cable operators
switched to Terayon's system. Benefiting immediately from
S-CDMA's RF noise immunity, Access reported a 95 percent
decrease in RF related service calls.......Rogers found that
Terayon's modems superior regardless of its highly upgraded
cable plant which is already 85 percent two-way activated."
The second best example of not providing all the facts.
First is the omission of what system Terayon replaced in
both the Access in Rogers systems. Both Zenith and LANCity
were both deployed in Canada. I know Rogers was using
Zenith in their initial roll outs. Why is this important?
Zenith's system couldn't support heavy loads, didn't use
forward error correction, and was an Ethernet repeater
rather than a switch, wasting bandwidth all over the place.
The technology worked ok. That is why it was deployed and
made useful. Comparatively though, any newer system would
provide superior RF performance. I respect the Zenith
product. However, its architecture was from the original,
or old way of doing high speed data. LANCity's system,
while being well thought out, suffered from not using
Forward Error Correction (FEC) in either the upstream or
downstream. You must use FEC. Even Terayon uses FEC. Any
modern digital system using FEC will get way better results,
with respect to errors in the upstream as compared to
LANCity. Also, LANCity upstream could only be in the
cleanest part of the upstream spectrum. Access speaks of
95% reduction in RF related service calls. Out of
curiosity, what happened to the balance of all service
calls? Did others go up or did everything go down? Again,
no mention of the stock warrants with Rogers. Yet another
omission. I do credit Terayon for getting into these
markets and selling modems. However, their spin is being
heavily boosted by the omission of all the details.
Third paragraph, page 7, first column. "..at TCA in
Texas..., which reports that an amazing 85 percent of its
Terayon customers have been able to install the equipment
without any outside help". This could be absolutely true.
However, I'd like to know if the equipment worked after the
installation or did the cable operator have to roll a truck
to re-configure something in the plant for that subscriber?
In comparison, and before Terayon's success at this, Palo
Alto Cable Co-op, which uses Com21 equipment was having
subscribers self install the cable modem. However to become
operational, the outside plant may have needed changes, or
the drop cable to the house needed to be changed, which is
not a subscriber issue. Yes, a subscriber could install the
cable modem in their home, but the cable operator had to
roll a truck to get the service working. At Palo Alto, they
could've claimed a high percentage of customer installs
also, but it's not the full picture. There are missing
details in this Terayon account.
I have no comments about the rest of the report with regards
to technical inaccuracies or omissions.
Mark |
