5/18/98 Barron's. Our experts pick the likely winners in the global telecom wars
[No ASND reference. Article most discusses impact of European deregulation.]
interactive.wsj.com
Excerpt on ISP benefit from VoIP:
Jack B. Grubman, a managing director of Salomon Smith Barney, who heads that firm's global telecom team.
Q: For all our talk about wireline and wireless communication, will the Internet and satellites hurt or help phone companies as we know them?
Grubman: To encourage use of the Internet for voice traffic and enhance
phone services, the Federal Communications Commission ruled that, unlike
long-distance carriers like AT&T, MCI, Sprint, whatever, Internet service
providers -- ISPs -- need not pay access charges to local phone companies.
That's the only reason ISPs have a cost advantage over regular carriers of
domestic long-distance voice traffic.
Before ISPs, if you wanted to call from New York to Chicago, here's how it
worked: You'd pick up the phone in New York, and your long-distance
carrier would pay Bell Atlantic on the originating end, and Ameritech on the
receiving end, three cents per minute.
New long-distance carriers quickly figured out that if they could offer voice
service over Internet protocols, or IP, they could avoid paying access
charges. Guess what? If you look at the unit cost per minute for AT&T and its
ilk, 60% of that sum is the access charge they pay to the RBOCs. If 30% of
long-distance calls went over the Internet, this traffic would be carried by
long-distance providers, which own Internet backbones, and the Bells would
lose all these access fees. Rest assured, this won't be allowed to happen.
Voice-over-the-Internet basically is an arbitrage to avoid access fees.
If the playing field becomes level again, and I'm sure it will, the threat to
RBOCs from the Internet, at least for domestic calls, largely will go away.
Q: How's the quality of Internet telephony?
Grubman: When you make a regular circuit-switch call, you use a dedicated
circuit. When you send any kind of message over routers, which happens on
the Internet, there's no network management, no queuing. A bunch of data hits
the routers. The data go off on their merry way, looking for their destination.
The problem is, the data may not arrive the way they were sent.
Q: Please elaborate.
Grubman: Let's say you send five packets of
data. Each packet goes looking for a path. They
all do arrive at the destination point, but the third
packet may get there before the first one does.
Q: How much sooner?
Grubman: We're talking seconds. That's okay
for data, meaning numbers, but not okay for
voice. It needs to be recompiled in milliseconds or
fractions of milliseconds. That's why you have a
delay on voice over IP. You can't have full duplex conversation, meaning you
can't both talk at once. You'll never hear voice over IP with quite the same
total quality and lack of delay as you get on circuit switches.
Again, if the entire cost advantage of voice over IP is a lack of access
charges, we know that advantage will disappear one way or another. In the
developed world, I don't view voice over IP as being that much of a threat.
There's another factor here.
Q: Tell us.
Grubman: There's a huge shortage of bandwidth, and it will continue. If you
own a lot of bandwidth, why would you commit it to what amounts to a
penny-per-minute voice service when you can use it for data services that
generate higher revenues? That's why I don't believe voice over IP will take
over the world.
If a voice call is a unit of one voice-grade circuit, the typical data message that
corporations need is anywhere from a unit of 2,000 to a unit of 100,000. Data
messages are outgrowing voice by 10-to-1. Here's the key: Everything rides
the same fiber, whether it's a voice call, a data call or an IP message. More
importantly, voice networks are engineered for 6-9 of reliability. Data
networks provide 20-9.
Q: Huh?
Grubman: Quite simply, what all this mumbo-jumbo means is that data
networks carry much more mission-critical stuff than voice networks. If you
and I are conversing and the line goes dead, okay, fine, we'll call back. But if
you're running someone's global data network around the world -- for money
transfers, foreign-exchange reconciliations, inventory transfers and the like -- a
line going dead is a very big problem.
In 1996, when MFS bought UUNet, virtually 100% of UUNet's commercial
customers were connected to it with a T-1 line that handles 1.5 megabits of
transmission. Today, probably 33% of new customers signing up with UUNet
want OC-12 lines, which offer 36 times as much capacity. All this has
occurred in less than two years.
My point is, demand for bandwidth, fatter pipes if you will, keeps going up
and up and up from corporations around the world. Therefore, to be perfectly
blunt, voice over IP is a waste of fiber capacity. It's a little niche business that
will remain that way. And I bet that providers of voice over IP will be forced
to pay access fees to local phone companies within 18 months.
Castro: Let me add something. Outside the U.S., especially in Europe,
incumbents are the biggest providers of Internet access in general, and will be
the natural providers of IP telephony. They're not doing it now because it
doesn't make any sense. |
