Werbach's Supernova blog

James Rogers jamesr@best.com
07 Jan 2003 12:25:10 -0800

On Tue, 2003-01-07 at 10:03, Mr. FoRK wrote:
> Here's a quote:
> "In about 2-3 years you'll be able to have the bandwidth of a medium-sized
> city's telephone CO, for about $1000."
> What does this mean? Who is 'you' (individual, company, etc)? Is this $1000
> yearly or single-time capital cost? How is this technologically possible?
> How do the bits move? Who owns or controls those pathways?

That would be $1000/month for a very fat pipe to and through the core
with no over-subscription.  If you knew how much a savvy Tier-1 company
actually pays per Mbit through the core today, you'd realize that we
aren't that far away from this ideal.  With improvements in last/first
mile technology, this is definitely feasible and realistic.

90% of the direct cost of bandwidth right now is peering with crusty
old-school transit providers who are trying to amortize their
unreasonably expensive network infrastructure and corporate overhead. 
As the need for peering with these guys diminishes, the effective cost
of bandwidth will plummet.  In three years, I would estimate that as
alternatives get built out and become more ubiquitous (re: peering with
old farts) the actual bandwidth costs will probably be reduced by a
factor of two or three due to reductions in peering overhead.  Combine
this with a ten-fold stepping of bandwidth per lambda (which will
effectively arrive in the next year or two), and you have a significant
reduction in per Mbit costs.  In three years, retail prices of around
$20/Mbps through the core is not at all farfetched, and the direct
operational costs per Mbit would be several times less.  As a home user,
who never actually purchases bandwidth through the core, the retail
price will be even less.

The retail price for bandwidth right now is artificially inflated
several-fold.  The new breed of non-telco network providers are finding
that there is plenty of market so that they don't have to compete with
each other yet.  As a result, they are intentionally pricing their
bandwidth just below the telco floor.  The telcos have already been
priced out of the market in many areas, but for now the telcos are being
used as a way to prop up the prices (everyone is "competing" with the
telcos on price).  When/If the telcos cede the bandwidth battle and/or
the non-telco companies become big enough that they start seriously
competing with each other, you'll see a very marked drop in bandwidth
prices to more market driven levels.  I think we will see this
competition emerge in the next couple years, and this will coincide with
the dramatic drops in real bandwidth costs for the technological and
operational reasons I mentioned above.

This is my analysis of it, and people who are well-connected to what's
going on in the center and the fringes of the industry (e.g. Dave
Isenberg) would be aware of these trends as well.


-James Rogers