From "Jubimer" on the Ampex board (relevant to ISLD as an issue, and because AXC is a customer)
Some interesting comments from Kagan
ITG (Internet Technology Group, an Ampex sub -- mo) has interesting implications for cutting bandwidth costs.....
June 30, 2000
STREAMING MEDIA INVESTOR (SMI)(No.21-3)
A line in the recent iBeam IPO filing got us thinking about what it will take to make the economics of streaming competitive with broadcast and cable TV. (The short answer: three to five years and rapid drops in
bandwith costs, coupled with healthy CPM growth.)
In essence, the iBeam comment was that it currently costs about $70 to deliver a three-minute segment of full-screen video via streaming media to 1,000 viewers, for which $40 could be generated in advertising.
Wall Street has become particularly sensitive to economic models that lose money and this one stuck out as a red flag, so we investigated further. The tables below show the analytical iterations and our projections.
Breaking the paradigm down to a three-minute segment made sense, as on average there are 10-12 minutes of ad inventory in one hour of ad-supported programming, which works out to one 30-sec. unit every three minutes or so.
Current broadcast network cost per thousand viewers (CPM), priced against mass household delivery, run in the single digits during the day and in the teens in prime time, with events such as the Super Bowl getting into
the $30-$40 CPM stratosphere.
The CPMs against targeted demographics, however, can run much higher--and streaming media is the ultimate in ability to target viewers one-on-one, so a $40 CPM is probably in the ballpark for a full-screen ad.
Still working backwards, we then calculated how many megabytes were contained in a three-minute video segment, which turns out to be about 11MB with current codecs which can squeeze near-VHS quality out of 500Kbps.
Dividing the $70 by 11MB works out to $0.0064 in bandwidth cost per megabyte delivered, which matches up with the price range of $0.01-$0.005/MB
iBeam noted in its filing, which it charges its best high-volume customers. (This figure probably is forward-priced a bit--the industry average for moderate bandwidth users is probably closer to two cents-three cents per
MB.) The bottom line: In today's market it costs about 2.3 cents to deliver a minute of full-screen streamed video.
To make the economic model work--given that programming costs usually wind up at 25%-50% of total net revenue--it will probably require bandwidth costs to drop below $20 per three-minute segment to reach 1,000 viewers, or less than $0.0007/viewer/min.--at least a threefold drop from current levels. The next step in the analysis was to figure out how quickly bandwidth costs might come down. Over the last year, there have been dramatic improvements in codecs, with RealNetworks' RealServer8 taking a near-VHS
full-screen stream from 1Mbps down to 500Kbps. While there is R&D going on which may improve compression efficiencies by an order of magnitude, tweaks to the incumbent codecs are
beginning to approach theoretical limits. Our projection model postulates that by 2006, barring another breakthrough, the near-VHS bit rate drops to about 260Kbps.
The largest gains will come from competition between edge-servers and infrastructure/backbone providers. Prices have dropped by 35%-40% over the last year and should continue to drop at 15%-30% a year for the next few.
Coupled with a projected increase in CPMs (about 7% a year), the current unprofitable $70/$40 equation drops to $11/$60 by 2006, which sports a gross profit margin of 78%--enough not only for a healthy programming budget but to cover SG&A and even a positive cash flow margin.
What the model doesn't factor in--and this is streaming media's killer upside--is any revenue from transactions, e-commerce, interactivity or PPV. |
