Friday, May 23, 2008

Reasons CDN's Are In Trouble

The other day I posted about my view that p2p in Flash 10 will put a serious hurt on the CDN (content distribution networks) business. But that is by far not the only problem I see for CDN land.

As background, CDNs are services that provide high speed servers for hosting videos or other files that you expect to be accessed a lot, and that you want to be delivered very quickly.

The problem with CDNs is that they are inherently a commodity businesses. Yes, there will be niches for which they can and will continue to succeed. But for much of what CDNs offer, Amazon, and eventually Microsoft and Google (AMG) will do it cheaper.

Today, Amazon offers a service called S3 for Simple Storage Service, that is essentially a CDN, but without the high performance guarantees. S3, when its not down, has pretty good performance. People are *flocking* to it. I am totally confident that Amazon, with all its resources, will increase up time and get at or near CDN speed over time. They have invested heavily in all their developer platform services and I don't really see a way for anyone other that Microsoft and Google to compete with them doing what most will perceive as being a higher priced version of the exact same offering.

The reason AMG will have a cost advantage is obvious, but just to be clear, all three members have massive buying power that will be hard to compete with. And perhaps the most disturbing issue that must be facing the CDN world is that AMG does not need to make much, or perhaps any money at this business because they are all looking for developer relationships that will yield value in other ways.

Of course, if the CDNs really can offer unique services that AMG can't compete with, they will be less threatened. But I don't see how. The switching cost between CDNs is low, and there is no network effect or other good lock-in mechanism. So even though today there are performance benefits to CDN services from companies such as Akamai, Limelight, and Panther over lets say Amazon S3, I do not see those lasting. And I would even suspect Amazon and the other AMG members will ultimately offer different more focused products that aim more directly at the heart of the CDN market.

Finally, others are predicting a shakeout, because the entire CDN business had only $450 million in revenue last year while $281 million of venture capital poured into the market. In otherwords we are in a massive CDN bubble right now.

So I while I do see niche CDNs surviving for very specialized purposes, perhaps like live event Flash video streaming, I can't see CDNs really being able to make significant money at more basic stuff like hosting JPEG picture files or progressive download video (like YouTube) -- which are today the heart of the market.

As I see it, CDNs will have to be much more creative and really will need to move up the software stack to survive. This means offering tools that do more than just act like a big fast hard drive. They will need software tools that integrate with the storage that developers need. They will need to lock developers in by providing software tools that are unique and that get baked into the developer's application in ways that make switching hard, but also provide great value. This strategy is key because in most cases, the cost of development and time to market are far more important than the cost of storage and bandwidth.

This is my prescription, but I don't really believe the CDN guys have it in their DNA. We will see. But one thing is for sure. Whatever they are going to do they better hurry because AMG is coming and time is indeed of the essence.


  1. Taking your theory out 5 years...all youtube videos will be streamed p2p resulting in ISP's wanting to change their pricing models for consumers up bandwidth, of course only benefiting the ISP's in the end and resulting in a shift back to non p2p flash.

    Just a thought

  2. marc: I definitely agree that it's possible that this will ultimately affect the demands and economics of consumer-grade internet access. But I don't think it will necessarily have the impact you're predicting. IMHO, the responsibility to deliver a quality user experience will remain shouldered by the media outlets who will be making decisions like whether or not to rely on P2P delivery, or to what extent to rely on it. Internet connections provided by many popular (and crappy) ISPs routinely gag on aggressive upstream demands from its users. If users experience this at YouTube or wherever, they will ultimately react to this with egoist pragmatism and force media outlets to rethink their media delivery strategies when they start visiting other sites that deliver higher quality media more quickly.

    Hank: You ought to consider displaying a little more shame when you proclaim business imperatives such as a need to "lock in" your customers and "make switching hard." Consumers are getting wiser about computers -- slowly, but it is happening -- and greatest on the uptake is an understanding of how perilous and anticompetitive these underhanded tactics are. That kind of attitude will draw more than a little well-deserved hate.


Note: Only a member of this blog may post a comment.