Article Info

Level3 and Akamai – The Investment Paradox

Level3 (LVLT) tries very hard to appear hip with the Web 2.0 vibe, and today’s WSJ article cites content distribution (namely video) as the reason for a resurgence of investor interest in carriers. Simultaneously, investors continue to flood liquidity into content delivery networks (CDNs) like Akamai (AKAM), trumpeting them as key enablers of the Web 2.0 content rage. It is impossible for both to be right, as each business is designed to eliminate the need for the other.

Om Malik has his take on the WSJ article here, filled with lots of good linkage to previous Level3 articles.

The CDN business model is about taking content and storing it close to the consumer. Whether it be Apple’s (AAPL) iTunes, Microsoft (MSFT) software patches, or the latest silly video (my favorite), geographically distributing the data achieves the following:

  • Reduces the need for core bandwidth. Rather than everyone downloading iTunes tracks from Cupertino from around the world they download the file from a server 50 km away.
  • Improves latency. There’s no getting around the speed of light, and each router hop in the Internet adds latency.
  • Improves QoS and reliability. If the closest server is down or overloaded, others a little further away can take over.

Hence, every time your computer fetches data from an Akamai server, it’s essentially robbing a carrier like Level3 of the right to carry that traffic, and reduces the overall demand for carrier transport services.

There are very valid technical arguments for NOT using distributed caching. Some would argue it’s better to centralize the distribution of content for all but the most popular files, and use techniques like SIP to ensure QoS. There are big economies of scale when it comes to building datacenters (example: Google (GOOG)), and others would argue that core bandwidth will be so cheap there is no need for the hassle and expense of caching at the edge.

I’m not trying to say one approach is better than the other. Experience has shown me that Akamai investors are second only to Apple and Google investors in their rabid devotion. I think it’s an excellent technical debate and far too early to call it one way or the other.

However, I don’t think there is any way that companies like Level3 and Akamai will both profit in an outsized way from the distribution of digital content. It is a zero-sum game; either cached at the edge or transmitted from far away. Not both.

This type of market paradox exists in other areas, and leads me to believe we are in a Web 2.0 equity bubble, albeit localized to a handful of stocks I like to call the ‘Web Twenty’.

Full Disclosure: I hold shares in Akamai but am market neutral


Comments are disallowed for this post.

  1. The paradox is easy to resolve. More than half the bandwidth on core networks is P2P video distribution. Akamai doesn’t service this (LVLT does). On the other hand, as traditional media outlets increase their download business, Akamai wins. P2P is still very large and will likely increase. So Level 3 gets its increased traffic too.

    Posted by Scott | December 21, 2006, 5:01 PM
  2. That is a very interesting point.

    But exactly how do you monetize traffic that exists only because of a reluctance to pay? What carrier is going to buy more bandwidth to pay for more P2p traffic? None. They’ll just police it until it is legit.

    Ultimately, if P2P does go legit, it too can be cached, and it will (or will not) be in the best interests of carriers to do so. The only reason P2P isn’t cached today is because carriers would essentially be storing and distributing unlicensed content.

    See for p2p caching.

    So you are right but P2P is either a short term technology or something that has high revenue elasticity.

    Posted by Andrew Schmitt | December 21, 2006, 5:11 PM
  3. I disagree completely Andrew- you’re assuming you can stamp out piracy online or even bring it to some kind of ‘acceptable’ level. Carriers will of course keep on buying up bandwidth because their consumers demand it, regardless of the legality. They play ball with recording associations, dutifully process C & D requests and pass details onto the courts.

    P2P is going legit- see BBC tie up with an American p2p content distributor this week, and Bram Cohen’s deals with various media companies. I’d argue that many people use online services for say TV shows or movies because there isn’t a serious alternative quite yet- although clearly efforts are under way to sort this situation out. There will obviously always be people put of by paying, but one need only recall the impact (supposedly) itunes et al. had on the way people got music online.

    In this sense, you might be optimistic and say p2p share will stay the same, but the ratio of legal to illegal content will shift to the former. Or you could conclude that as more people come online, and find a goldmine of content available for free, it will keep on getting worse!

    Returning to my point however- as p2p becomes legit, does akamai get stiffed in a zero sum game? Unlikely. p2p assumes people will share a diverse range of content for free (or rather at their own expense). What do you do when you have a very niche item, of little mainstream interest, that doesn’t bounce around a p2p network because demand is low? You’d have poor QOS for end user – so you can see situations in which a mix would make a good deal of sense.

    Posted by Tom | December 21, 2006, 5:59 PM
  4. I think it is a big assumption to say P2P goes legit. I will agree it is an interesting debate.

    I don’t think it has any bearing on this debate though. If P2P goes legit, then it almost certainly will be cached. For scarce content, the owners of that content will take responsibility for seeding.

    P2P is actually the MOST inefficient way of distributing content from a network utilization standpoint. If it were to become big, the network operators themselves would pay for caching. So I still don’t see how this factors into the debate.

    Posted by Andrew Schmitt | December 21, 2006, 9:40 PM
  5. Level 3 owns a vast metro fiber footprint, so alot of Akamai’s communications between its facilities will still run over their fiber. Akamai is well suited to pull traffic down that doesn’t change much or is used by many people. For unique content or content that changes, centrally serving is still the way to. Finally, it’s alot easier for a Level 3 to get into Akamai’s business than it is for Akamai to get into Level 3’s.

    Posted by Anonymous | December 21, 2006, 9:42 PM
  6. Not too sure about that- how much of Level3’s metro fiber footprint is leased vs. owned?

    100% agree on your observation of one company entering the others business.

    Posted by Andrew Schmitt | December 22, 2006, 7:43 AM
  7. When I read your paragraph that said:

    “However, I don’t think there is any way that companies like Level3 and Akamai will both profit in an outsized way from the distribution of digital content. It is a zero-sum game; either cached at the edge or transmitted from far away. Not both.”

    I immediately thought of Internap (INAP). A company that recently acquired a CDN by the name of VitalStream. I believe INAP’s business model is going to be providing both network traffic (high performance required traffic through PNAPs) and Content Delivery to the edge. So, in effect they are going to do both.

    I find this model interesting because it appears that Internap is going to capture an end to end solution for Web 2.0 customers seeking colo, traffic, and CDN. They claim to be going after a possible niche of high end customers who require 100% uptime for their customers. This means higher margins.

    Some investors appear to be very pleased with their model of doing both as their stock has risen over 350% in the last year.

    I wanted to point these guys out to you Andrew and see if you considered their business model when you researched the article? If so, what are your thoughts on what appears to be the only company doing “both”.

    Posted by Kevin | December 22, 2006, 9:34 AM
  8. Andrew –

    I enjoy your blog, but have never posted. Before I go on, I’ll fully disclose that I work for a CDN called Panther Express

    With all due respect, I disagree with your premise. As much as us CDN’s want to peer with as many networks as possible – creating the natural symbiotic relationship of those who have content and those who want to access it – we all have to buy upstream transit at some point. The only folks who don’t buy transit, as you probably know, are the Tier 1’s.

    For instance, Panther is a Level 3 customer. We’re also a GLBX customer, a Telia customer, a PCCW customer, etc. The more traffic we push through our content delivery network, the more we have to buy from transit providers. Again, some of this traffic will be offloaded via peering relationships with the smaller guys, but at some point a CDN customer’s content has to reach an eyeball thats sitting on a Level 3 downstream somewhere – and every Mbps that touches a paid transit link gives that ISP another few bucks in their pocket.

    IMO, CDN’s don’t compete with Tier 1 ISP’s. We enable them! I guarantee that if Akamai is a Level 3 customer, LVLT wants nothing more than for Akamai to keep growing. And if AKAM is not a LVLT customer, they’re knocking at the door trying to get a piece of their traffic. Granted, ISP’s don’t necessarily like CDN’s – b/c they would get a higher margin off a customer buying a 100Mb commit vs. a CDN buying a 10Gbps+ commit – but they’re still making bank on those kind of volumes.

    CDN’s WOULD/do compete with the smaller ISPs/Hosting providers that are buying bulk transit and trying to get the hot new Revver or Veoh to colocate with them and push all the Mbps out of their network. However, these video sites (and the like) need the scalability and performance increase that a CDN provides, so thats why the CDN space (as well as the transit space) will just continue happily grow :).

    Posted by Jason Evans | December 22, 2006, 5:12 PM
  9. Any insight to rumor that level3 is buying Savvis CDN operations? That will make them indifferent.

    Posted by Nocturnien | December 22, 2006, 6:46 PM
  10. What’s being missed here by the author is the technical understanding of what Akamai and Level 3 provide. Level 3 is NOT a CDN. Level 3 provides pipe and co-location services for those who want to operate their own servers, like a MySpace. Akamai is a CDN that has it’s own servers that customers basically rent. Two completely different products.

    It IS possible for both Akamai and Level 3 to win because they offer different services and go after two different types of customers. Level 3 is not a CDN, they are not in the Microsoft or Adobe certified hosting provider programs and can’t deliver Flash or Windows Media. Don’t take my word for it, look at the Adobe and Microsoft pages:

    This article is also factually incorrect when it states that “Hence, every time your computer fetches data from an Akamai server, it’s essentially robbing a carrier like Level 3 of the right to carry that traffic, and reduces the overall demand for carrier transport services.”

    No, Akamai does not own or operate it’s own network. It owns no pipe. Akamai owns and operates it’s own servers, sitting in other companies data centers and they lease pipe from numerous carriers INCLUDING Level 3. Akamai is a customer of Level 3.

    Again, it is factually incorrect when the author says “However, I don’t think there is any way that companies like Level 3 and Akamai will both profit in an outsized way from the distribution of digital content. It is a zero-sum game; either cached at the edge or transmitted from far away. Not both.”

    Companies like ARE putting content at the edge, but they are doing it via their own servers co-located at multiple data centers that are conected to pipe like Level 3. They are doing the same thing Akamai is doing, but doing it themselves, on their own hardware and with their own pipe contracts with the carriers because it is cheaper, for them, then using Akamai.

    Posted by Dan Rayburn | December 23, 2006, 10:29 AM
  11. Does not the purchase of SAVVIS CDN
    by Level 3 pose a threat to AKAM?

    Seem with all the relationships Level3
    has with the “new, internet 2 players
    and the academia crowd of Internet 2,
    and Comcast…” they are in a great

    Posted by Jon Smith | December 26, 2006, 1:04 PM
  12. There is a lot of tech lingo here that most people do not understand, including myself and probably the author of the thesis that Akamai and Level 3 are “mutually exclusive”– James Cramer disagrees because he loves BOTH stocks, and I trust his analysis, but more I trust Bill Miller of Legg Mason who has owned LVLT for a while BECAUSE HE KNOWS THAT THERE WILL BE A BANDWIDTH SHORTAGE (as Cramer says) and that LVLT WILL BE IN THE DRIVER’S SEAT THEN.

    Posted by Herb Luria | December 26, 2006, 3:23 PM
  13. and gee whiz we all know Cramer is a Tech genius.

    I’m not saying L3 and AKAM compete (though a recent announcement may change that). What I am saying is increased demand for CDN services reduces what the network utilization would have been without them.

    The media and casual investors mistakenly believe both transport bandwidth and CDN’s will benefit from Video. What I am saying is that the profit to be made will be highly asymmetrical.

    Posted by Andrew Schmitt | December 26, 2006, 11:44 PM
  14. Tech Garble!? I was trying to speak as tech-laymen as possible. :)

    Obviously, with the Savvis CDN acquisition by L3, this entire post with comments has been altered, but my initial point hasn’t, so I’ll stress this point once more:

    Akamai doesn’t own a network backbone. Therefore, they MUST buy transit from ISP’s such as Level3.

    A CDN is just another avenue for an ISP to make $$ on transit revenue. CDN’s give an ISP higher global network utilization, not less.

    Posted by Jason Evans | December 29, 2006, 8:36 AM
  15. Yes, but 99% of the traffic generated by a CDN is probably peered with the ISP delivering the content. L3 doesn’t benefit unless the ISP needs to lease dark fiber.

    Posted by Andrew Schmitt | December 29, 2006, 2:35 PM
  16. Sorry, but thats not correct. Akamai would love to offload 99% of its transit via peering, but it doesn’t. They pay Level3 – and all other Tier 1 ISP’s – for transit.

    Posted by Jason Evans | January 3, 2007, 8:09 AM
  17. One vote Dan and one vote for Jason. Symbiotic relationship between the two (Level3 and Akamai) with Akamai enabling and continuing to enable more growth for both concerns. There may be some customer-envy and tech envy (either way) but Akamai has provided content providers a way to NOT become a hostage to peering and peering limitations both technical and political/business… conceptually and practically, they have flattened/unified the Internet for their customers such that ‘the cloud’ works with one point of contact for their customers web apps. As an added benefit, ‘the cloud’ is more efficient and the networks are better off for someone having made sense of the clan wars and making the whole of the web more commercially practical. It’s about ‘core competencies,’ sorry for the cliche but both companies have their own and a CDN I believe really needs to be network independent, something that a L3/Savvis combo would have a hard time pulling off?. I think.

    Posted by Damian J | January 11, 2007, 1:41 PM
  18. If/when there is a bandwidth shortage then LVLT obviously benefits but the value of AKAM’s service becomes more valuable too. I remember one of AKAM’s early problems was that because bandwidth had been so overbuilt the prices were much lower than expected and AKAM couldn’t offer the savings advantage they anticipated.

    Posted by jc | January 23, 2007, 4:33 PM
  19. What impact will AKAM buy of Netli have? Just buying out another upsatrt competitor or do they offer something significant to AKAM offerings?
    Feb 5 (Reuters) – Akamai Technologies Inc. (AKAM.O: Quote, Profile , Research) said it agreed to acquire Netli for about 3.2 million shares to enhance its products that improve the performance of Web- and other Internet-based applications.

    The deal values Netli at about $178 million based Akamai’s closing stock price on Friday.

    The company, which speeds up delivery of digital information via the Web, said the deal is expected to be neutral to its 2007 earnings. (Reporting by Amitha Rajan in Bangalore)

    Posted by jc | February 6, 2007, 8:56 AM
  20. Well, to put the L3 vs Akamai debate to rest, and I quote an earlier post here “Finally, it’s alot easier for a Level 3 to get into Akamai’s business than it is for Akamai to get into Level 3’s.”

    And that’s exactly whats happened. L3 is now selling their own CDN on their own network and is in direct competition with Akamai in the CDN space. So L3 doesn’t care how you want to deliver, from the origin servers or cached on their network. And L3 is working on streaming CDN services and should have that available by end of the year. It’s already in beta now… And I doubt that legitamizing p2p distribution will have much effect on either model, as the legit distribution of content is their bread & butter anyways,, espn, cnn, etc…

    Posted by Andy Kiefer | September 20, 2007, 7:41 PM
  21. How does selling CDN that works within a single network, in L3’s case their own, really work?

    Akamai’s strength is having servers in many different networks, allowing both network and geographic closeness to end-users.

    Posted by Abe | January 7, 2008, 3:03 AM