$23.77
-0.40 (-1.65%)
Akamai Technologies, Inc. (AKAM)
CAPS Rating:
The Company provides services for accelerating and improving the delivery of content and applications over the Internet from live and on-demand streaming videos to conventional content on web pages to tools that help people transact business.
RSS Headlines
Fool UK
Recs
Akamai is now facing the biggest competitors it has ever faced in its 10 years of existence. The MIT buzz that they their math is somehow better just isn't true any longer. 10 years has passed since they started and others have had plenty of time to write code and catch up. The CDN math is complicated but it is not Einstein level. Content delivery is now a commodity business using commodity products. They are now going head to head with competitors that manage far bigger, more complex global networks that have greater visibility into the Internet traffic flows for static http objects and streaming and hosted internet video. Both Level 3 and ATT have recently deployed new working content management networks that are in place and will be getting customer wins happily with prices that are half of what Akamai charges. The difference is that they are both running global networks alongside the CDNs which Akamai does not. Moreover, they have deeper pockets to invest in the new faster, emerging hardware.
A little history is in order: Akamai emerged as the early content player in the late 1990s during the time when the majority of users were on dial up. Their distributed server architecture made sense at that time because there were acknowledged problems and congestion at most ISP peering sites. Since then these backbone ISPs and peering sites have now purchased routers with gigabit and 10 gigabit network interfaces that have 1000 times more performance of the Cisco 7500 routers of that era (which were 10 and 100-megabit (.1 gigabit) interfaces. The 7500 were pre-gigabit and pre-10gigabit network interfaces at that time. Many backbone providers have purchased Cisco GSRs and CRS-1 routers, which will scale to 90 terabits or 90,000 gigabits in a multi-rack chassis. The network interfaces used on core routers will be running at 100gigabits by late this year (See announcements on www.nxtcommshow.com). As an example, the new edge Cisco ASR 1000 Quantum flow router that replaces the older 7200 series has the performance equivalent of 160 7200s. The core routers are thousands of times this.
SEE (http://www.networkcomputing.com/immersion/virtualization/showArticle.jhtml?articleID=206901409)
"Leveraging the 40-core Quantum flow processor (announced by Cisco in February and which can process 160 simultaneous threads at a whopping 49 billion transactions per second), the ASR supports the common services found on Cisco's routers, including stateful firewalling, QoS, VPN, and multicast."
SEE( http://telephonyonline.com/access/news/cisco-edge-router-0304/ )
"In terms of performance, said Suraj Shetty, Cisco’s senior director of worldwide service provider marketing, the new router is equivalent to 160 of Cisco’s popular 7200s"
Akamai gets bandwidth and colocation space from other providers but does not own the fiber or networks it runs over. Most of their IP address blocks are from their colocation providers. They play around with DNS CNAME functions and have a large network of caching servers that are colocated around the Internet. These servers are getting cheaper all the time and using VMware, it is now possible to replicate their server architectures for a quarter (1/4) of what the cost is on their balance sheet all with faster quad processors which are much faster than some of the older legacy Akamai servers. Nowhere does Akamai breakout in detail the generation of the servers they have by number or type. It is possible for an edge customer to be stuck behind an aging, slower Akamai server, and have problems getting the page objects and sites they serve up.
Akamai has an amazing NOC that makes for a great dog and pony show. However, the fragility of their CDN architecture, which rests on the unsecure DNS protocol, makes them very vulnerable to attacks. There have been attacks (search on Akamai problems) which have affected major customers. DNSSEC would help on a private backbone but really the core argument is that the CDN world is changing and changing very fast. CDNs play nicely with the Doubleclicks of the world for web advertising but one cannot expect that to last much longer since Google now owns it and could announce a CDN at any time on their deployment of global servers and suddenly there are three majors CDNs out there.
They advertise the fact that the US government uses them for content delivery does not validate a new customer’s purchase decision. Both Level 3 and ATT won the multibillion NETWORX US government telecommunications bids. (http://www.gsa.gov/Portal/gsa/ep/contentView.do?pageTypeId=8199&channelId=-13259&P=&contentId=22827&contentType=GSA_BASIC)
AND
( http://www.gsa.gov/Portal/gsa/ep/contentView.do?pageTypeId=8199&channelId=-13259&P=&contentId=23096&contentType=GSA_BASIC)
You can bet sure that some of the government business that is going to Akamai may go away. And why does this “bright” management give their customer list away on their website to AT&T and Level 3? These competitors now have the real customer lists to target for their new CDNs and just have to offer a lower price until Akamai misses their numbers and the stock price drops like a rock. Employees will exit because the options drop when that happens.
Lastly do not forget the off-the-shelf distributed server load balancers are getting much faster and better for enterprise and ISP customers that can use them as cheaper replacements to any companies CDN. See the recent announcements by A10 networks ( http://www.a10networks.com/) for its new product line of distributed content devices that are selling like hotcakes.
To summarize, I do not own Akamai stock. But even if you don't believe a word of what I've said so far, I believe the multiple they are commanding is too high given that they own thousands of commodity PCs that should be written down (write off 4 or 5 billion of assets) and no network. Remember the Internet will still keep going and run properly if they go out of business and will probably be faster because the slower caching servers will no longer be in the way. Be wary of this stock.
With all due respect, you don't know what you are talking about. Long term, Akamai is much, much more diversified than any other CDN. Many aspects of their business are unique and very difficult to emulate. Further, again with all due respect, owning the "network" has nothing to do with the viability, security, and cost basis of their business. The entire premise of ANY CDN is that they help other network operators by lowering their (ISP, etc) network costs. The result is that ISPs are incented to deploy Akamai's CDN boxes which are deployed inside the ISP's firewall. Owning the network would in no way help Akamai in fact it could actually hurt them. Finally the DNS comment isn't accurate either. The entire internet uses DNS while in theory Akamai is vulnerable they are no more/less vulnerable than any other company. In fact you could argue they are less vulnerable because Akamai's machines only do a specific function and don't have direct interaction with consumer machines. So, do I think Akamai is invulnerable? No, all businesses have vulnerability and Akamai is no different. Do I think they have a unique value proposition and a defensible business? Yes, they are far ahead of their competitors.
Actually, I do know what I am talking about. Let us clarify further:Assets vs RealityTake the price of quad xeon processor Dell Server with 6 Gig of RAM with high availability RAID drives, about US$3K. Let us be generous and say that Akamai has purchased brand new servers for their entire global network of 32000 servers in the last six months. Let us also assume that Akamai is strange and has not heard of VMware that allows the virtualization of more then 10 servers onto one quad XEON processor server which would cut the number of servers needed to 3200 globally. So by the math; 32000 servers x $3000 per server = US$96,000,000 investment in new servers. Now let us assume in this next example that it's not 32000 servers but that they use VMware and its 3200 servers each with 10 IP addresses on each one which makes it look like 32000 servers on their Network Operating Center (NOC) network maps. That math suggests that 3200 x $3000 per server=US$9,600,000 to replace the entire servers every six months. Now how one gets to a market capitalization of 3.7billion as of August 1st 2008 is really good MIT math or really “fool” investors listening to marketing briefs.In this next example let's assume that Akamai, Level 3 and ATT are each rebuilding or building their CDN networks every 12 months to get the fastest possible servers. We are assuming open source free caching software, a free operating system, LINUX, and free DNS BIND server software, which are used by all three. In this example, I try to justify the high market cap and look at what the competiors are likely doing to build out their servers. Assume all CDNs are buying top of the line HP Blade servers at 10k a dual quad blade (HP Proliant BL680c G5 Server blades with 8 XEON processors cores and 8 Gigs of RAM) which is what I would do and then adding in the HP chassis with multiple power supplies and multiple cooling fans. In addition, let us assume that they all using VMware to lower their cost per server, a wiser, better investment. Let us say the investment is US$50K per chassis each including only 4 $10k blades each. And for performance, let us say they have 10 gigabit ethernet adapters to really move the cached content on to the Internet backbones.SEE: http://h71016.www7.hp.com/dstore/ctoBases.asp?ProductLineId=431&FamilyId=2063Then the math is a tad more complex; 32000 IP servers/10 VM IP numbers per physical server blade = 3200 physical blades each with 8 processor cores (which is twice the Dell example). 3200 blades times US$10k per blade is a US$32,000,000 total cost for the blades. The chassis to hold the blades are roughly US$10k each so 3200 blades require 800 chassis to hold them (3200/4 per blades per chassis=800) which adds another 8,000,000 to the cost. The total hardware cost is thus US$40 million to build a brand new network of 32000 IP global caching servers running on 3200 blades in 800 chassis in 100 plus countries every year ($50K times 800). Add the other costs of running a business, personnel, open source free caching software, open source free LINUX and monthly bandwidth and collocation charges. The point I am making is that it does not cost that much to build a network for an AT&T and Level 3. So the Akamai valuation cannot any longer be based on the cost of its server network or its server software and math, it is only based on the realized revenue and net income. It 2007 Akamai reported US$636.41 million in revenue and a 100 million in net income. If I was a fly on the wall at AT&T or Level 3: they each already have the head count to easily support another 3200 servers, so they don't need any more people, they already have the bandwidth so they don't have to buy that,they already have the colocation facilities to house the 800 chassis so they won't have that cost. Both AT&T and Level 3 could be thinking that all we need to do is spend $40 million for servers and start cutting into the customer list that Akamai provides on their website. If they each get 10% of Akamai business or $60 million in revenue, they have paid for their server investment in the first year and added customers that may buy other services. This is a good business for AT&T and Level 3. This however is not very good for Akamai, which could lose 120 million in revenue. Akamai reported in 2007 US$100 million in net income so losing $120 million in revenue in a short period would not be very good and would have a headcount impact at Akamai.ABOUT THE COMMENTS:So now to answer geogia07’s comment that Akamai is diversified;>“With all due respect, you don't know what you are talking about. Long term, Akamai is much, much >more diversified than any other CDN. Many aspects of their business are unique and very difficult to >emulate. “If they lose 20% of their business to AT&T and Level 3 suddenly, any diversification they have will not matter in the long term, because they will be crushed in the short term and the customers will move off their network if they are concerned about viability or head count loss.>”Further, again with all due respect, owning the "network" has nothing to do with the viability, >security, and cost basis of their business. The entire premise of ANY CDN is that they help other >network operators by lowering their (ISP, etc) network costs. The result is that ISPs are incented to >deploy Akamai's CDN boxes which are deployed inside the ISP's firewall.I agree with Geogria07 that ISPs use Akamai right now because it saves them some upstream bandwidth costs by caching content and because some of them get a check. When you start adding 100 gigabit network interfaces into the ISP networks the cost per upstream bit will come down. The need to cache static http content to save bandwidth costs won't be real anymore and the backbone networks will be so fast that bigger hosting companies (like IBM's new green datacenter facilities) will be able to do a better job. This is not dialup anymore. PTP movie downloads are not cached in Akamai. The only content worth caching for the ISP in the future may be videos, which are more randomly used and cache differently. These types of cached content will be effectively handled at massive CLOUD CACHING SITES like what IBM, GOOGLE and AMAZON are building and that may require Akamai also face them as competitors.>“:Finally the DNS comment isn't accurate either. The entire internet uses DNS while in theory Akamai is >vulnerable they are no more/less vulnerable than any other company. In fact you could argue they are>less vulnerable because Akamai's machines only do a specific function and don't have direct interaction >with consumer machines.” I would argue that DNS is a much bigger problem for Akamai because their whole business depends on it working properly. It is precisely because they do a specific function that when their DNS servers do not work properly the missing partial page content that exists on their cached servers will prevent the rest of the page from loading properly from the origin customer site. They don’t recover and the user gets a blank page. Their servers roll over the DNS CNAME entries constantly for a hidden cached content domain like ebaystatic.com so that they can shake off hackers attacking them. This up and down visibility leaves the content connection in flux and unstable. This is where load balancing at the origin site is a better solution and produces a better customer experience.As for the last comment about Akamai servers "don't have direct interaction with consumer machines" it is clearly false. Simply download from http://www.nirsoft.net the free software called CurrPorts and you will see very quickly, what Akamai cached sites your computer is using and how often. In fact, if you are having problems reaching a site and you open this software, you can see what Akamai servers are hung up.I recommend downloading IPNetinfo from the same site to look up the IP numbers; you can link the two programs. Frequently Akamai does not properly label their or their advertising partners reverse DNS entries and it is a real problem when troubleshooting connectivity issues or for cyberdefense.
No....sorry, you don't understand the company - although you do have an enviable knowledge of routers and systems. Just 2 quick points:1) Akamai is the only company offering performance improvements, failover, and enhanced security for mission critical business applications in the internet cloud. Their Application Performance Solutions (APS) solutions business is the fastest growing part of their business, and comprises about 40% of their revs today. They are the perfect SaaS platform, and the perfect platform for true cloud computing. Forget CDN, it's the commodity side of their business, and it's only one facet of what they do. They deliver, protect, and offload business applications - nobody else does that2) No matter how fast the pipes get or how great the routers get, the internet is composed of peering relationships that are based on money exchanging hands between ISPs. As long as those relationships exist, Akamai can provide faster routing through the cloud, better security, failover, and disaster recovery. They are the business platform within the internet itself, and the small thinking CDN competitors aren't competing against that. End of story.
I will add that I have read business plans that intend to use distributed cached storage in novel ways, for example, in a way that can benefit security. There's definitely innovative potential for the use of Akamai's core infrastructure. I guess time will tell.