What’s The Barrier To Entry In The CDN Business? A Few Hundred Million

One of the most common questions I get asked from those who track public CDNs in the industry is what the barriers to entry are for new CDNs who enter the space. With so many new content delivery networks popping up in the last 24 months and the technology having evolved quite a bit over the past ten years, it's a fair question.

Today, the online video platforms and the necessary hardware that is required to run a CDN are completely commoditized and delivering video on the web is not as hard as many CDNs make it out to be. That said, it's pretty easy to enter the CDN market with an investment of tens of millions of dollars and offer a solution in the market that gets some decent customers. But that alone is not enough to seriously compete with any of the major CDNs in terms of scale or revenue. While many of the newer CDNs coming to the market always want to say they are going to "challenge Akamai", the fact of the matter is they won't challenge Akamai's revenue, scale or market share, ever. Sure, if you are Level 3 and you put hundreds of millions of dollars into building out a CDN offering, you have a chance over many years to compete with Akamai. But most CDNs aren't raising and spending that kind of money.

It takes hundreds of millions of dollars to become a major CDN in the market, not to mention years to build it all out. It's taken Limelight five years to go from $20M to roughly $150M and taken Level 3 almost three years to go from zero to a projected $100M in CDN revenue. And that's with both companies spending hundreds of millions of dollars on their CDN offerings to compete in this space. It's also why Akamai, Limelight and Level 3 do the vast majority of all large-scale live events on the web, because they have spent a lot of money to have the scale needed. Some want to imply that with CDNs having been around for over ten years now, it's cheaper than ever to build a CDN and you don't need a much capital to do it. I would agree that it is cheaper today to build a CDN than it was five years ago and you get more for your money. But you can't build a CDN to compete with the major players today simply by spending $50M. Even AT&T who spent at least $70M last year to build out their CDN, has roughly less than 15% of Limelight's total capacity today. That's puts things into perspective.

At the Content Delivery Summit last month, Jeff Cohen from Microsoft stated during his keynote speech that it took "several hundred million dollars" to build out a CDN that has the scale and performance that Microsoft needed. While he said you could build a CDN for probably half that, he questioned what kind of performance it would have and what level of service it would provide. That gives you a good indication of just how expensive it is to build, operate and run a CDN, from a company who has spent hundreds of millions in the past three years to build it out. You can't just stick a bunch of boxes out on Internet like some people suggest.

While some are going to suggest that there will be a lot of major players in the CDN market through consolidation, that's not reality. Today, Akamai, Limelight, Level 3 and CDNetworks control more than 80% of the market, globally, based on revenue, for video delivery. And while many of the telcos will enter the market in a bigger way next year, most of them will do so via acquisitions of the bigger players. And if they acquire smaller players, someone like EdgeCast, they will then have to sink a lot of money into the network even after the acquisition to build it out to scale. The market will never have ten major CDN players accounting for the vast majority of the revenue. We're always going to have four or five vendors tops, who make up the vast majority of the video CDN market, based on revenue.

Not every CDN offering on the market needs to be at the scale or size of what Akamai, Limelight or Level 3 offer. There is nothing wrong with having smaller CDNs in the market that focus on small and medium sized customers or are trying to attack a specific vertical. We need them in the market just as much as we need the major players. But when most financial analysts ask questions about the barriers to entry, they are asking from the perspective of new entrants grabbing significant market share or revenue from any of the public CDNs, which is something that's not just not going to happen.

  • Kevin

    I’m curious, with all of these CDN’s selling their services for so cheap, do you think they will ever be profitable or even be able to stay in business? Limelight is not profitable and loses money every quarter (though it is a smaller loss every quarter so if the trend continues they may see profit in 2010.) Now the little guys like SimpleCDN, Panther, EdgeCast (Edgecast may be getting to the not so little guy point)are selling their services for far less than Limelight is. I use Limelight as an example because they are a public company the other CDN’s I mentioned are not public so I don’t know how much they are losing. If Limelight can’t make a profit with it’s enormous scale and charging 20% + more than the others, how could the little CDN’s ever survive?
    I just don’t get it, 85% of the current CDN’s in the market have to go bankrupt unless acquired, I just don’t see how they will ever survive.

  • Andres

    As an answer to Kevin, surviving depends basically on revenue. And this depends on two factors: incomes and costs. Despite EdgeCast or other CDNs are raising less money than LimeLight, maybe their costs are lower (I’m just guessing, don’t have any figures).

  • shaun noll

    great posts as always dan,
    i’m curious why you think it will happen next year though? what makes you put that time frame on it as opposed to this year or in 2 years etc. ?
    “telcos will enter the market in a bigger way next year, most of them will do so via acquisitions of the bigger players”

  • shaun noll

    “85% of the current CDN’s in the market have to go bankrupt unless acquired, I just don’t see how they will ever survive.”
    i would also say that its likely roughly 85% of CDNs (by raw numbers of CDNs, not revenue) have really not sustainable edge or patented software with scale also. so there is also little reason for a big CDN to acquire them. sounds alot like 2001

  • Hi Kevin, no, I don’t think many of the CDNs will stay in business. In 2000 we had almost 60 CDN vendors. In 2002 we had about a dozen, in 2004 we had six. We’ve seen this before. We are already starting to see some go under. Panther Express was out of cash as was Grid Networks. More will be having money problems very shortly.
    This entire business is all about the economics of scale, something I think a lot of people, even those in the industry, forget. CDNs have to push more traffic on their network to lower their overall costs, yet sign up enough customers during that time to help pay for the buildout they need to have to get to scale. Most CDNs will never be able to do that unless that stay small and focus on medium sized customers and don’t try to take on someone like a Netflix.
    To Kevin’s answer, he’s right in that 85% of the CDNs may go under, but they only account for 20% or less of the overall CDN revenue.

  • Hi Kevin, I think the telocs jump into this space next year simply because that is what they keep saying. At the CDN Summit last month, many of the major telcos spoke about how they felt next year would be the tipping point and they would have to jump in with both feet by then. But for all we know it could be two years from now instead of next year. No way to truly know, but not is the time to make acquisitions if you are a telco as the evaluations of the CDNs are lower due to the economy and market conditions.

  • shaun noll

    thanks for taking the time to answer Dan. i agree, the valuations of some of these CDNs is quite low. i know Akamai would be a big bite but damn do they have a cheap valuation and amazing cash flow…..

  • Rob

    Small CDNs can not only survive but do well – but they not if they take VC capital. The problem with many of the companies in the space is that they have taken huge rounds of money and therefore need huge revenue growth to survive. VC’s typically want a 10x return on their money. That means if a company takes $20M then the exit needs to be around $200M.
    This leaves these companies chasing the same customers as Akamai and those customers don’t have a solid reason to switch. Further, there just aren’t that many large media customers in existance. Media companies are vertically and horizontally integrated and there are really less than 20 of them WW that really have the scope and size required to justify some of these huge investments. Therefore, if the startup is unsucccessful in landing a few of these companies in a meaningful way, they fail.
    Worth pointing out too – many of the investments were made when Akamai’s market cap was over $8B. That’s money chasing that type of valuation. Now that Akamai’s valuation has receeded significantly I don’t think you’ll see anymore $50M investments in the near term.
    Small CDNs need a niche and they need to be experts at that niche. They also need to be competent building and running a business that may not have a public offering.

  • Hmmm you may want to look at that number. Lets take a case study of CDNs purchaed by Akamai post dot.com bubble:
    Speedera spent far far less than “a few hundred million”
    Netli I’m sure spent far less than “a few hundred million”
    In fact, I show often provide models to our clients of how much it costs to build a CDN- even if you write all your own software from scratch, and deploy a 25 datacenter international CDN, its going to take far far less than “a few hundred million.”
    If you look at a 10 year operating history, massive scale, and acquisitions, then maybe you’ll be able to come up with that number- but I am not so sure that number is correct for the costs of starting a CDN.

  • Speedera proves my point. Just before Speedra was acquired, they were on a run rate of about $40M a year. So while they didn’t spend hundreds of millions of dollars, they were very small in size and weren’t even doing 10% of Akamai’s revenue.
    If you want to be the size of Akamai, Limelight and Level 3, you have to spend hundreds of millions of dollars. You are not going to get the size and scale of those CDNs by investing $50M into the business.
    No where did I say you have to spend hundreds of millions to “start a CDN”, in fact I said the opposite in the post. You can have a smaller CDN with a lot less investment and there is nothing wrong with that.
    But when those same CDNs are much smaller in terms of network size and revenue but then say they are going to “challenge Akamai”, it’s really not a fair statement from them.

  • Eric

    To support a few of the commenters that the start-up capital costs aren’t as large as one my contend, I think the dynamics are changing in that edge caching is no longer as important as it once was. All of the major CDN players built out their infrastructure when broadband access in the US, Europe and Asia was deficient. When you combine adaptive bit-rate streaming with bandwidth optimization (server-side Fast TCP) you’ll be able to reach the entire world with just 3 data centers. (1) USA (1) Europe (1) Asia.
    Not many consumers will be able to tell the difference between 3 millisecond response and 300 millisecond response.
    With more and more OEM storage vendors offering Cloud Storage on a Bytes stored model, it would allow new entrants to offer origin CDN services with a cloud-based content library streaming directly to consumers without the need for an edge cache. I’m getting 20Mbps from my broadband supplier on the east coast. I get perfect quality of HD video streaming from an origin server hosted in Akanok in CA.

  • jason

    Level(3) does NOT, I repeat NOT do $100M in CDN revenue. They do about a quarter of that. The rest is just customers grouped into the content markets group.

  • Hi, Jason. No where did I say they “do” $100M in CDN revenue, I said they are “projected” to do $100M for 09′. They did $50M last year in CDN revenue from software downloads, small object delivery and streaming, both live and on-demand.

  • U.S. telcos did an estimated $100 million in video transmission business in the U.S. in 2008. That’s amazingly large relative to the worldwide video CDN business. See
    Where do you envision telco video transmission revenue going? Any thoughts on why the share of channel termination revenue has grown and inter-telco-office revenue has shrunk?

  • Hi Douglas, to be honest, I don’t know as I don’t track the video transmission business as it pertains to traditional broadcast signals.

  • Adam

    As opposed to starting a new small CDN and start acquiring customers, would it be a potential opportunity for a private equity firm to come in and consolidate a few smaller CDNs, merge and streamline the integrated networks, expand a little, and then take on the big boys?
    And what I mean is not grabbing significant marketshare, but something along the lines maybe 5-15% of the total CDN revenues?

  • Hi Adam, I think that’s a great question. Personally, I don’t think that would work. Acquiring a lot of companies is all about the execution and merging of multiple platforms, billing, moving over customers, reporting systems, etc… is not easy. Also, once acquisitions happen, the CDN that does the acquisition never keeps 100% of the current business, typically a good portion of it goes somewhere else. So even if a private equity firm consolidated say 3-4 CDNs, all doing $5M each in revenue, you’d still be left with a company doing less than $20M a year after all is said and done, IF it was executed correctly. That’s a lot of work for very little revenue.

  • Adam

    Thanks Dan for the response (gosh that’s quick!).
    So say I have a super secret hardware technology in my back pocket that lets me build out the infrastructure at 60% of the cost. But that’s just the network. In the CDN market today, what can an emerging network company offer above and beyond content distribution?
    What are all the current CDN customers out there looking for but haven’t seen with the Akamai’s, Limelights, and Level 3’s? And are even the smaller CDNs, i.e. EdgeCast with video, addressing any of these market needs?

  • Hey Adam, that’s exactly it, the network is not the biggest cost, it’s the people and all the overhead. Bandwidth is not even 40% of the cost of running a CDN, on average.
    Content owners are looking for help with the ecosystem of video, of which delivery is just one component.