There's a great analysis of the Limelight IPO here.
The company is one of a handful that has managed to build a global RCDN (rich content delivery network), and has benefited tremendously from the largess of customers such as YouTube and Brightcove.
However, with a serious legal threat to its business from much bigger competitor Akamai yet to get to court and the gradual loss of its biggest customer as Google rolls out its own infrastructure I can't but help wondering if the company is 'doing a PartyPoker' as it's called in the UK (ie cashing out while the going is good).
Using third party RCDNs is breathtakingly expensive, but does allow capital costs to be outsourced, and as end delivery networks such as Verizon, AT&T, BT and AOL ramp up to the challenges of video delivery, the importance of edge caching and dynamic network replication will diminish in favour of 'networks within networks' (such as those being built by Narrowstep with customers such as Telefonica) and peer to peer models (in fact I'd more accurately use the term 'grid computing').
More important still are the economics of delivering content at high data rates and quality whilst covering the cost through advertising, pay-per-view, subscriptions and other revenue models. Rather than being an overhead, the cost of RCDNs becomes a cost of sale, biting into margins.
That's a lot of risk, but Limelight is one of the few sizable companies from the world of TV 2.0 at present and is almost profitable, so it's not surprised to see names like Goldman Sachs, Morgan Stanley and Piper Jaffray backing the IPO.