One of the highlights of Day 1 at VON Canada was Niklas Zennström keynote speech right before lunch.  Mr. Zennström is the CEO of Skype, the peer-to-peer VoIP client that is taking the world by storm, earning huge coverage in the blogsphere, from Andy Abramson to ZDNet blogs (by authors such as Russell Shaw). 

Despite the skepticism of some bloggers such as Om Malik, the results of Skype over the past year are nothing short of amazing.  According to the numbers in Niklas' presentation, the company is attaining a growth rate of 1000% a year, reaching an installed base of roughly 35 million users worldwide.  Given that Verisign is pegging the global number of Internet users at 900 million, Skype has captured almost 3.89% of the total number of cybernauts out there.  But that also includes users still bogged down by low bandwidth, so a more telling statistic would be the penetration of Skype into the global installed base of broadband users.  A quick scan at ITFacts.biz reveals that 150.5 million users around the globe have broadband access.  This yields a Skype penetration figure of about 23.3 percent, which is pretty good for a company having a product that has been launched less than two years ago.

More importantly, Skype has also been very adept at increasing the take rate of premium services (i.e. getting the subscribers that have been obtaining the free service to pay for extra services such as SkypeOut or SkypeIn).  Thus far, the company is counting mostly on SkypeOut, and is getting a take rate of roughly 3.43% of its users (corresponding to 1.2 million SkypeOut accounts).

Figuring out the annual ARPU for SkypeOut might be a bit tricky, but here are a few datapoints: James Enck figures that $75 per annum is a reasonable figure.  After consulting the SkypeOut pricing list, one can see that the per-minute rates for the U.S., France, Germany, Italy or the U.K. are 2.3 cents US$ per minute.  So the $75 translates to roughly 272 MOU per month, which might be a bit optimistic (granted that the assumption there is that most users make calls to cheap locations instead of India (18.5 cents/min.), a mobile phone in Yugoslavia (28 cents/min.), Vietnam (35.8 cents/min.) or a U.K. wireless number (27.5 cents/min.).  I would suggest 3 hours per month at the "low" rates is a more conservative assumption (want to err on the low rather that high side).  That means $4.14 per user per month or an annual ARPU of $49.68 (let's just round it up to $50 for convenience).  Of course, this can be done in a more scientific way factoring in the top countries by users (Rodrigo Sepulveda Schulz has a list here), but the numbers wouldn't be that far off (Poland has a charge of 3.5 cents/min. and Taiwan 2.9 cents/min.). 

Now, fast forwarding a bit to the short term future, it would not be unreasonable to see Skype get 25% penetration of 200 million broadband subs by the end of this year (about 50 million subs).  Assuming that the take rate for SkypeOut grows to 5%, that yields 2.5 million paying customers by the end of 2005, at an annual ARPU of $50 resulting $125 million in revenues for SkypeOut.  In Niklas' presentation, he mentioned a SkypeIn annual fee of 30 € per user, which translates to roughly $39.10 (let's just use $40 for simplicity).  With a take rate of 2.5% (assume half of SkypeOut due to the fact that it is a newer service), this means 1.25 million customers generating revenues of $50 million for SkypeIn, or a combined total of $175 million for Skype services (for 2005), not including extras such as SkypeVoicemail or headset partner deals.

That is a pretty amazing result, particularly considering the very low marginal cost per call minute and the near zero customer acquisition cost.  How much will this make the company worth?  Valuation depends on the sensitivity analysis of various DCF scenarios, but regardless of the multiple, with this much revenues, the company should be able to invest in R&D and enhance the product even more.  For kicks and giggles, why not partner with an speech rec vendor and bundle in some speech rec features (that would really make Om Malik's discussion about passing the "mom test" academic)?  Cannot use a simple interface to make a call?  Simply say the number you want to call!  Of course the speech engine does not even need to be that sophisticated (it could have a very small vocabulary, which does not add that much to the cost and/or complexity of the application).  But this is just one of many ideas of what Skype can do with the revenue streams.  Another one certainly is to hire top talent.  One sign of good things to come is the hiring of Microsoft guru Lenn Pryor, who left the Redmond,WA software giant to join Skype - a great coup for Skype.

Of course, despite all this optimism, a certain dose of reality is also needed.  For Skype to succeed, peering agreements will be sine-qua-non, and while Skype has the jump on the competition, more of such agreements are going to be required over time.  Moreover, for SkypeIn, Niklas also highlighted in his presentation that the company will need authorization from telecom authorities around the world to offer local numbers.  Thus far, the only countries with SkypeIn numbers are the U.S., U.K., France, Sweden, Norway, Finland, Denmark, Poland, and Hong Kong (ed. note: later, I asked Niklas about why Canada was not on the list, and he mentioned nothing happening yet, but my suspicion is he will probably wait for the CRTC VoIP decision in May before making a move in that direction).