Mobile-optimized Skype

There is a clear need for a mobile-optimized service that brings the benefits of Skype while addressing its limitations. Skype has chosen a proprietary technology targeted at a limited market – talking on PCs. The market needs a system that enhances a standard Internet technology to target a very large market – talking on mobile devices. Below is an analysis of Skype’s business model and technology as it relates to the mobile opporunity.

Skype’s market lead is small now and they are fighting against the entire SIP industry.

Skype has 55M downloads, but only around 2M active users. The active users are the actual size of the network; these are the people available to call on Skype at any moment. Most of these users are international and thus its strength is international toll bypass. Only about 10% of these users live in the US.

Even as a portion of the total IP telephony accounts, Skype has only a small market lead. The current broadband phone subscribers are likely in the range of 500,000 worldwide. Free SIP-based services probably total another 100,000. In addition, every Windows XP PC has a SIP client embedded in the included IM software. Cisco has shipped 1M IP phones that are software upgradeable to SIP. Sip.edu is a university project that has provided SIP accounts to reach 80,000 student phones at Yale, MIT, Penn and Indiana. All of the Vonage-like services are competing for mindshare. Vonage is spending $40M this year and will spend $75M next year on advertising. AT&T is likely to spend similar amounts.

Skype’s potential market is limited on PCs.

The vast majority of Skype’s clients are on PCs. In fact rather focusing on broad mobile device support, they have created Linux and Mac versions. However, people do not normally talk on PCs, PCs are not always-on appliances and people are not always near them. The only reasons people make an effort to use Skype is when they are at their PC anyway or they are trying to avoid expensive calling charges.

Skype’s technology is not well suited to mobile devices.

When Skype does attempt to move their software to mobile devices, they will fail or be forced to move away from their p2p technology. The restrictions on mobile devices include high-cost mobile data, loosely-connected mobile data networks, limited battery power, and limited processing power. In short, the mobile world eliminates the idle PC resources that Skype exploits with its p2p technology.

An example of this limitation is Skyp’se first PocketPC client. The client only runs on the most expensive, fastest processor handheld devices. It cannot run on the first dual-mode cellular WiFi device, the HP 6300, due to its high processing requirements.

SIP is a server-assisted p2p technology.

SIP was designed as a p2p technology to empower connected devices to manage multimedia sessions. It is typically deployed with a network-based service to manage routing and policy. However, the network services typically only touch the signaling traffic and not the media traffic. The media traffic is routed peer to peer over the best available connection, unless a proxy is necessary for restrictive firewalls. In a 3 minute voice call, maybe 50 kb are used for signaling and 23 Mb are used for media. Less than 1% of the traffic is not p2p.

Skype is not the only solution for high quality, private, easy installation VoIP services.

Publicly available SIP-based technology can provide the same benefits as Skype, but optimized for mobile devices.

Skype is not cheaper to create.

A quick analysis of Skype spending shows that it’s cost to develop is similar to the cost of purchasing third-party SIP application and feature servers.

Total development cost to date: ~$20M
Total active subs: 2M
Development cost per sub: $10
A typical SIP application server is $5-10 per sub for a highly scalable, reliable and flexible platform on which to develop services. An integrated feature server which includes the application server and features is typically $10-40 per sub depending on target market.

Skype is not faster to develop new features.

For each new substantial feature, Skype must extend the protocol, create the logic for the SuperNode “server” and develop the client.

Connect the Dots

I've spent the past year analyzing the potential for Voice over WiFi. Here are some of the most interesting data points:

1) 70% of cellular usage is indoors, according to Jupiter Research.
2) Broadband penetration is >30% in US households and much higher in high ARPU cellular households.
3) Low cost WiFi chipsets, adaptors and access points are everywhere.

What are the implications for mobile carriers?

On the positive side, Voice over WiFi can improve coverage where people live and work without more spectrum, capital, or zoning approvals. However, if up to 70% of cellular traffic can switch to unlicensed spectrum with WiFi, than what is the value of the licensed spectrum? T-Mobile US may suddenly find itself on a level playing field with Verizon. Looking at the question another way: what would the carriers do if the FCC handed them 70% more licensed spectrum? Would they lower voice pricing to keep the network fully utilized or rollout more 3G with its low margin and unclear payback? Either option is ugly for them.

Worse yet, Voice over WiFi enables any Tom, Dick or Vonage to siphon off up to 70% of cellular traffic since it is as easy for Vonage to provide Voice over WiFi as Verizon. The only missing piece is a good, inexpensive WiFi handset.

When will we see mass market handsets with embedded WiFi?

Carriers certainly have reasons to hold them off but competition will ultimately force their hands(ets). All it takes is one carrier to launch a good solution and the rest will follow. My guess is that there are more than 20 WiFi-enabled handsets currently in development from nearly every vendor. Some of these will even support UMA, a form of Voice over WiFi. Expect several to launch in 2005.

The Southwest of Mobile Networks

All of the press around Intel’s investment in Clearwire and their collaboration on 802.16e should have people thinking. Not about how Intel is greasing the industry to get support, but rather about Intel’s larger vision: defeat Qualcomm at their own game.

Intel’s public announcements tend to focus on fixed wireless and notebook remote access applications. Yet these are very small industries so why bother? With the addition of mobility in 802.16e, presumably you can download your email in your car at 80mph. Maybe that opens the door for next generation Blackberry’s, networked iPods and storage-less digital cameras. Still seems too speculative to bother.

The real story is that Intel is using these nascent markets to get some industry adoption while they complete 802.16e with support for VoIP. According to public documents, Intel expects this in 2007.

The general consensus is that 2007 is too late as WCMDA will have broad adoption globally with excellent device support while 802.16e will just get off the ground. However, I believe the battle is not so much about technology, but rather the business model rationalization that Intel’s insurgents, like Clearwire, will force. Technology is an enabler but the battle will be fought with new business models.

The mobile carrier business model is unsustainable, with high acquisition costs, complex pricing plans, onerous contracts and walled gardens. The incumbent carriers have high operating costs and fight their value chain—handset vendors, content owners, and application developers—to maintain their preeminent position.

The opportunity for Clearwire and others is to create the Southwest of mobile networks. Run the least operating cost network based on a pure IP infrastructure. Outsource legacy PSTN interconnect to others that already have scale. Open the devices at the application and network layers to enable a true consumer electronics business and eliminate the carrier subsidy. Allow anyone to develop and deliver innovative applications and content.

How will they make money? In the near term, they will have high growth with low marketing expenses. In the long term, they will be the arms dealer to the consumer device, content and services giants who will fight the mega-battles to win the consumers wallet.

How to Win the VoIP Wars

As AT&T, Vonage and others are proving everyone right that VoIP is a race to the bottom by lowering prices repeatedly, it's worth thinking about how to create a profitable VoIP company. Since the main value of VoIP providers today is to offer connections to the PSTN, one can draw meaningful comparisons to ISPs, who offer connections to the Internet. Similar services, different networks. Below is a discussion of various business models, the history of their application in ISPs, and implications for VoIP providers.

Least cost provider - As the market for dialup ISPs matured, United Online emerged as the least cost operator. UOL outsources the network services to Level3 to minimize CapEx, markets a simple value proposition through mass media to minimize acquisition costs, and manages intensely its customer service costs. The critical elements in this model are a maturing market since this provider cannot afford to create the category, and scale. This strategy is the likely fall-back plan for the insurgents like Vonage if they fail to achieve their current plans. However, the market is not evolved enough yet.

Mass marketing brand differentiation - The ISP example is AOL. With the largest number of subscribers and highest cash flow, they were able to spend the most on marketing and maintain their lead. This is the classic BCG growth-share model at work. Clearly, AT&T is attempting this model in VoIP. It may work if somebody else doesn’t quickly execute another more powerful model.

Feature differentiation - This is Earthlink's ISP plan. From the beginning, Earthlink offered tools to make it easy for new users. Now Earthlink strives to enhance the users experience and relieve critical inhibitors, such as pop-ups and spam. The trouble with this model is that it only works as well if the company continues to innovate and outpace the competition. In a way, the risks are similar to Hollywood studios or videogame publishers...you are only as good as your last hit. Vonage says that it will create new features that people want and differentiate it from the competition. I am skeptical but wish them luck.

Positive feedback loop lock-in - A positive feedback loop is when the addition of users increases the value of the service and therefore attracts new users. The lock-in comes from closed, proprietary technology. There are no ISP examples anymore as AOL, CompuServe and MSN were forced to open themselves to the Internet. eBay and ICQ are other Internet examples. The VoIP world, this is what Skype is trying to accomplish. However, with the high level of competition today and the active open source community, it will be very difficult to gain the critical mass necessary to execute this model. Skype might need 50-100M active subscribers, which would be 2-5% of all phone connections worldwide, to maintain a dominate position.

Platform lock-in - This is the Windows model and I believe the most likely to succeed model in the VoIP world. It relies on 3rd party developers and content owners to provide innovative solutions for users. The lock-in comes from owning the API's for the developers and having the largest user base for developers to sell into. Applying financial portfolio theory here tells us that the lowest risk approach to creating a killer application is to have a large number of innovators trying (and mostly failing) to create different solutions. The question is how to execute this model. Apple is attempting the same in media with iPod - consumer electronics packaging for the iTunes distribution service. Unfortunately, Apple is so far forgetting why Mac lost and Windows won, but that’s another discussion. Another approach is salesforce.com - software as service with generic devices.

In future postings, I'll discuss what Microsoft, Yahoo and Apple should do with VoIP based on this framework.

Wireless Stupid Network

Among the digerati, it’s accepted that the rise of the stupid network is draining the value out of the wireline incumbents. As voice revenues are attacked by intermodal competition and VoIP insurgents, the profitability of these networks will diminish rapidly. In the near term, Wall Street is focusing on the cash flow generation driven by cost cutting efforts and the revenue impact of wireless and broadband. There are many fallacies in their analysis, but they are well documented.

What strikes me is the lack of attention on the same phenomenon for wireless networks. The current generation of wireless technology is the equivalent of a voice line plus modem line. Internet access is slow and display technologies are weak, leaving the possibility for wireless carriers to control the Internet experience. However, as 3G and 4G networks approach DSL speeds and display technologies allow rich media applications from the Internet, the wireless networks begin to look a lot like mobile stupid networks.

A mobile stupid network has the same properties as a fixed stupid network. The wireless carriers will lose control over services, and VoIP and SMS (the highest revenue per bit application on the planet) insurgents will appear to siphon profitability.

Certainly, the race to the bottom is a longer term process for wireless carriers as there is more value in mobility and adoption is still increasing. As a result, there is not a near term short opportunity here. But wireless investors need to keep in mind that the party is going to end, it’s just a matter of when.

Skype Hype

Skype is free, so how do they make money?

The Skype business plan is to:

1. build a huge captive island of users by leveraging the arbitrage value of free calling around the world, and
2. charge users to connect to other islands--PSTN, SMS, IM, email--and for premium services to be determined.

This makes sense from a capitalist's viewpoint but has a critical failure. People do not use PCs as always-on, always-available devices like they do with phones. The form factor is all wrong. So the advantage Skype gains by utilizing the highly programmable, rich user interface and connected PC is lost due to its bulky, slow starting and stationary nature. People will simply not have Skype near them for enough of the day to create enough value.

The problem solver in you says, "put Skype on a PDA." But, PDAs are not good phones. Take a look at the penetration of PDA phones. The best selling PDA phones have become less like PDAs over time. The design goal of the latest PalmOne Treo was to become more like a phone than its predecessor.

Now you're thinking, "OK, then put it on a mobile phone." This is a great idea. However, during the timeline that matters for Skype, maybe the next 2-3 years, even 3G data networks will not have the capacity or quality necessary to allow Skype to arbitrage the cellular voice networks using VoIP.

Skype has proven again that there is unmet demand for talking at lower prices. To their credit, they have made a system that overcomes many of the hurdles to adoption of VoIP software. Their advances in firewall traversal and use of adaptive codecs are worth noting. They have, as others before them, shown that the integration of IM, presence and voice is valuable. However, Skype is not a "killer" telephony application. They need to think clearly about where they bring value to users, and where not.