Dead End: Skype
Abstract
Skype launched on August 29, 2003 and made a verb of itself. Free computer-to-computer voice calls over the internet, at a time when an international phone call still cost real money, spread fast enough that “to skype” someone entered ordinary speech in a dozen languages. eBay bought it in 2005 for $2.6 billion, could not explain why an auction site needed a phone network, and wrote down $1.4 billion two years later. Microsoft bought it in 2011 for $8.5 billion, its largest acquisition to that point, and then spent a decade redesigning it, folding it into other products, and finally building a competitor, Teams, that it liked better. When the pandemic arrived in 2020 and the whole world suddenly needed to make video calls, Skype was the incumbent with the household name, and it lost the moment to Zoom, a company most people had never heard of. Microsoft shut Skype down on May 5, 2025, after twenty-two years, and told the remaining users to move to Teams. Skype was a product that got everything right early and was then owned, in succession, by companies that did not know what to do with it and a company that would rather sell something else.
Kazaa’s Second Life
Skype’s technology was recycled from a lawsuit magnet. Niklas Zennström, a Swede, and Janus Friis, a Dane, had built Kazaa, the peer-to-peer file-sharing network that succeeded Napster as the music industry’s chief enemy in the early 2000s (see The Warez and Filesharing Era). Kazaa’s design pushed the work onto its users’ machines: instead of a central server, well-connected computers became supernodes that routed traffic for everyone else. That architecture was built to be hard to shut down, because there was no center to seize.
Zennström and Friis realized the same trick worked for voice. If users’ own machines relayed the calls, a voice service needed almost no infrastructure of its own, which meant it could be free. They hired four Estonian engineers who had built Kazaa, Ahti Heinla, Priit Kasesalu, Jaan Tallinn, and Toivo Annus, and pointed the peer-to-peer network at telephony instead of MP3s. Skype was the result, developed in Tallinn and launched in August 2003. The name was meant to be “Sky peer-to-peer,” shortened to “Skyper,” and then to Skype when the domain was taken. Skype became one of the first global products built by Estonia’s engineers, and the money and talent from it seeded a generation of Estonian startups.
The pitch was direct: call any other Skype user, anywhere, for nothing, at better audio quality than a telephone. Calls to actual phone numbers cost a few cents a minute through a paid add-on called SkypeOut. For anyone with relatives on another continent, the math was obvious, and Skype spread by word of mouth through exactly those people.
The eBay Mistake
In September 2005, eBay bought Skype for about $2.6 billion. The logic offered at the time was that buyers and sellers would want to talk before completing an auction. They did not. eBay’s business ran on text, ratings, and PayPal, and a voice-calling network bolted onto it found no use. The integration never happened in any way that mattered, and in October 2007 eBay took a $1.4 billion impairment charge, conceding that it had badly overpaid for something it could not fit into its business.
eBay spent the next two years looking for an exit, complicated by the fact that Zennström and Friis still controlled the underlying peer-to-peer technology through a separate company and were in litigation with eBay over it. In 2009 eBay sold 65 percent of Skype to an investor group led by Silver Lake (with Andreessen Horowitz and the Canada Pension Plan) for about $1.9 billion, valuing Skype at $2.75 billion, and the technology dispute was settled as part of the deal. Skype had now been bought, misused, and sold in four years, without ever finding out what it might have become under an owner who understood it.
Microsoft, and the Slow Reengineering
On May 10, 2011, Microsoft bought Skype for $8.5 billion in cash, its largest acquisition to that date. The number was startling for a service that made little profit, and the rationale was strategic: Skype’s brand and its hundreds of millions of registered users would strengthen Microsoft across Windows, Xbox, Office, and phones. Microsoft retired Windows Live Messenger and moved its users to Skype, and put Skype on every platform it could.
Then it began rebuilding the thing it had bought. The distinctive peer-to-peer architecture, the Kazaa inheritance, was expensive to run on mobile devices, which slept and dropped connections and could not serve as supernodes. Starting in 2012, Microsoft replaced the user-hosted supernodes with supernodes it ran itself in its own data centers, and by 2017 Skype was a conventional service running on Microsoft’s Azure cloud. This solved real mobile problems. It also meant that calls now flowed through servers Microsoft controlled, a change noted by privacy researchers, because a centralized service is one that can be wiretapped, and Skype’s name surfaced in the Snowden disclosures about NSA access to communications platforms.
The bigger damage was self-inflicted through constant change. Skype was redesigned repeatedly, each version moving buttons, adding features nobody asked for, and breaking habits. Users who had relied on it for a decade complained that every update made it worse. The product lost the one quality that had made it spread: it just worked, and you did not have to think about it.
The Moment It Lost
By the mid-2010s the ground had shifted. On phones, people made voice and video calls with WhatsApp, FaceTime, and Facebook Messenger, apps already tied to their contacts. In 2017 Microsoft launched Teams, a workplace chat and meetings product aimed at Slack, and folded its business calling into it, retiring Skype for Business. Microsoft now owned two overlapping communication products and had chosen which one it cared about. Skype was left to consumers, under-loved, still being redesigned.
Then came the test. In March 2020 the pandemic sent the world home, and video calling went overnight from a convenience to a necessity. Skype was the incumbent. It had the recognized name, the verb, twenty years of users. Its traffic did jump, from about 23 million daily users to 40 million. And it lost anyway. Zoom, a company most consumers had never heard of, went from roughly 10 million daily meeting participants to over 200 million in the same months. Zoom won because it removed friction Skype still imposed: you clicked a link and you were in a meeting, no account, no app install required, no adding contacts. Skype asked you to create an account and connect to specific people, a model built for calling your family, not for a company or a school class assembling forty strangers on short notice. Microsoft, for its part, spent 2020 pushing Teams, not Skype.
The Shutdown
On February 28, 2025, Microsoft announced that Skype would close on May 5, 2025, and directed users to migrate to Teams, which could import their Skype contacts and chats. Skype had run for twenty-two years. At the end it still served tens of millions, a large number for most products and a shadow of what it had been, and a rounding error next to the audiences of the apps that had passed it.
Skype’s failure is a study in ownership. The technology was clever and early, and the brand became a common word, the rarest thing a product can achieve. What it never had, after its founders sold it, was an owner who wanted Skype to be the best version of itself. eBay bought it by mistake and wrote it off. The investor group flipped it. Microsoft bought it for its users, spent years re-engineering it into something more ordinary, and then built a rival it preferred. When the single largest opportunity in the history of video calling arrived, the company that owned the most famous calling product on earth was busy selling a different one.