The history of internet calling: from a crackling 1995 demo to the browser tab you're using now

Most histories of internet calling start in 1995, but the real origin is earlier. In 1974, researchers at Lincoln Lab and Culler-Harrison transmitted the first test packets of voice data between computers. By 1976 they had held a conference call over a packet network, and by 1982 they had connected that network to the public phone system.

None of this was something you could buy. The internet was a government research project, and placing a phone call over it was off-limits to civilians. But the proof of concept was there: voice could travel as data.

Two early efforts set the stage. In 1991, a programmer named Ron Frederick wrote a tool called nv, short for network video, that sent audio and video over the early internet. Around the same time, a company called VocalTec was building what would become the first commercial internet phone app.

Speak Freely, a free tool anyone could download, also appeared in the early 1990s. The quality was poor, but it worked, and it was the first taste of voice over a network for ordinary users.

The landmark came in February 1995, when VocalTec released InternetPhone. It ran on Windows 3.1, needed a 486 CPU and 8MB of RAM, and charged a per-minute fee on top of the registration cost. Both callers needed the same software, and the audio was, by every account, genuinely bad.

But it proved a point: people would pay for internet calling even at terrible quality, because the alternative, international long-distance rates, was so expensive.

The phone industry noticed, and did not like what it saw. In 1996, a group of US carriers petitioned the FCC to ban internet telephony outright, arguing it was unfair competition. The petition went nowhere, and VoIP was left to develop.

H.323, the first attempt at a standardized VoIP protocol, was ratified by the ITU in 1996. It worked, but it was complex and rooted in the circuit-switched mindset of the old phone network, so it did not scale gracefully.

In 1999, the IETF published RFC 2543, the Session Initiation Protocol, or SIP. SIP took the opposite approach: lightweight, text-based, modeled on HTTP and email, and designed for the internet rather than adapted from the phone network. It would go on to become the dominant VoIP protocol worldwide.

That same year, Vonage was founded and the first hosted PBX systems appeared. The idea that a business could scrap its physical phone system entirely was starting to look plausible.

In 2003, two developers, Janus Friis and Niklas Zennstrom, launched Skype. It was different in two ways that mattered enormously.

First, it was peer-to-peer. Skype's calls went directly between users' computers, with other users' machines helping route traffic when needed, rather than through central servers. That made it hard for carriers to block and kept infrastructure costs low.

Second, it was free between Skype users and cheap for calls to real phone numbers through SkypeOut. The gap versus international phone rates was so large that Skype created a new kind of user: the person who had simply not been calling abroad because it cost too much.

By 2005, Skype had 53 million users. eBay bought it that year for $2.6 billion, and Microsoft acquired it in 2011 for $8.5 billion.

The lesson was not only technical. It was that removing friction from communication creates demand that was not there before. Make calling easy and cheap, and people do it constantly.

In 2004, the FCC under chairman Michael Powell classified VoIP as an information service rather than a phone service. That single decision exempted VoIP providers from many of the state-level regulations and taxes that applied to traditional carriers.

The practical result was lower prices and faster innovation. Free of the regulatory overhead of legacy telephony, VoIP companies could iterate and compete in ways the old carriers could not match.

A year later, in 2005, the FCC added an important condition: VoIP services that connect to the public phone network had to support E911, so calls to emergency services would work and carry location information. It is also why browser VoIP still tells you not to rely on it in an emergency.

The iPhone launched in 2007 and Android in 2008. Almost overnight, VoIP had a new home: hundreds of millions of always-connected devices in people's pockets.

WhatsApp launched in 2009 and added voice calling in 2015. FaceTime arrived with iOS 4 in 2010, putting one-tap video calling on every iPhone. Facebook Messenger added calling in 2014. Each of these pulled tens of millions of people into internet calling who would never have called themselves VoIP users.

The technical turning point came in 2011, when Google, Mozilla, and Opera shipped the first implementations of WebRTC, or Web Real-Time Communication. WebRTC built calling straight into the browser. No plugin, no install, no separate app. Just a browser, a microphone, and a web page.

WebRTC reached W3C candidate status in 2017 and became a full W3C standard in 2021. It is now built into every major browser. Every service that lets you call from a browser tab, from Google Meet to Discord to FreeCallMe, runs on it.

In March 2020, roughly a billion people shifted to remote work in the span of a few weeks.

Zoom went from 10 million daily meeting participants in December 2019 to 300 million by April 2020. Microsoft Teams added tens of millions of daily users in a matter of weeks. Google Meet reached 100 million daily participants. The VoIP infrastructure under all of it, the servers, the TURN relays, the SIP gateways, absorbed a traffic jump no one would have planned for.

The stress test revealed something. WebRTC-based systems scaled better than expected precisely because peer-to-peer connections do not all funnel through central servers. The distribute-the-load-across-endpoints idea that Skype pioneered in 2003 turned out to be exactly right for a world where the endpoints were suddenly billions of home internet connections.

In May 2025, Microsoft shut down consumer Skype, the product that had turned internet calling into a mainstream habit, and pointed users toward Microsoft Teams. In a literal sense it was the end of an era: the app that made international calling free for a generation was gone.

The closure followed a decade of Skype losing ground, to WhatsApp for mobile calls, Zoom for video meetings, and Teams for work. Its architecture, revolutionary in 2003, had not evolved fast enough.

What survived was the behavior Skype trained: the expectation that internet calling should be free, frictionless, and require nothing complicated from the person on the other end.

The VoIP market is projected to pass $236 billion globally by 2028, growing at roughly 11% a year. More than 75% of enterprise employees now use a VoIP or mobile VoIP system as their main business phone.

The technical frontier has moved to AI: real-time transcription, live translation, sentiment analysis, and automatic call summaries are becoming standard. The audio-quality problem that consumed the first twenty years is effectively solved. The Opus codec, 5G, and WebRTC together deliver quality a copper landline never could.

And the browser call link, generate a URL, share it, click it, talk, is the distillation of everything the last thirty years produced. No software, no account, no install. Just a page, and the other person's voice.