How a 24-Year-Old Broadcast His Entire Life on the Internet From a Camera Strapped to His Head and Built the Company That Became Twitch

On March 21, 2007, someone called the San Francisco Police Department and reported a stabbing.

The address they gave was a two-bedroom apartment in the city where four young founders lived and worked in shifts so that someone was always awake to manage a website.

When the police arrived, they found a 23-year-old Yale graduate sitting at his laptop in a room scattered with clothes and paper coffee cups and the general debris of a post-college life. Hip-hop was playing. The apartment looked like what it was: the home of four young men who did not sleep at normal hours.

The graduate threw his hands up.

He was wearing a small cylindrical camera strapped to his head. A cluster of cellular modems in a backpack nearby was keeping the camera connected to the internet at all times.

An officer shone a flashlight around the room and asked if someone had been stabbed. The founder, panicked, said no, and then blurted the only explanation he could think of.

"Uh, tech company."

The police had not come for a crime. They had come because someone in his online audience had called in a false report as a prank. Because this man's life was being broadcast live to anyone with an internet connection, 24 hours a day, 7 days a week, with the address publicly visible on the site.

The man with the camera on his head was Justin Kan. The company was justin[dot]tv. The year was 2007.

Seven years later, Amazon paid $970 million for what grew out of that experiment.

The Setup

Justin Kan was born on July 16, 1983, in Seattle, Washington.

He grew up in a household that valued academic achievement. He was drawn to technology and to the kind of thinking that sat at the intersection of ideas. He attended Yale University from 2001 to 2005, graduating with a double major in physics and philosophy.

He did not go to Yale to learn to be an entrepreneur. He went because he was genuinely curious and the curriculum matched that curiosity. But Yale is where he met Emmett Shear, who would become his co-founder across two companies.

While other seniors were applying for jobs, Kan and Shear were building things. They co-founded Kiko, an AJAX-based online calendar application, during their final year at Yale. It was their first company and they got into Y Combinator's inaugural Summer 2005 batch. One of the first groups Paul Graham ever funded.

Kiko had real technology and genuine users. Then Google launched Google Calendar for free. The market Kiko was building for essentially evaporated overnight.

They sold Kiko on eBay in 2006 for $258,000. Not a failure in any dramatic sense. Just a first company that lost its market to a company with resources neither of them could compete with.

The proceeds went to their next experiment.

That experiment was a camera strapped to Kan's head. His entire life. Live. On the internet. Forever.

Most people who hear this idea immediately understand why it was unlikely to work. The more interesting question is why they did it anyway. And what they found out when they did.

The Constraint

justin[dot]tv was founded on October 1, 2006.

The technology required to broadcast live video continuously from a moving person did not exist commercially. Kyle Vogt, who dropped out of MIT to join the team, built it. A custom camera. A battery pack. A cluster of cellular modems in a backpack that maintained a live connection even when Kan was walking down the street, eating in restaurants, taking the subway, sleeping.

The constraint was not technology. Vogt solved that.

The constraint was the idea itself.

Kan and his co-founders genuinely believed that lifecasting, broadcasting your entire life continuously, would become a mainstream behaviour once someone proved it could be done. They thought they were building the infrastructure for a new form of human communication and that Kan's experiment would be the proof of concept.

They were almost entirely wrong about this.

Most people were not interested in watching someone else's ordinary life on the internet. The novelty wore off. The pranks from the audience increased. The content was uncontrollable because it was reality. Someone called in a false stabbing report. Someone called in a false fire report. The San Francisco emergency services put justin[dot]tv's number on a special list requiring confirmation before responding to any calls.

Kan's lifecasting lasted eight months. As a mass-market behaviour, lifecasting never became what they predicted.

But something unexpected was happening inside justin[dot]tv while the lifecasting experiment was running. Something that would take them years to understand fully.

The gaming content was growing. Spontaneously. Without any effort or strategy from the team.

Gamers were discovering that the justin[dot]tv platform, which had opened for public broadcasting in October 2007, let them stream their gameplay to an audience. And that audience was not the distracted, fragmented audience of general-purpose internet content. Gamers watching other gamers play were engaged, specific, loyal, and growing at a rate that dwarfed every other category on the platform.

By 2011, gaming was not just the most popular category on justin[dot]tv. It was the entire story.

The Opportunity

The insight that created Twitch was not strategic planning. It was observation.

Kan and his co-founders had spent four years building a general-purpose live streaming platform. They had dozens of content categories. Sports. Music. News. Lifestyle. People. Every category had users. None of them had the engagement and loyalty and growth trajectory of gaming.

Gaming on justin[dot]tv in 2010 was something that looked, from inside the company, obviously special. The numbers were different. The community behaviour was different. The way audiences participated was different.

Esports tournaments were being broadcast live and drawing thousands of concurrent viewers at a time when nothing else on the platform could approach those numbers. Professional players were building audiences that came back daily, week after week. Viewers were chatting with each other and with the streamers in real time in ways that created genuine community, not just passive consumption.

In June 2011 Kan and Emmett Shear made the decision. Spin off the gaming category into its own platform. Give it its own brand. Its own focus. Let it develop in the direction it was clearly already going.

They called it Twitch. Officially launched in public beta on June 6, 2011.

Within one year it had over 20 million unique monthly visitors. By 2013 it had 45 million unique viewers monthly. By early 2014 it was the fourth-largest source of peak internet traffic in the United States, behind only Netflix, Google, and Apple.

Google tried to acquire it for $1 billion in May 2014. The deal fell through over regulatory concerns.

Amazon acquired Twitch in August 2014 for $970 million in cash.

The entire journey from Justin Kan strapping a camera to his head in a San Francisco apartment to Amazon writing a $970 million check took seven years.

The Playbook

He committed to the weird idea completely before the market validated it.

This is the first and most important principle in Kan's story and the one most people resist.

Kan did not test the lifecasting concept with a survey. He did not build a minimum viable version and measure engagement metrics. He built a custom camera and a mobile broadcasting backpack, moved into an apartment with his co-founders, and broadcast his life 24 hours a day.

He had a theory. The only way to test the theory was to actually do the thing at full scale, not to do a reduced version of it and measure the response to the reduced version.

Most builders test a diminished version of their idea, get a diminished response, and conclude the idea does not work. Kan tested the real idea. He got a clear answer. The clear answer was not what he expected. But it was real, and it came with data he could actually use.

The clear answer was: people are not interested in watching someone else's ordinary life. But people are extremely interested in watching other people do things they are passionate about at a high level of skill. Gaming. Cooking. Music. Sport. The performance of expertise in real time.

That answer was only available because he went all the way in.

He opened the platform to the public before he knew what the public would do with it.

In October 2007, justin[dot]tv stopped being a single channel and opened to anyone who wanted to broadcast.

This was not a planned feature release. It was a pivot made because the single-channel lifecasting experiment was clearly not going to become the mass-market behaviour they had predicted.

The decision to open the platform was the decision that allowed the discovery. Once millions of people had access to the broadcasting infrastructure, the use cases no one had predicted emerged. Gamers discovered that the platform let them stream to an audience. The gaming community did what gaming communities do: they organised around it, talked about it, grew it, and made it their own without any intervention from the company.

Kan did not build Twitch by inventing it. He built it by creating the conditions for the right use case to emerge, watching what happened, and then following the evidence.

He followed the data even when it contradicted the original thesis.

This is the move that separates good founders from the ones who hold on to the original idea long after the evidence has spoken.

The original thesis was: lifecasting will become a mainstream behaviour. The evidence said: it will not. Most people, having raised money on a thesis and having committed publicly to that thesis, would not abandon it at the first sign of contradiction.

Kan and Shear looked at the gaming numbers and chose the evidence over the thesis.

They did not make this decision immediately. It took four years of watching the gaming category grow before they committed to it fully by spinning off Twitch. But they made the decision. And they made it in the right direction.

The willingness to follow evidence against your own conviction is one of the rarest and most valuable skills in building anything. It requires the ability to separate your identity from your hypothesis. To say: I was wrong about this, and that does not mean I am wrong about everything.

He built infrastructure first and trusted the use cases to emerge.

The broadcasting technology Vogt built for Kan's lifecasting experiment was genuinely sophisticated. A mobile live video streaming system that worked in 2006 was not a simple thing to build. The team invested heavily in the technology before they had any proof of the business model.

That infrastructure investment looked like over-engineering at the time. It turned out to be the thing that made every subsequent pivot possible.

When gaming audiences needed the ability to broadcast gameplay reliably and at scale, the infrastructure was already there. When Twitch needed to handle millions of concurrent viewers, the foundations had been built for a much more demanding use case than gaming.

The lesson is not to over-engineer your first product. The lesson is that the infrastructure you build for one use case will often enable a use case you did not predict. Do not optimise so narrowly for your current thesis that you cannot recognise the opportunity sitting inside your own technology.

He was willing to kill the parent to save the child.

In 2014, justin[dot]tv shut down permanently. The parent company rebranded as Twitch Interactive to focus entirely on the platform that had outgrown its origin.

That decision, ending the experiment that started everything to protect the thing it had accidentally created, is one of the clearest demonstrations of founder discipline in this entire story.

Most founders cannot do this. The original idea carries too much identity, too much history, too much of the original narrative of who they are and what they set out to build. They keep the original alive long after the evidence says it is time to let it go.

Kan let it go. He shut down the thing he had strapped to his head and broadcast his life through for eight months. Because the thing that mattered was not the original experiment. It was what the experiment had made possible.

Tools Used

A custom-built mobile camera and backpack broadcasting system that Kyle Vogt designed from scratch. Not available commercially. Built specifically for this experiment because nothing that worked existed.

A cluster of cellular modems providing continuous mobile internet connectivity before reliable 4G existed. The technical solution to the constraint of live mobile video was solved by the team before the product existed.

Hacker News and tech media for early distribution. The novelty of the concept attracted significant press coverage within weeks of launch. Kan appeared on the Today Show with Ann Curry on April 2, 2007, wearing the camera on his head.

Open platform architecture from October 2007 onwards. The decision to let anyone broadcast on justin[dot]tv was the infrastructure decision that allowed gaming to emerge.

Twitch Partner Program launched in 2011, allowing successful broadcasters to share in advertising revenue. This was the mechanism that created professional streamers and incentivised the best gaming content creators to commit to the platform.

Timeline

1983, July 16: Justin Kan born in Seattle, Washington.

2001 to 2005: Attends Yale University. Double major in physics and philosophy. Meets Emmett Shear. They co-found Kiko, an online calendar application, during senior year.

2005: Y Combinator's inaugural Summer 2005 batch. Kiko accepted. First funding.

2006: Google launches Google Calendar for free. Kiko's market evaporates. They sell Kiko on eBay for $258,000.

2006, October 1: justin[dot]tv formally founded. Kyle Vogt builds the mobile broadcasting system. Michael Seibel joins for business development.

2007, March: justin[dot]tv launches. Kan's 24/7 lifecasting begins. Police arrive at apartment after audience member files false emergency report. National media coverage follows.

2007, April 2: Kan appears on the Today Show with Ann Curry wearing the camera.

2007, October: justin[dot]tv opens to public broadcasting. Anyone can create a channel and broadcast live. Gaming begins emerging as the fastest-growing category.

2008: justin[dot]tv reaches 1 million registered users.

2010 to 2011: Gaming consistently the largest and fastest-growing category on justin[dot]tv. 17 million monthly unique visitors. Decision made to spin off gaming into dedicated platform.

2011, June 6: Twitch officially launches in public beta.

2012: Twitch raises $15 million in venture capital. Signs deal with CBS Interactive for advertising sales. 20 million unique monthly visitors.

2013: Twitch raises additional $20 million. 45 million unique monthly viewers. Becomes profitable. Gaming represents 91% of its content. Fourth-largest source of peak US internet traffic.

2014, May: Google attempts to acquire Twitch for approximately $1 billion. Deal collapses over regulatory concerns.

2014, August 25: Amazon acquires Twitch Interactive for $970 million in cash. Announced as Amazon's largest acquisition in its history at the time.

2014: justin[dot]tv shuts down permanently. Parent company rebrands fully as Twitch Interactive.

Mistakes and Lessons

The most important mistake in the Justin Kan story is not one Kan made. It is the mistake most people make when they hear the story.

They conclude that the lesson is about pivoting. That you should start one thing, watch it fail, pivot to another thing, and succeed. That the Twitch story is a story about the value of being flexible.

It is not. It is a story about what happens when you commit fully to a specific thesis, let the evidence speak clearly, and then follow the evidence even when it contradicts the thesis you built the company around.

Pivoting implies a casual change of direction. What Kan and Shear did was something harder. They had built their entire public narrative around lifecasting. They had raised money on that thesis. They had attracted press coverage as the people who were going to make lifecasting mainstream. Every stakeholder relationship they had was built around that story.

Abandoning it required admitting publicly that the original idea was wrong. Not partially wrong. Fundamentally wrong in the prediction that mattered most.

Most founders will not do this. The public commitment to the original idea becomes a cage they cannot escape because escaping it requires admitting failure.

Kan escaped it. The $970 million is the result.

The second mistake was in the early monetisation strategy. justin[dot]tv struggled for years to build a sustainable revenue model around general-purpose live streaming. The platform had millions of users and genuine engagement but the content was too broad and unpredictable to attract the kind of advertiser relationships that produce real revenue.

Twitch solved this by narrowing the audience to gamers. Narrower audience, deeper engagement, much clearer advertiser value proposition. The same content economics that made justin[dot]tv difficult to monetise made Twitch straightforward to monetise.

The lesson: a specific, deeply engaged audience is worth more to advertisers than a large, diffuse one. Narrowing the scope of your platform is not a retreat. It is often the move that makes the business model work.

The Psychology

Three things about how Kan thinks that explain results most founders cannot replicate.

He was genuinely curious about the experiment, not just committed to the outcome.

Kan has described the lifecasting experiment as exactly that: an experiment. He had a hypothesis. He was genuinely interested in whether it would prove out. That curiosity gave him the ability to watch the experiment produce unexpected results without experiencing it as failure.

When the audience started engaging most intensely with the gaming content, he was curious about why, not defensive about what it implied about the lifecasting thesis. That curiosity is what allowed him to see the real opportunity before it was obvious.

Most builders are too attached to their original hypothesis to be genuinely curious about what the evidence is saying. Kan's scientific background, physics and philosophy at Yale, probably helped him treat the company as an experiment rather than an identity.

He was not afraid of looking strange.

Wearing a camera on your head and broadcasting your life 24 hours a day in 2007 was genuinely eccentric. The mainstream reaction was not enthusiasm. It was bewilderment and, frequently, ridicule.

Kan did not seem to care. He had committed to the experiment and he was going to run it properly regardless of how it looked from outside.

The willingness to look strange before the market understands what you are doing is a prerequisite for building anything genuinely new. Every category-defining product looks strange before it defines the category. The people who can tolerate that strangeness for long enough to find out whether the idea is real are the ones who build the categories.

He understood that the asset was the infrastructure, not the use case.

Kan and his team built a sophisticated live streaming infrastructure to serve a single use case: one person broadcasting their life. When that use case did not produce the outcome they expected, the infrastructure was still there. And it was capable of serving use cases they had not anticipated.

That infrastructure became Twitch's competitive advantage. By 2014, Twitch was handling millions of concurrent viewers across thousands of channels. The technology to do that at that scale was not easy to replicate. The moat was not the brand. It was the years of engineering investment made originally for a completely different purpose.

The 2026 Builder Translation

justin[dot]tv launched in 2007. The tools Kan used are irrelevant. The principles underneath them are more applicable in 2026 than they were then.

Commit to the weird idea at full scale before you know whether it works.

In 2026 the cost of going all-in on a specific, unusual idea has never been lower. You do not need to build custom hardware. You do not need a team of engineers. You need enough conviction to test the real idea, not a diminished version of it.

The builder who tests a half-hearted version of their idea gets a half-hearted response and concludes the idea does not work. The builder who commits to the real version gets a clear signal. Kan got a clear signal. It contradicted his thesis. It pointed directly to the thing that was worth $970 million.

You cannot get that signal from a reduced experiment. Only the full one tells you the truth.

Build for a specific community, not a general audience.

The justin[dot]tv lesson that matters most in 2026 is about specificity. A general-purpose platform serving a broad audience is almost impossible to monetise well. A specific platform serving a passionate, specific community is highly monetisable and produces the kind of engagement that compounds.

In 2026, every major platform is general. The opportunity is in the specific. A live streaming platform for competitive chess. A community platform for indie game developers. A content platform for clinical nurses. The specific audience that a general platform serves badly is your entry point.

Watch where the unexpected energy comes from.

Kan and Shear spent four years watching what their users did with the platform before they built Twitch. The gaming energy was visible for years before they acted on it definitively.

In 2026, if you have a product with any number of users, something unexpected is probably already happening. Someone is using it in a way you did not design for. Some segment is growing faster than every other segment without any particular encouragement from you. That unexpected energy is not noise. It is the market telling you something.

Watch it for long enough to understand it. Then build what the evidence is pointing to.

Open the platform before you know what people will build on it.

justin[dot]tv opened to public broadcasting in October 2007 because the original concept was not working as predicted. That forced openness created the conditions for the discovery.

In 2026, the builder who creates infrastructure and then opens it to a community before they know exactly how it will be used is the one who finds the use cases that nobody predicted. This is the principle behind every open platform that has produced unexpected value. The creators always find possibilities the builders did not imagine.

Modern Opportunity Radar

Three real opportunity spaces in 2026 that share the structural DNA of Twitch's origin.

Live streaming communities for non-gaming expertise categories.

Twitch proved that audiences will watch other people perform expertise in real time, for hours, regularly. Gaming was the first category. In 2026, cooking, music production, writing, coding, fitness, and dozens of other skill categories have active audiences who want to watch practitioners at work. The infrastructure for building a community around live expert performance exists. The specific communities are waiting to coalesce around the right platform that serves their needs specifically rather than as one category among thousands on a general platform.

Mobile broadcasting tools for specific professional events.

In 2007, Kan's team had to build custom hardware to broadcast live from a moving person. In 2026, every smartphone is a broadcasting device. The gap is not the technology. It is the platform that serves a specific professional event category well. Live court hearings. Industry conferences. Agricultural site visits. Medical procedures for educational purposes. The technology is universal. The platform that serves a specific professional category with the right features, the right community infrastructure, and the right monetisation model does not exist yet in most categories.

Community-owned content platforms with revenue sharing.

Twitch's Partner Program, which allowed successful streamers to share in advertising revenue, was the mechanism that created professional streamers and made the platform's best content sustainable. In 2026 the creator economy tools to replicate that structure exist for any community builder. A platform that shares revenue meaningfully with its best contributors creates a flywheel that general platforms cannot replicate. The creator has an incentive to grow the platform. The platform grows with them.

How You Can Replicate This

You probably do not have an idea that requires a custom camera and a backpack full of modems.

But you might have an idea that feels too strange to take seriously.

The question worth asking is: does it feel strange because it is genuinely bad, or does it feel strange because it is genuinely new?

Bad ideas feel strange because they do not solve a real problem. New ideas feel strange because the market has not yet formed the vocabulary to describe what they are.

Kan's idea felt strange because nobody had lived their life on the internet before. It was not bad. It was new. The fact that it did not work in the way he predicted does not change that. What it built, the infrastructure and the community and the discovery of gaming as a live content category, was only possible because he went all the way in on the strange idea.

Here is the question for your week.

What idea do you have that you have not pursued because it feels too strange to explain? Not dangerous. Not impossible. Just strange. Just something that makes people tilt their head when you describe it.

Write it down. Give it a fair hearing. Ask what it would look like if you committed to it at full scale. Ask what evidence would change your mind about it and what evidence would confirm it.

Then decide whether to run the experiment.

The people who build the things worth $970 million are usually not the ones who had the most sensible idea. They are the ones who committed to the strange one long enough to find out what was inside it.

Related Playbooks

The Markus Persson playbook covers how Minecraft spread through a community that formed around it without any marketing design from its creator. Twitch's gaming community formed the same way. Both cases show what happens when a product gives a specific, passionate group of people a place to gather and the infrastructure to express what they care about.

The Pieter Levels playbook covers how a Dutch developer built a $3 million business from a problem he personally experienced. Kan's story starts from the opposite direction: a hypothesis about human behaviour that the experiment partially disproved. Both paths produce clarity. The hypothesis test is just faster and more expensive.

Premium Insights

Here is the insight that almost never surfaces in coverage of the Twitch acquisition.

By the time Amazon paid $970 million, Twitch had over 55 million monthly active users, was the fourth-largest source of peak US internet traffic, and was profitable. Amazon was not buying a bet. Amazon was buying a platform with dominant market position in a category that was growing faster than anyone had predicted.

What made that possible was not the Twitch strategy. It was the justin[dot]tv years.

The four years of building and operating a general-purpose live streaming platform from 2007 to 2011 produced the technical infrastructure, the operational knowledge, and the community understanding that made Twitch possible. The team that built Twitch had already solved the hardest problems in live video at scale. The infrastructure that could handle millions of concurrent viewers had been built originally to handle much more chaotic and unpredictable content than gaming.

When Twitch launched in 2011, it was not a new company making its first attempt at something difficult. It was an experienced team with four years of live video operations applying that experience to a specific, well-understood, rapidly-growing use case.

The competitors who tried to challenge Twitch were starting from scratch. Twitch started from four years ahead.

The lesson is about what happens to the knowledge and infrastructure you build in your failed experiments. They are not lost. They are accumulated. The thing that does not work as a product often produces the exact capabilities that make the next thing possible.

Kiko's failure taught Kan how to build and ship software products under pressure. justin[dot]tv's lifecasting experiment taught the team how to build and operate live video infrastructure at scale. Those lessons were not wasted. They were the foundation of the $970 million outcome.

Nothing you build that teaches you something real is ever wasted. It is always the preparation for whatever comes next.

Your Move This Week

Write down the idea you have not pursued because it feels too strange or too hard to explain. Give it fifteen minutes of honest consideration. Ask three questions: Does this solve a real problem that a specific group of people has? What is the smallest version of this I could test at full commitment rather than half-heartedly? What evidence would tell me within 90 days whether this is real? If you can answer all three, you have the beginning of an experiment worth running. That is how Twitch started. Not with a plan. With a strange idea and an honest attempt to find out if it was real.

* * *

Kan did not know he was building Twitch.

He thought he was building the infrastructure for a world where everyone broadcast their life.

He was wrong about that. But he was right that the infrastructure for live human connection at scale was worth building.

The camera on his head was not the product.

It was the question.

And the answer was worth $970 million.