He Nearly Went
Broke. Twice.
Then He Moved
to Japan and
Rebuilt Everything
From Nothing.

Every issue of The Real How follows the same structure. The Setup. The Near-Death. The Rebuild. The Playbook. The Traction Framework. The Mistake. The Landscape ahead. And your move for the week.

This issue is different from the others in one specific way. Every other founder in this series built something that worked, kept working, and scaled. Ivan Zhao built something that failed, rebuilt it in a foreign country with no team and almost no money, and then watched it become 100 million people's second brain.

That arc — from broke and invisible in Kyoto to a $10 billion valuation where Sequoia's partner said the price was "very painful" but paid it anyway — is the most important story in this entire series for anyone who is stuck right now. Not starting. Not scaling. Stuck. Sitting on a product that is not working. Wondering whether to push through or shut it down.

Zhao answered that question in 2015. He did not pivot. He did not shut down. He did something stranger and harder and ultimately more interesting. He went to Japan and rebuilt the whole thing. What came back was worth more than anything he had been building before.

This is that story. And unlike the others in this series, it is not just about the past. It is about what is coming — what the window looks like right now for builders who understand what Notion understood before the market did. Let's get into it.

The Setup: A Canadian-Chinese Designer Who Wanted to Give Everyone the Power of a Programmer

Ivan Zhao was born in China and grew up in Canada. He moved to San Francisco as a young man looking for a job as a designer and posted a note on Hacker News. His future co-founder Akshay Kothari saw it. They stayed in touch for years. In 2013, when Kothari sold his company Pulse to LinkedIn for $90 million, Zhao was ready to start something himself.

His obsession was an old one. A question that had been sitting inside computer science since the 1960s and had never been properly answered. Computers are among the most powerful tools human beings have ever made. But their full power was only accessible to people who could program them. Everyone else got the surface layer — the applications someone else had already built, organised the way someone else had decided, doing what someone else had decided they should do.

Zhao wanted to close that gap. He wanted to build a tool that gave non-technical people the power to build their own systems, create their own workflows, organise their own knowledge in the exact structure that matched how they thought. Not a word processor. Not a spreadsheet. Not a note-taking app. Something that contained all of those and let the user decide how to combine them.

He was inspired by HyperCard — Apple's 1987 application that let ordinary people build their own interactive documents and databases without writing code. By the early visions coming out of Xerox PARC. By the question of what computers could be if someone genuinely designed them for the way humans think rather than for the way computers work.

The First Near-Death: Building on Shifting Sand. Running Out of Money. Facing the Question Every Stuck Builder Faces.

By 2015, Notion had been in development for two years. They had a product. They had early users. They had poured every dollar they had into building it. And it was not working. The software was built on Polymer, Google's Web Components framework — a technology that was changing fast and pulling the product with it every time it changed. Zhao described it later as building on shifting sand. The platform crashed constantly. Features broke without warning. The codebase had become so unstable that maintaining it consumed most of the engineering time, leaving almost nothing for building the things that would actually make the product better.

The user feedback was brutal. People found it confusing. Some questioned whether there was a need for the tool at all. Traction was essentially zero. The money was almost gone.

Here is where you have to stop and sit with what Zhao was facing. Because if you have been building something that is not working, this paragraph is the one you know from the inside. The product is not good enough yet. You know it. You can feel the distance between where it is and where it needs to be. But you cannot quite see the path from here to there. Your friends are telling you it is time to move on. The sensible thing — the kind of thing that prudent, rational people do — would be to accept the evidence and stop.

He did not stop. But he also did not simply push harder on the same direction. He made a more interesting decision than either of those.

He shut down everything. Fired his employees. Packed his belongings. Left San Francisco, where the costs were enormous and the pressure to become something legible to investors was constant. And moved to Kyoto, Japan.

That decision is the bravest and strangest one in this entire series. Not quitting. Not doubling down. Retreating to a place of silence, lower costs, and complete freedom from every external pressure — in order to figure out what better actually looked like before spending another dollar on the wrong version of it.

The Kyoto Rebuild: 18 Hours a Day. Zero Noise. One Question: What Does Better Actually Feel Like?

Kyoto in 2015 was cheap. Quiet. Beautiful in a way that does not compete with your attention the way San Francisco's ambition does. Zhao coded eighteen hours a day. Simon Last built alongside him. Just the two of them, working in silence in a city where nobody expected anything from them, with no team to manage, no investor updates to write, no press cycle to feed.

They threw out the entire codebase. Not iterated on it. Not patched the worst parts and kept the best. Threw it out completely. The architectural foundation was the problem and there was no way to fix a foundation without rebuilding the structure on top of it.

They rebuilt Notion from the block up. Literally. The core innovation of the new architecture was the block — a single flexible unit of content that could be anything. Text. An image. A to-do item. A database. A linked mention of another page. Blocks could be nested inside pages that were themselves blocks inside other pages. The structure was infinitely flexible because the primitive itself was flexible.

Notion 1.0 launched in August 2016. Still quiet. Still tiny. But the feedback was immediately different from anything that had come before. People who found it understood it on contact. Early users did not just use it. They built their entire professional and personal workflows inside it. They created templates and shared them. They wrote about how they used it. They recommended it to colleagues with the enthusiasm of people who had been looking for this specific thing for years and had finally found it.

By September 2019 — three years after the rebuilt launch — Notion had 1 million users. By January 2020, they raised $50 million from Index Ventures at a $2 billion valuation. Index moved so fast that the deal closed 36 hours after Zhao started looking. By October 2021, Coatue and Sequoia led a $275 million round at a $10 billion valuation, with Sequoia's partner later saying the price was "very painful" — but they paid it anyway because the growth metrics made it impossible to argue.

The Playbook: Five Decisions That Make a $10 Billion Company Out of Two People in Kyoto

The first decision: give people a canvas, not a feature. The entire productivity software industry in 2016 was competing on feature count. Notion's rebuilt version went in the exact opposite direction. It gave users fewer, more powerful primitives and trusted them to build the structure that served them. Every competitor added more. Notion built the thing that let users add their own. The canvas beat the feature list because the user's specific needs beat anyone's prediction of what those needs would be.

The second decision: stay small enough that quality is possible. When Index Ventures invested $50 million at a $2 billion valuation, Notion had fewer than ten employees. When the $10 billion round closed, the team was still remarkably small for a company of that valuation. Zhao and his team did not treat each funding round as permission to hire fast. They treated it as proof that staying small worked. The quality of the product, the precision of the user experience, the consistency of the voice — all of it was only possible because the team was small enough for one person's taste to run through all of it.

The third decision: keep control at all costs. Zhao raised $343.2 million and gave away none of the board. Zero investor board seats. He still owns approximately 30% of the company. No venture capitalist has a vote on what Notion becomes. This is an almost unheard-of outcome at this scale and it was only possible because Notion was profitable before it needed to raise. When you do not need the money, you can set the terms. When you can set the terms, you protect the thing that made the product worth funding in the first place: the founder's specific, uncompromised vision for what it should be.

The fourth decision: let the community build the distribution. Notion's growth from 1 million to 100 million users was not primarily an advertising story. It was a community story. Users built templates — elaborate, sophisticated systems for managing projects, running personal CRMs, tracking habits, managing knowledge. They shared those templates. They wrote about their workflows on YouTube and Twitter and Reddit. They brought colleagues into shared workspaces and those colleagues brought their teams. The product spread the way genuinely useful tools spread — person to person, with the enthusiasm of people who have found something that genuinely works for them.

The fifth decision: catch the AI wave before it was a wave. In 2022, Notion integrated AI into the product. Not as a separate product bolted on afterward. As a native layer inside the workspace that could write, summarise, translate, and analyse within the exact context of each user's specific pages and databases. When AI became the dominant theme of the technology market, Notion already had it running inside the tool that 100 million people used to run their thinking. That is not luck. It is the compound interest of having built the right foundation years before anyone knew what the right foundation was for.

The Traction Framework: How 100 Million People Found a Product That Did No Traditional Marketing

Weinberg and Mares in Traction describe a framework for finding the distribution channel that unlocks your business's next stage of growth. Most businesses get zero channels to work. Notion found one — community building — and ran it so well that the product is now the default workspace for a generation of knowledge workers who discovered it through someone they trusted, not through an ad they were served.

There is a Weinberg and Mares insight that perfectly explains Notion's growth that almost nobody applies to this story: the product itself is the best distribution when people can see what it does before they sign up. Every shared Notion page, every embedded public database, every template someone downloads from a YouTube video description — all of it is a live demonstration of the product's capability. Notion's architecture made sharing native. Not as a deliberate growth hack. Because the product was genuinely more useful when things were shared. The distribution was baked into the design of the thing.

In 2026, that lesson is worth more than it has ever been. The products that spread are the ones where the output of using them is something worth sharing. Not a screenshot of a metric. Not a referral code. The actual output — a beautifully structured knowledge base, a project tracker that makes a team look organised, a personal wiki that captures how someone thinks. That output is the marketing. Build the thing that produces output people are proud of.

The Mistake: Power Without Guardrails Is Paralysis. The Blank Canvas Intimidates Before It Liberates.

The thing that made Notion great was also the thing that made it hard to start using. The infinite canvas is genuinely powerful. It is also genuinely overwhelming for a new user sitting in front of a blank page with no instructions and no clear starting point. The same flexibility that made power users build extraordinary personal systems made first-time users bounce after thirty minutes of not knowing what to do.

Notion's onboarding was, for years, the weakest part of the product. The friction between "I have heard this is amazing" and "I have a working system that serves my actual needs" was measured in hours of setup that most new users were not willing to invest. The churn from this specific friction — powerful tool, steep initial learning curve — was the most important limiting factor on growth for the longest period of the product's history.

Zhao has spoken about this honestly: the blank canvas is a feature for power users and a bug for new ones. The journey from one to the other requires a bridge. Templates were that bridge. Building them earlier, curating them better, and putting them at the front of the onboarding rather than in a gallery users had to discover would have accelerated the growth curve significantly. The lesson is not unique to Notion. It applies to every tool that gives users genuine power over their own structure.

The 2026–2030 Landscape: The Window Notion Saw Before the Market Did — and the One That Is Open Right Now

Here is the thing about building in the years ahead that Notion's story teaches more clearly than any playbook I could write from scratch. The most important products of the next four years will not be built by the people who wait for the market to define the category. They will be built by the people who look at what is currently fragmented, annoying, and held together with workarounds — and build the version that removes the fragmentation before the market has named it.

Notion saw that knowledge workers were using a different tool for every type of content — one for documents, one for wikis, one for tasks, one for databases — and that the switching cost between all of them was destroying the actual thinking. The market in 2016 did not have a name for the problem. It just had the symptoms. Zhao named the problem before the market did and built the solution before the market knew to ask for it. That is worth billions. Every time.

Right now, in 2026, there are several of those same unnamed problems sitting in plain sight. I want to name them for you, specifically, because you are reading this to understand what is possible in the years ahead.

The context collapse problem. The average knowledge worker uses 275 SaaS applications. They switch context between nearly ten different tools per day. AI has made each tool smarter in isolation. Nobody has made the space between the tools work the way Notion made the space between documents work. The product that collapses context switching — that becomes the single place where you can think across all your tools rather than inside each one — is sitting in the same gap Notion occupied in 2016. It does not exist yet. The builder who makes it will own the next decade of knowledge work the way Notion owned the last one.

The AI-native workspace problem. Every existing productivity tool is adding AI on top of a structure that was designed before AI existed. The filing cabinet was not designed for a world where you can have a conversation with your documents. The to-do list was not designed for a world where your assistant can prioritise it and execute the items. The product built from scratch for the AI-native way of working — where the AI is not a feature but the interface itself — has not been built yet at the scale and quality that Notion achieved for the block-based workspace. That product is being built right now by someone reading something like this. It might be you.

The personal data problem. Every platform you use owns your data, your graph, your history of interactions. The product that lets an individual own their own knowledge graph — their connections, their history, their patterns of thought — and then bring AI to bear on it without surrendering ownership to a platform, is a product the market will pay a great deal for in the years ahead. Notion pointed at this. Nobody has solved it completely.

These are not predictions. They are observations about where the friction currently is. The friction is where the product sits. The product is where the company starts. You do not need to build the whole system to begin. You need to find the one place where the friction is worst and make it disappear. Zhao found it in 2015, in a Kyoto apartment, with almost no money left. The tools available to you in 2026 are incomparably more powerful than what he had. The window is wider. The question is the same.

Your Move This Week

In 2015, Ivan Zhao was broke, invisible, and sitting in a Japanese city with a product that had not worked and a team he had to let go. He was 28 years old. Most sensible people would have stopped. He coded 18 hours a day for two years instead. He came back with the thing he had been trying to build the first time. It just took going to Japan and stripping everything away to find out what it actually was.

The product that came back was not better because of more features. It was better because of fewer, more powerful ones. The product that came back was not bigger. It was smaller. The product that came back did not tell you what to build. It gave you the blocks and trusted you to figure it out.

That product is now used by 100 million people to run their thinking. Every single one of them found it through someone they trusted who told them it changed how they worked.

You do not need to go to Kyoto. But you might need to ask the same question he asked in that apartment: not what more can I add, but what fewer, better things would actually make this real.

Next issue: The validation framework I use to test any idea in an evening — before a single line of code, a single investor conversation, and a single dollar of infrastructure.