Share
Some of the most valuable businesses ever built were built by small teams. Instagram sold for $1 billion with 13 people. WhatsApp sold for $19 billion with 55. Cursor hit $1 billion in annual revenue in 24 months with 300 employees. Peter Steinberger built OpenClaw alone, in three months, and had OpenAI and Meta competing to hire him before the paint was dry.
Every time one of these stories breaks, the reaction follows a familiar sequence: shock, then a scramble to explain it away, then the quiet wait for the next one.
But these companies are not anomalies. They are a pattern. And the three things they share were not invented by the AI era — they were just waiting for it to make the consequences of ignoring them too large to overlook.
1. They hire for density, not volume
The conventional approach to building a team is additive. More people means more capacity. More capacity means more output. It feels logical. It compounds into mediocrity.
These companies think about the quality of each hire relative to the whole. One exceptional person lifts the people around them. One weak hire dilutes the rest. At 13 people, there is nowhere to quietly absorb a poor decision.
Reed Hastings built this idea into the foundation of Netflix. In No Rules Rules, he argued that in creative roles, the best person can be ten times more effective than an average one. Netflix, with fewer than 20,000 employees, generates around $3 million in revenue per employee — roughly double Google’s figure. His argument: a small group of exceptional people, paid at the top of the market and given real freedom, will outperform a large group of adequate ones every time.
Sam Altman has said that OpenAI completed GPT-4 with roughly 250 people, competing against organisations with thousands. He did not frame that as a constraint. He framed it as a strategy.
The lesson for any leadership team is uncomfortable: adding people to solve a capability problem often makes the capability problem worse. Talent density is not about being harsh. It is about being honest — with yourself, about what and who you actually need.
2. Their people believe in the mission, not just the money
Jan Koum built WhatsApp because he believed messaging should be private and simple. When Facebook offered $19 billion, he did not suddenly discover that mission. The mission was why the product was worth $19 billion in the first place.
Peter Steinberger turned down what was likely a larger financial offer from Meta. His reason: he wanted to change the world, not scale a company.
This distinction — missionaries versus mercenaries — was articulated by venture capitalist John Doerr and developed by Marty Cagan in Inspired. Missionaries are engaged, genuinely connected to the problem they are solving. Mercenaries execute instructions without feeling. Both can be technically skilled. Only one builds something worth a billion dollars.
You cannot write a job description that recruits a missionary. You can only create conditions where one would want to stay, and then pay close attention to whether candidates are drawn to the problem or just the package. The signals are usually there. Most hiring processes are not designed to look for them.
3. Nobody has a narrow job
At 13 people, Instagram had no room for specialists who stayed in their lane. Everyone was close to the product, the customer, and the decisions that shaped both. The same is true at Midjourney today — fewer than 100 people, over $200 million in annual revenue.
This is not a cost-cutting measure dressed up as culture. It is a structural feature that large organisations consistently struggle to replicate: people with the full picture, making good decisions quickly, without waiting for permission.
Daniel Coyle studied high-performing groups across completely different fields and wrote about what he found in The Culture Code. The common thread was not individual brilliance. It was the safety and permission to act across boundaries. In small, high-trust environments, people naturally reach beyond their formal remit — and the decisions that emerge are better for it.
General Stanley McChrystal made a related argument in Team of Teams. Rigid functional specialisation fails when conditions change fast. What wins is shared consciousness — everyone knowing enough about the whole to act without waiting for instruction from above. In organisations built for complexity, permission structures are the enemy of pace.
The compounding cost of ignoring this
None of these three things are new ideas. Hastings, Cagan, Coyle, and McChrystal were writing about them well before the current AI wave arrived. What has changed is the cost of ignoring them.
When a team of 300 people can generate $1 billion in annual revenue, the old assumptions about headcount, hierarchy, and specialisation no longer hold. The organisations pulling ahead treat talent density, mission alignment, and role fluidity as strategic priorities — not HR policies to be revisited at the next away day.
The question is whether yours does.
