
The two-pizza rule makes teams cheap to dissolve.
When Amazon limits each team to the number of people two pizzas can feed, the constraint is organizational liquidity. A six-person team can be disbanded, reassigned, or spun off without triggering the gravitational collapse that happens when you try to break up a thirty-person department. The pizza limit is a forcing function for disposability.
We treat team size as a collaboration problem. We worry about too many people in the room, too many voices in the decision, too many dependencies slowing the work. All true. But the deeper constraint is how cleanly teams can stop existing when the work changes.
Permanent teams optimize for their own survival. They accumulate process, territory, and identity. They develop institutional memory that becomes institutional inertia. A team that's been together for three years has shared context and shared assumptions about what the team is for, what problems it owns, and what success looks like. When the market shifts or the strategy pivots, that shared understanding becomes shared resistance. The team bends the new work to fit the team.
Small teams stay liquid because they never accumulate enough mass to resist dissolution. A four-person squad working on a feature doesn't have time to build an empire. They ship. Then they disband. Then they reform around the next problem. The work defines the team. When Shopify spins off an internal tool as a separate company (like when Shopify Plus became its own business unit), they're harvesting a team that outgrew its original container and giving it a new one before it calcifies.
The conventional wisdom says you keep teams together to preserve institutional knowledge. You invest in team cohesion, you build trust over time, you let people develop deep expertise in their domain. All of this is correct for stable work in stable markets. But stable work in stable markets is increasingly rare. What's common now is work that changes faster than teams can adapt, and organizations that need to reallocate talent faster than traditional restructuring allows.
The cost of permanent things
A permanent team is a bet that the work won't change. That's a reasonable bet in industries where the work is predictable: manufacturing, logistics, regulated services. But in product development, in technology, in any domain where the next opportunity looks different from the current one, permanence is a liability.
Permanent teams aren't bad at their jobs. They're often excellent. But excellence in one domain doesn't transfer cleanly to the next, and the team's identity becomes anchored to the old domain. A team that spent two years building a mobile app has deep expertise in mobile development, user experience, and the specific technical constraints of that platform. When the company decides the next big opportunity is in enterprise integrations, that expertise doesn't disappear, though it doesn't help either. Members can learn the new domain, but they'll learn it through the lens of the old one. They'll try to make the new work look like the old work, because that's where their competence lives.
Amazon's two-pizza teams constantly reorganize. A team ships a feature, the feature goes into maintenance mode, and the team dissolves. Some people move to the next high-priority project, some stay to support what they built, some get reassigned to entirely different parts of the business. The team never becomes permanent enough to develop an immune system against change.
Large, stable teams create a different problem. When you have thirty people working together for years, you can't easily split them up. The social cost is too high, the knowledge transfer is too complex, and the political resistance is too strong. So you don't. You keep the team together and find new work for them, even if that work isn't the highest-value thing they could be doing. The team becomes a resource allocation problem: you have this group of people, they need to be doing something, so you find something for them to do. The work serves the team instead of the team serving the work.
Build teams to harvest
Shopify's approach to spinning off internal tools reveals the underlying principle. When an internal tool grows beyond its original scope (when it starts serving use cases the parent company didn't anticipate), Shopify spins it off as a separate entity with its own team, its own roadmap, and its own business model. The team that built the tool becomes the team that runs the new company.
This only works if the team was small enough to spin off cleanly. A four-person team building an internal tool can become a four-person startup without triggering a corporate restructuring. A twenty-person department can't. The organizational overhead of extracting twenty people, reassigning their responsibilities, and backfilling their roles is too high. So the tool stays internal, the team stays embedded, and the opportunity to create something independent gets lost in the friction of organizational complexity.
The two-pizza rule creates teams that can be harvested. When the work outgrows its original container, you don't have to dismantle a department: you just redirect a small group of people. The rest of the organization barely notices. This is organizational optionality: the ability to pursue new opportunities without having to restructure the entire company first.
The conventional model treats team stability as an asset. You invest in building a cohesive team, you give them time to develop shared practices, you let them mature into a high-performing unit. Then you keep them together because breaking them up would waste that investment. This logic is sound if the work stays constant. But if the work changes (if the market shifts, if the strategy pivots, if a new opportunity emerges), the investment in team cohesion becomes an anchor. You can't easily redirect a mature team to new work without losing the cohesion you spent years building. So you don't redirect them. You keep them on the old work longer than you should, because the switching cost is too high.
Small teams invert this logic. You invest in the ability to reconfigure quickly. A team that's been together for three months can be disbanded without trauma. A team that's been together for three years can't. The shorter the team's lifespan, the lower the switching cost, and the faster you can reallocate talent to the next high-value problem.
Discipline in dissolution
Keeping teams small is the easy part. Actually dissolving them when the work changes is hard. Most organizations create small teams and then let them drift into permanence. The team ships the initial project, then takes on maintenance, then picks up adjacent work, then becomes the de facto owner of that entire domain. Three years later, you have a permanent team that started as a temporary one.
The discipline is in the dissolution. When the work is done, the team disbands. It doesn't reassign to adjacent work or transition to maintenance mode. It disbands. People move to new teams, new projects, new problems. The team's identity dissolves with the work. This feels wasteful. You spent months building a high-performing team, and now you're breaking it up just as it hit its stride. But the alternative is worse: you keep the team together, you find new work for them, and you slowly drift into a world where teams exist to justify their own existence rather than to solve the company's highest-priority problems.
Amazon enforces this through organizational churn. Teams form, ship, and dissolve on a regular cadence. People expect it. The culture normalizes it. When a project ends, nobody's surprised that the team breaks up: it's just how the company works. This only functions because the teams are small enough that dissolution doesn't feel like a layoff. A six-person team disbanding is a normal rotation. A thirty-person department disbanding is a restructuring.
The two-pizza rule creates the conditions for this discipline. Small teams can be dissolved without drama because they never accumulated enough organizational mass to make dissolution traumatic. Large teams can't be dissolved without drama, so they don't get dissolved. They persist. They accumulate. Eventually they become obstacles to the next strategic shift.
What Happens Next?
The organizations that win the next decade will be the ones that can reconfigure fastest when the work changes. The ability to form a team, ship something meaningful, and dissolve cleanly is becoming more valuable than the ability to maintain a team for years.
This doesn't mean every team should be temporary. Some work requires deep, sustained expertise that only comes from years of focus. But that work is rarer than we pretend. Most work changes faster than teams can adapt, and most organizations would benefit from treating team formation as a temporary state rather than a permanent one.
The two-pizza rule is a structural commitment to organizational liquidity. Keep teams small enough that you can discard them when the work changes, and you'll spend less time managing the politics of restructuring and more time pursuing the next opportunity. The work will define the team.