Share with your CTO
David Pan, field CTO at Cursor (the AI coding tool recently acquired by SpaceX for $60 billion), is calling time on Jeff Bezos’ two-pizza team rule. The rule, roughly fewer than 10 people per team, shaped engineering org design for over two decades. Pan’s argument is direct: AI tooling has compressed productive team size so dramatically that even a small group is now oversized for many engineering tasks. Two pizzas, he says, is too much pizza.
What this means for your business
Pan isn’t making an HR observation. He’s describing a structural shift in how engineering output is produced. A three-person team with Cursor, GitHub Copilot, or a comparable AI coding assistant can ship what a seven-person team shipped two years ago. That changes headcount planning, sprint capacity, and how you scope new product initiatives. If you’re still sizing engineering pods using pre-AI assumptions, your org chart is carrying dead weight.
The two-pizza rule was never really about pizza. It was about communication overhead: the number of interpersonal channels in a group grows exponentially with headcount, so smaller teams move faster. AI doesn’t eliminate that dynamic, it collapses the baseline. When an AI assistant handles code review, test generation, and documentation scaffolding, the human coordination burden shrinks before you’ve even reduced headcount. The team that used to need eight people to stay aligned now needs four, not because people were redundant, but because the AI absorbed the connective tissue work.
The question worth holding: does this mean you hire fewer engineers, or that the same engineers now own a dramatically wider surface area? Those are different strategic answers with different risk profiles. A smaller, higher-autonomy team moves fast but creates concentration risk on critical systems. The signal worth watching is whether elite engineering orgs start publishing new internal norms for team sizing, the way Amazon’s six-pager memos and two-pizza rule propagated through the industry after the early 2000s.
Concept deep-dive: Communication overhead compression
In any team, the number of communication channels scales roughly as n(n-1)/2, so a team of 8 has 28 potential pair relationships, while a team of 4 has 6. Bezos built the two-pizza rule on this math. AI tools compress overhead differently: they absorb async coordination tasks (documentation, code review, status summarization) that previously required human-to-human cycles. Think of it as reducing the effective team size without reducing the headcount on paper. The business implication is that your coordination costs are now a design choice, not a staffing constraint.
Based on reporting from Cursor field CTO David Pan says it is ‘goodbye time’ for Jeff Bezos ‘pizza team meeting rule’ that shaped engineering organisations for 20-plus years, as in AI era …, originally published 2026-07-10 00:44:00.
