Amazon founder Jeff Bezos’ famous two-pizza rule which is the idea that Amazon meetings should be small enough to be fed with two pizzas has long influenced how the engineering teams are structured. Amazon itself describes ideal teams as fewer than 10 people, arguing that smaller groups minimize communication overhead and foster faster innovation. Now, David Pan, field CTO at Cursor and a former Amazon engineering manager, feels that the metaphor no longer fits, as reported by Business Insider. In a post shared on social media platform X (formerly known as Twitter) Pan called Bezos’ rule “an all-time great metaphor” but declared: “RIP to the two-pizza team. In the AI era, two pizzas is too much pizza.”Pan’s point is that artificial intelligence tools have made teams so efficient that they can shrink even further. What once required a small group can now be handled by just a handful of people, aided by AI assistants.
Read David Pan’s complete post here
“RIP to the two pizza team. Jeff Bezos gave us an all-time great metaphor: keep the team small enough to feed with two pizzas. It shaped how engineering orgs got built for 20+ years. He was right about small teams. Still is. But in the AI era, two pizzas is too much pizza.”
The debate over team sizes
Pan’s post sparked lively debate online. Some users suggested new metaphors like the quarter-pizza team or the three-slice team. Others joked that they could eat the pizzas themselves, even on smaller teams. Not everyone agrees the rule is outdated. Spencer Rascoff, CEO of Match Group, told Business Insider earlier this year that he still supports the two-pizza principle. Meanwhile, some managers argue for larger, flatter teams to spread responsibilities, even as AI enables smaller groups to do more.
SpaceX buys AI coding startup Cursor
Last month, days after taking SpaceX public in the largest stock market debut ever, Elon Musk confirmed that he is buying Anysphere, the startup behind the AI coding tool Cursor, for $60 billion in an all-stock deal. It is the biggest acquisition of a venture-backed startup on record, beaten only by Musk folding his own xAI into SpaceX earlier this year. Cursor had the product and the users but kept slamming into the same wall: compute. Its in-house model, Composer, hit frontier-level performance at a fraction of rivals’ cost, yet the team admitted it was “bottlenecked by compute.”SpaceX sits on the opposite problem. It owns Colossus, xAI’s million-H100-equivalent supercomputer in Memphis, and has been renting the spare capacity out, including roughly $26 billion a year in cloud deals with Anthro, so the moment Cursor and Grok demand picks up, SpaceX can claw the compute back for itself. The two have already been training a model together, due to land in Cursor and Grok Build soon.
