The modern AI industry is usually described as a stack — foundation models, infrastructure, applications. Underneath that stack sits a denser structure: a network of people, ideas, and capital circulating between a small set of laboratories. Who came from where, and what they built next, often explains more about how the field actually evolves than any product roadmap can.

We have assembled an interactive map of this structure. It contains 333 nodes — companies, people, universities, and investors — connected by 565 edges representing seven relationship types: founding, employment, acquisition, investment, spawning, advising, and education. Four viewing modes let you isolate corporate spin-outs, individual career paths, or academic genealogy independently.

The map is a work in progress. We will continue extending the dataset and welcome corrections from researchers in the field.

Two things become visible quickly. The first is the lineage of frontier labs. Anthropic was founded in 2021 by seven former OpenAI researchers. Mistral assembled its founding team in 2023 from DeepMind and Meta. xAI drew its founding cohort from DeepMind, OpenAI, Google, and Microsoft Research. Thinking Machines Lab, founded in February 2025, is largely composed of senior OpenAI alumni who left during 2024. The map makes the entanglement of these "competing" organisations explicit.

The second is the academic substrate. Many of today's senior industry researchers trace their PhD lineage to a handful of advisors — Hinton, Bengio, LeCun, Ng, and Abbeel chief among them. The intellectual provenance of the modern field is far more concentrated than its current corporate distribution suggests.