The findings the whole industry is standing on

Stanford's edge is measurement and framing: it names the category (foundation models), then publishes the numbers everyone else argues over.

Berkeley builds the plumbing. Most of what runs underneath the products in this tracker — the serving layer, the distributed runtime, the instruction set, the open evaluation — started in a Berkeley lab and left as open source. The same instinct produced CRISPR: a biological mechanism turned into a programmable interface anyone can build on.

MIT holds both ends: the founding theory of the field, and the journalism that tells you which of today's claims will still stand next year.

CMU is where robotics became an engineering discipline rather than a demo: navigation, manipulation and autonomy with error bars attached.

Georgia Tech's contribution is behaviour and scale: how robots act in the world, and how machine learning gets taught and deployed at volume.

Caltech supplies the theory that tells you what is physically possible — and the vocabulary the rest of the industry then markets with.

Virginia Tech is where autonomy meets the regulator: FAA-designated test ranges, naturalistic driving data, and the certification evidence products need before they can fly or drive.

Harvard's Belfer Center converts research capability into geopolitics — the country-level view of who can actually build this technology.

Princeton is the field's fact-checker: rigorous work on what AI systems can and cannot do, and on the evaluations that mislead.

UW, with AI2 next door, is the strongest institutional advocate for genuinely open models — weights, data and training recipe together.

Toronto kept neural networks alive through the winter, then detonated the current era with one ImageNet result.

Illinois builds public compute and the software that made the internet usable — the open counterweight to hyperscaler infrastructure.

Michigan owns the proving ground: autonomy claims tested against a repeatable, instrumented city.

UT Austin pairs one of the strongest robot-learning groups with a grand-challenge program on AI ethics.

Europe's strongest legged-robotics lineage, plus a serious push on trustworthy and open models.

Most of China's open-weight model lineage runs through Tsinghua labs and their spinouts — the supply side of the China column in this catalog.

Not a university, but the body that turns research consensus into deadlines enterprises must actually meet.

The first mathematical model of a neuron was written here — the 1943 paper every neural network in this catalog inherits from.

The field got its name at a summer workshop here, along with its founding over-optimism.

Cornell built the first machine that learned from examples rather than instructions — and then took the credibility hit when it was oversold.

The first stored-program computer ran here, and the question of whether a machine can think was posed here in testable form.

Oxford pairs deep vision research with the field's most-cited work on long-run risk and governance.

Bengio's lab produced the attention mechanism and the generative-adversarial idea — two of the three pillars the current era stands on.

NYU is where the field gets audited: benchmarks that expose what models cannot do, and the convolutional groundwork that started the modern run.

EPFL pairs neurotechnology and open-model work — Europe's most credible attempt at building frontier models under public governance.

Cambridge supplies the long-horizon arguments — computability, Bayesian machine learning, and existential-risk framing that policy now quotes back.

Delft did the experiments that turned quantum networking from thought experiment into hardware — and keeps setting the fidelity bar.

Peking supplies much of the open-source Chinese model ecosystem's tooling — training frameworks, alignment datasets and video generation research.

Waterloo owns the retrieval end of the stack and Canada's quantum institute — two things every serious AI product now depends on.

UCLA hosts the first ARPANET node and the causal-inference framework that the current interpretability debate keeps rediscovering.

Hopkins connects language technology to medicine — the two places where evaluation actually has consequences.