AI concentrates twice. It concentrates at the metro level — in specific cities, specific neighborhoods, specific commute radiuses around specific frontier labs. And it concentrates at the university level — in the specific institutions producing the researchers, founders, and technical leaders inside those cities. My firm publishes benchmarks measuring both.

The AI City Index (Volume 01, 2024) mapped where AI capital and talent concentrate at the metro level. The AI Higher Education Index 2026 (Volume 02, just published) maps where AI is produced at the university source. Read together, they explain each other. Read separately, each answers half the question.

Where the two indexes agree

The Bay Area (rank 1 on the AI City Index) contains Stanford (rank 1) and UC Berkeley (rank 4) on the AI Higher Education Index. The concentration is total. Every dimension our data measures for the Bay Area — frontier lab founder pipeline, research output, technical talent density, capital concentration — traces institutionally back through Stanford and Berkeley. This is the most concentrated AI production geography on the planet and the correlation between city and university is essentially perfect.

Beijing (rank 2 on the AI City Index) contains Tsinghua (rank 5) and Peking (rank 7) on the Higher Education Index. Same story, translated. The Beijing frontier AI ecosystem — DeepSeek, Zhipu, ByteDance's AI lab, Alibaba's DAMO Academy — draws its founding technical leadership from Tsinghua and Peking at rates that match Bay Area concentration around Stanford and Berkeley.

Toronto (rank 6 on the AI City Index) contains the University of Toronto (rank 6 on the Higher Education Index). Boston-Cambridge (rank 3 on the AI City Index) contains MIT (rank 2) and Harvard (rank 17). London (rank 4 on the AI City Index) contains Cambridge (rank 11) and Imperial (rank 41). Tel Aviv (rank 5 on the AI City Index) contains Tel Aviv University (rank 28) with dense Technion (rank 25) commute-radius participation.

Where the two indexes diverge

Cities can host universities that outrank the city, and cities can outrank the universities they host. The divergences are diagnostic.

Pittsburgh does not rank in the AI City Index top 15. Carnegie Mellon ranks 3rd on the Higher Education Index. The gap means CMU's AI production is not being fully captured by Pittsburgh's metro-level AI investment ecosystem. CMU graduates disperse to the Bay Area, to Boston, to Seattle; Pittsburgh does not compound the AI production CMU generates into local metro capital density.

New York City ranks in the AI City Index top 10. Its highest-ranked university is Cornell (rank 14 on Higher Education) followed by Columbia (rank 18) and NYU (rank 33). The New York AI ecosystem is stronger than any single New York university would suggest — because it draws from cross-metro talent flow, financial capital density, and Cornell Tech's Manhattan positioning combining with Columbia and NYU output.

Zurich does not rank in the AI City Index top 15 despite hosting ETH Zurich (rank 9 on Higher Education). Similarly, small metros can host disproportionately strong universities without producing metro-level AI concentration. The academic-to-metro conversion is not automatic.

Why the two-index framework matters

Universities produce researchers, founders, and technical leaders. Cities concentrate capital, industry adjacency, and the infrastructure researchers and founders need to build. When a university produces AI talent and its host metro concentrates AI capital, both benchmarks compound. When one produces without the other concentrating, the production leaks — talent flight, capital drain, or both.

Toronto is the working example of the compound-loss risk. Toronto ranks 6th on both indexes — Tier I on university production and comparable metro capital concentration. But Toronto's structural talent flight to San Francisco means the AI City rank compresses relative to what the University of Toronto's production would predict for a fully-retentive metro. Canadian federal AI policy is fighting this compression through Vector Institute investment and Pan-Canadian AI Strategy funding.

Boston-Cambridge is the working example of compound-gain. MIT (rank 2) and Harvard (rank 17) combine with a dense metro AI ecosystem including OpenAI-adjacent research investment, Microsoft Research Cambridge, and the broader biotech and enterprise-AI cluster. The metro converts university production into local capital density at high efficiency.

What this means for policy and strategy

For state governments and metro-level economic development agencies, the two-index framework separates two policy questions. Should we invest in strengthening our host university's AI production? Or should we invest in strengthening our metro's AI capital concentration? The answer depends on which index our metro currently lags on. Pittsburgh's answer is capital concentration. New York's answer is metro-level differentiation from Bay Area competition. Zurich's answer is scale-up capital availability.

For corporate AI talent recruiters, the framework separates where AI talent is trained from where AI talent lives. Stanford graduates who now live in San Francisco score for Stanford (Higher Education Index) and for the Bay Area (City Index). Recruiting strategies need to account for both — the university pipeline and the metro concentration where the pipeline output currently resides.

What comes next

The AI City Index (Volume 03, 2027) will publish next year with an expanded universe and improved methodology drawing on lessons from both prior volumes. The AI Higher Education Index Edition Two, planned for May 2027, will include a "Reshuffle Report" tracking composite-score change between editions. And a growing set of cross-index derived benchmarks — including regional AI ecosystem scores that combine city and university variables — will publish across the coming quarters.

The larger project is to build a coherent measurement infrastructure for the AI-answer economy. Cities. Universities. Companies. Executives. Institutions. Each with a benchmark, each with a methodology, each cross-referenceable. This is what my firm has been building for the past three years and what we continue to publish through Everything-PR and 5W AI Communications directly.

Read both indexes: the AI City Index (Volume 01) and the AI Higher Education Index 2026 (Volume 02). Together they explain more than either does alone.