GRIP

India

Artificial intelligence

A public mission exists. The labour market is still mostly application work and services, with frontier-model training a much smaller activity.

Startup EcosystemsIndiaArtificial intelligence

The public programme

MeitY implements an IndiaAI Mission through IndiaAI, an independent division of the Digital India Corporation. The mission’s own description is about compute, datasets, skills and application challenges, including work with public services. That is an industrial policy. It is a programme, not a census of companies, and this page leaves programme targets and prize money in the challenge notices.

A startup that fine-tunes a model for an Indian language or a government form is doing a different thing from a lab training a foundation model. Both can be serious. Calling both ‘AI companies’ without saying which one is how a brief becomes a slogan.

The work that is actually hired

The deep pool in Indian cities is people who can apply models, label data, integrate an API, and sell the result to an enterprise that already had an IT budget. That pool overlaps the services industry, and it is valuable. A research group with the capital and the power budget to train large models is a different, smaller pool.

Bengaluru and Hyderabad hold more of the applied engineering. Delhi is where a public-sector pilot is sold. A pilot and a procurement are different events. Data rules, including what a company may send to a foreign model API, belong in the same note as the demo.

Other sectors

Across GRIP

Elsewhere in the research

The same place, on a related page.

Continue with GRIP

Cities and sectors are different maps.

A sector note says which rule and which customer. A city note says who can be hired. Neither is a ranking.