Andhra is counting its skills before it counts on them
A statewide skill census covering 3.5 crore people, and a deep-tech training programme targeting ten lakh learners. The idea is to stop guessing what the workforce can do.
Andhra Pradesh has launched a statewide skill census, mapping the capabilities of roughly 3.5 crore people aged between 15 and 59. The stated purpose is to inform investment decisions, workforce planning and upskilling programmes with data rather than assumption.
It is an unglamorous initiative, and possibly the most consequential one in the state’s development programme.
The problem it addresses
Every state courting manufacturing investment makes the same claim: a young, trainable workforce. Almost none can say what that workforce can actually do.
Investors evaluating a site want to know whether five hundred people within commuting distance can be trained to a specific standard within a specific period. “Large youth population” does not answer that question. A census that records skills and stated interests at individual level, in principle, does.
It also works in the other direction. If the census shows a district’s workforce concentrated in skills for which demand is falling, that is an argument for intervention before the problem becomes visible in migration statistics.
The deep-tech programme
Alongside the census, the Chief Minister has announced a deep-tech skilling initiative run with IBM SkillsBuild, the Andhra Pradesh State Council of Higher Education, the state skill development corporation and the Amaravati Quantum Valley. The target is two lakh learners in 2026–27, scaling to ten lakh over five years.
The state has also built a unified skilling portal connecting students, training providers and employers, with career guidance and resume support.
The scepticism this deserves
Skilling programmes in India have a poor record, and the reason is consistent: they are measured by enrolment rather than employment. Two lakh learners trained is a number that can be reported. Two lakh learners employed in the field they trained for is a number that usually is not.
The specific risk with deep-tech training at this scale is mismatch. Quantum computing and AI infrastructure generate a small number of highly specialised roles, not a mass employment category. Training ten lakh people for a labour market that needs a few thousand of them produces credentialled disappointment.
The defensible version of this programme is narrower: train enough people to staff the Google campus, the Quantum Valley and the electronics nodes, and train a much larger number in the adjacent, less glamorous skills those facilities actually consume in volume — electrical, HVAC, network cabling, industrial maintenance.
The P4 framing
The government has placed some of this under its P4 policy, which encourages the top 10 per cent of society to support the bottom 20 per cent, combining state welfare with private participation.
Whether that produces meaningful resources or mainly produces a framework is not yet clear. Voluntary redistribution schemes have a thin track record globally.
What to watch
Publication of the census findings, and whether they are made available in enough detail to be independently analysed. A skill census whose results are held internally is a planning tool. One that is published is also an accountability tool — and a genuine asset for any investor deciding between Andhra Pradesh and somewhere else.