Engineering India’s next engineer

As artificial intelligence raises productivity and changes the skills employers demand, the old model of hiring large numbers of graduates and training them on the job is beginning to lose its relevance. The next phase of India’s technology story will therefore depend less on producing more engineers and more on transforming how they are taught, trained and assessed

India’s technology sector will cross $315 billion in revenue this financial year, growing at 6.1 per cent. Its workforce will grow at 2.3 per cent. Revenue and headcount, which moved together for three decades, have begun to separate, and that single divergence defines the task before our engineering colleges. India’s emergence as a global software powerhouse stands as a defining achievement of the last 25-30 years. The current transition towards AI represents the next phase of this unfolding developmental trajectory. Rather than a departure from our legacy, it challenges the same foresight that anchored the industry’s inception.

The industry that built modern Indian IT recruited graduates in bulk and trained them in-house, absorbing the distance between what colleges taught and what clients needed, forming a hand-holding safety net. It employs nearly six million people and carries a generation into the middle class. But that model depended on growth being linear in headcount. As AI raises output per engineer, firms will hire fewer people and expect far more of each one.

Capacity is not our constraint. The All-India Council for Technical Education (AICTE) approved 15.98 lakh engineering seats this year, and enrolment is at an eight-year high. The constraint is capability: by the Ministry of Electronics & Information Technology’s estimate, only about 16 per cent of India’s IT professionals are AI-skilled. We are not short of engineers. We are short of engineers who can build with these technologies.

According to the Competition Commission of India, the global AI market has surged from $103.6 billion in 2020 to nearly $289 billion today. Buoyed by deliberate policy, India’s own AI economy has grown from $2.97 billion to $7.63 billion over the same period, on a trajectory to reach $131.31 billion by 2032 at a 42.2 per cent annual clip. With analysts projecting 2.73 million new technology jobs by 2028, our difficulty is not a scarcity of opportunity either. The `10,371.92-crore IndiaAI Mission has taken national common compute past 38,000 GPUs, available at roughly `65 per GPU-hour, alongside several hundred open datasets. That capacity was built for students and researchers as much as for start-ups. Ensuring that every engineering department knows how to access it is the lowest-cost, highest-return intervention available to us today.

The Prime Minister’s announcement on  August 15, that one crore youth will be trained in AI skills over the next year sets the right order of ambition. Achieving a target of this magnitude requires a combination of national targeted skilling initiatives and systemic shifts within academia, which must be anchored in a rigorous definition of what AI proficiency entails.

Here is the core of the problem. A student can pass a machine-learning paper and remain unable to solve a real problem with AI. Capability is built in four stages: understanding the concepts, using the tools competently, building applications that work, and finally solving problems no one has pre-solved. Most programmes deliver the first two and stop. The economic value sits almost entirely in the last two. Closing that distance is not a curriculum exercise; it is closer to coaching, and it requires students to spend sustained time with practitioners, on messy problems, with the freedom to fail on a project without it wrecking a grade.

Three institutional changes would do most of the work. Curricula must be designed with industry, not merely shown to it. The partners are already at the door: more than 90 per cent of India’s leading Global Capability Centres work with universities on talent pipelines and joint research. That relationship should be converted from guest lectures into something structural; practitioners sitting on curriculum committees, sponsored laboratories, live industry problems set as coursework, and engineers from those firms co-evaluating final projects. Where syllabus revision cycles run to several years, they must be shortened. A curriculum that cannot be updated annually cannot teach this subject. Faculty must be kept current, and funded to stay that way. No department can teach beyond the experience of its teachers. Industry immersion, sabbaticals into technology firms and continuous upskilling should be treated as core institutional expenditure and built into approval norms, not left to individual initiative. This is the least glamorous of the three changes and probably the most decisive. Assessment must measure what students can build. Employers have already moved: approximately 40 per cent now say they prefer demonstrable AI skills or certifications to a degree. Universities should follow, giving formal academic weight to working prototypes, deployed applications, portfolios and research output. Assessment ultimately determines what institutions teach and what students trouble to learn. Reform it, and curriculum and pedagogy follow on their own.

At the same time, a monitoring framework must shift from auditing infrastructural inputs-such as enrolment figures or tool availability-to quantifying graduates’ demonstrable proficiency in executing AI deployments. We must define success by the substantive results of applied student innovation, ensuring our colleges are measured not by the volume of their recruitment, but by the calibre of their creative output.  AI is reshaping manufacturing, electronics, energy, biotechnology and aerospace, and our ambitions in semiconductors, space and clean energy will need engineers who can work across disciplines and niche skills such as designing chips. India, as the Prime Minister has said, “must not merely consume technology but create it”. What is needed is another UPI moment, this time with AI. UPI showed that India can drive a technological transition at a national scale by setting a common architecture and a clear direction rather than a uniform template. Engineering education needs precisely that: industry-aligned curricula, practitioner-led teaching, applied projects, real apprenticeships and portfolio-based assessment.

The first revolution was built on scale, and India delivered it. The second will be built on capability. An engineering degree takes four years; the demand is here now. The arithmetic leaves us no room for delay.

Curricula must be designed with industry, not merely shown to it. The partners are already at the door: more than 90 per cent of India’s leading Global Capability Centres work with universities on talent pipelines and joint research

The writer is a Member of Parliament (Rajya Sabha) and former Foreign Secretary of India; Views presented are personal.