A million is a round number, and round numbers in policy announcements usually signal ambition rather than a plan. That is not automatically a criticism — targets exist to organise effort. But it is worth doing the arithmetic out loud.

What Was Said

Speaking on 5 November 2026 at the Pakistan-Russia IT Collaboration: Industry Engagement Framework in Islamabad, Federal Minister for Planning, Development and Special Initiatives Ahsan Iqbal said Pakistan aims to develop one million AI professionals by 2030, framing it under the National AI Policy.

Alongside it:

  • Seven new AI centres planned, to link research, skills development, startups and industry.
  • A trillion-dollar economy by 2035 as the broader goal, with e-Pakistan one of five core pillars of the National Economic Transformation Plan.
  • More than 75,000 graduates annually, cited as the base the country is working from.
  • A call to move beyond IT services exports towards products and digital platforms.

At the same event, five Russian companies held 16 meetings with 11 Pakistani firms, organised by the Pakistan Software Export Board and P@SHA.

The Arithmetic

From late 2026 to 2030 is roughly four years. One million people over four years is about 250,000 a year.

Set that against 75,000 graduates annually across every discipline — engineering, medicine, humanities, everything. Even if every single graduate were trained in AI, the target would still require more than three times that number each year.

So the pipeline cannot be universities. It has to be short-course training of people already in the workforce or outside it: upskilling existing developers, converting adjacent technical roles, and running large-scale programmes for non-graduates. That is a legitimate approach and other countries have done it at scale.

It does mean the target should be read as "a million people will complete AI training" rather than "a million people will be employable as AI engineers". Those are very different claims, and the announcement did not define which one is meant.

The definition is the whole thing

Without a published definition of "AI professional", the target cannot be assessed or, later, honestly evaluated. A two-week course on prompt engineering and a master's in machine learning both plausibly qualify under a loose reading.

This matters because the measure chosen determines the behaviour. If the metric is completions, the system will produce completions. If it is people in AI roles earning a living, the system has to produce capability. The second is harder, slower and far more valuable — and it is the one that moves export earnings, which our look at the roadmap to $5 billion covers.

What Would Have to Be True

  1. Trainers at scale. Teaching 250,000 people a year requires tens of thousands of competent instructors. Pakistan does not obviously have them, and the people qualified to teach AI are the same people industry is bidding for.
  2. Compute. Practical AI training needs access to hardware. At this scale that means substantial subsidised cloud or domestic capacity, which is a budget line rather than a policy statement.
  3. Reliable power and connectivity. Both were cited among the reasons IT exports missed the $5 billion target last year. Training programmes face the same constraints as the firms that would hire the graduates.
  4. Demand at the other end. Training a million people into a market that cannot absorb them produces frustration, not an industry. This is the argument for the minister's own point about moving from services to products — products create domestic roles that services outsourcing does not.
  5. Retention. People who become genuinely good at this become internationally mobile. That is not a reason to avoid training them, but it belongs in the plan.

The Fair Assessment

Targets of this kind do useful work even when they are not met. They direct budgets, give universities a mandate, and signal to students that the field is a national priority. Pakistan has existing programmes to build on, including the HEC AI fellowship across 50 universities and curriculum reform work, so this is not starting from nothing.

What would make it credible is detail: the definition, the annual milestones, the budget, and who is accountable for reporting progress. Until those exist, a million by 2030 is a direction of travel rather than a plan — and it should be reported as one.

For a view of where the existing developer base actually sits on skills and pay, our developer survey is the more grounded reference, and our overview of AI in Pakistan covers what is already being built.

Key Takeaways

  • One million AI professionals by 2030, announced 5 November under the National AI Policy.
  • That implies roughly 250,000 people a year against 75,000 total annual graduates.
  • The pipeline must therefore be short-course upskilling, not degrees.
  • No public definition of "AI professional" yet — which decides what the target means.
  • Seven AI centres are planned; funding and governance detail has not been announced.

Written by the WritePures newsroom. Spotted something that needs updating? Tell our editors.