Artificial Intelligence (AI) is reshaping economies worldwide. Globally, AI’s market is projected to grow rapidly and Pakistan’s AI market is expected to expand ~27.6% from 2025–2030 to ~$3 billion. The Government of Pakistan has signaled its intent to participate in this trend. In 2018, the President launched the Presidential Initiative for Artificial Intelligence and Computing (PIAIC), aiming to “revolutionize education, research, and business by adopting cutting-edge technologies” and to make Pakistan a “global hub for AI, data science and cloud-native computing”. The draft National AI Policy echoes this vision: it cites PIAIC’s target of training 100,000 students in AI and cloud computing and states the government’s “clear intention to employ AI for economic growth”.
Despite these bold goals and new initiatives, Pakistan faces significant gaps in infrastructure and skills that could limit its AI ambitions.
• Current State of AI Adoption
Pakistan has taken early steps to spur AI development. Beyond PIAIC, several research centers have been established (e.g. the National Center of Artificial Intelligence (NCAI) and a Sino-Pak AI center). The government’s Digital Pakistan agenda also underscores the importance of a digital economy.
The current administration’s Digital Pakistan Vision pledges to expand broadband access, improve digital infrastructure, and invest in digital skills (the State Minister for IT has emphasized ending the rural–urban digital divide as part of this vision). In 2024, these efforts contributed to notable progress in connectivity; according to the Ministry of IT, some 13 million Pakistanis (including 8 million women) began using mobile internet last year, narrowing the gender usage gap substantially.
These trends suggest growing awareness, but Pakistan’s base of AI-related infrastructure and talent remains shallow. As the UNDP notes, substantial “gaps in infrastructure and AI compute needs” must be overcome and the workforce upskilled for AI development. Currently, most Pakistani universities and labs lack local high-performance compute (HPC): researchers often rely on public cloud services and open-source models for AI work. In short, early programs like PIAIC lay a foundation, but structural weaknesses persist.
• Infrastructure Challenges Pakistan’s AI ambitions are constrained by several infrastructure hurdles.
1. Data Center and Compute Shortages: High-end AI research and deployment require powerful hardware: clusters of GPUs or specialized accelerators in data centers. Pakistan’s existing data-center capacity is tiny by comparison. In fact, the largest data center in Pakistan can handle only about 3 megawatts (MW) of load, and the entire country has roughly 20 MW of data-center capacity. That is far below what major AI projects demand. With so little local compute infrastructure, academia and industry must depend on foreign cloud services. The National AI Policy draft explicitly warns that “limited computational resources are a shared obstacle by various sectors of Pakistan” and urges creation of centralized AI compute facilities. Without new GPU clusters or national HPC centers, Pakistan risks stagnation in AI R&D.
2. Energy Constraints: AI hardware is power-hungry yet Pakistan struggles with an unreliable electricity supply. Over the past year, millions have endured hours-long power outages to conserve fuel, with rural areas hit hardest. Frequent blackouts disrupt training of AI models and even routine development work. The government has recognized this gap; for example, authorities have set aside 2,000 MW of surplus electricity specifically for data centers and AI infrastructure. However, large-scale AI compute still competes with Pakistan’s chronic energy shortfall and seasonal outages. Any national AI cloud or data center strategy will need dedicated, stable power (ideally from green sources) to be reliable.
3. Digital Connectivity and Internet Access: Fast, affordable internet is essential for AI (for data access, cloud services, collaboration, etc.), but Pakistan’s connectivity remains uneven. Nationwide broadband penetration is only around 58% (as of late 2024), and most rural areas fall far below the national average. In cities, multiple providers offer fiber-based service, but villages often rely on outdated 2G or 3G cellular links. Even though 4G networks cover about 81% of the population, this coverage is heavily urban-centric. In practical terms, many small towns and countryside regions suffer slow or intermittent internet, which makes deploying AI solutions (or even remote education in AI) difficult. The “rural–urban digital divide” is widely acknowledged by policymakers. The government’s Digital Pakistan initiatives aim to narrow this gap (for example, targeted programs drove millions of rural women to adopt mobile internet).
Still, gaps remain: analysts note that in some provinces broadband penetration is below 10%, compared to 80%+ in major cities. Without broader high-speed access (fiber, 4G/5G) and reliable power in rural areas, nationwide AI adoption will leave huge swathes of the population out of reach.
• Skill Development Gap
Alongside hardware, Pakistan faces a skills shortage in AI and computing. Employers and policymakers repeatedly note a mismatch between academic output and industry needs.
A 2022 industry report warned of a “national digital skill emergency”: of roughly 25,000 IT/CS graduates each year, only about 10% meet the practical requirements of employers. (The State Bank of Pakistan similarly reported that only one in ten fresh IT graduates are actually employable).
In many universities, curricula emphasize theory and legacy IT topics; few degree programs offer intensive AI courses, project-based learning. Internships and hands-on labs are rare especially outside major cities. Consequently, even top students often lack experience with real data sets or modern AI workflows.
Beyond rote skills, Pakistan also needs researchers and engineers with cutting-edge expertise. Opportunities for AI internships, co-op programs, or industry-led research projects are limited. Some initiatives are emerging – like, Pakistan Software Houses Association (P@SHA) runs “Academia Bridge” courses and TechLift bootcamps to train graduates with in-demand tech skills.
These programs have trained thousands of students, but the scale is still small relative to the need. Meanwhile, few Pakistani universities have dedicated AI or machine-learning labs, and faculty may have little applied AI experience to pass on. The result is a talent pipeline that is growing too slowly and yielding too few industry-ready graduates.
The UNDP observes that closing this skills gap – by training and upskilling the workforce – is “key for AI development” in Pakistan.
• Policy Framework
Pakistan’s own policy framework acknowledges these problems. The National AI Policy (Consultation Draft V1) explicitly targets both infrastructure and skills issues.
It calls for large-scale investment in computational infrastructure and data storage, and for establishing national data standards and repositories. It also sets an ambitious education agenda: for example, the policy states that “Integrating AI into the National Curriculums at all levels is essential” and recommends overhauling coursework to align with global AI demand. These policy goals reflect widely recognized gaps.
Similarly, the government’s Digital Pakistan program (launched in 2018–19) envisions improved connectivity, more digital skills training, and a supportive IT ecosystem. In speeches and press releases, officials reiterate these aims: for instance, the State Minister for IT has affirmed a commitment to “Digital Nation” objectives and stressed the need to eliminate rural–urban divides. Together, the AI Policy and Digital Pakistan initiatives form a blueprint: they identify the problems (compute shortfall, power and internet gaps, skills mismatch) and outline high-level strategies (national AI fund, Centers of Excellence, curriculum reform, industry partnerships, etc.).
• Risks of Inaction
Failing to address these infrastructure and skill shortfalls would carry serious consequences. Innovation will slow if researchers must constantly outsource compute or struggle with unstable electricity. Indeed, experts warn that “owning AI models isn’t enough; the differentiator is who owns the infrastructure that delivers them.” Countries that do not build their own AI data centers risk dependence on foreign providers – losing control of data, models, and strategic capacity. In Pakistan’s case, underinvestment already drives talented engineers abroad. Pakistan is experiencing a severe “brain drain” of skilled youth: in 2022 alone, over 92,000 highly educated professionals (including engineers and IT experts) emigrated for work.
If Pakistani AI graduates find no local opportunities or tools, many will seek careers overseas, further depleting the domestic talent pool. At the macro level, a weak AI ecosystem could make Pakistan less competitive globally: as neighboring countries (India, Saudi Arabia, UAE) invest in sovereign AI infrastructure and local talent, Pakistani firms and researchers may fall behind in areas like fintech, agriculture tech, or public services. In short, the cost of inaction is a perpetuation of low-tech growth and missed opportunities for job creation and efficiency gains.
• Recommendations for Policymakers
To bridge these gaps, a coordinated policy push is needed. Key recommendations include:
1. Invest in AI Infrastructure and Sovereign Cloud:
Pakistan should allocate public funds (e.g. through PSDP or the proposed National AI Fund) to establish domestic AI data centers and high-performance computing clusters. A “sovereign” AI cloud – operated by government or local entities – would reduce reliance on foreign cloud providers and ensure data sovereignty. For example, plans for national computing resources already appear in the AI Policy (ensuring secure access to “computational facilities … for complex R&D”). Policymakers could further incentivize private investment (e.g. tax breaks, low-cost land) in green-powered data centers. Recent moves, like earmarking 2,000 MW of power for tech parks, should be expanded into comprehensive energy guarantees for AI hubs. Over time, building local GPU clusters and edge AI nodes will pay off in faster model development and national security of data.
2. Reform Education and Training:
University curricula must be updated with AI and data-science content. The HEC and provincial education boards should integrate AI modules into computer science programs and professional degrees. Practical labs, project courses, and AI research initiatives should be encouraged. The National AI Policy itself directs that AI be woven into curriculums at all levels and calls for bootcamps and MOOCs to upskill the workforce. Government scholarships and grants (e.g. HEC’s schemes) should target AI specialization. Vocational and technical institutes can add courses on machine learning tools and data analytics.
Expanding successful short-term programs (like PIAIC’s one-year courses) can produce skilled graduates quickly. Equally, incentives for faculty exchange or industry-partnership research labs would improve graduate employability.
3. Forge Industry–Academia Partnerships:
Strong collaboration between universities and the tech industry will help tailor training to real needs. Joint programs – such as internship pipelines, co-developed course content, and sponsored research projects – can give students hands-on experience. Industry bodies like P@SHA are already moving in this direction with bootcamps and the Academia Bridge program, the government can amplify such efforts by matching grants or tax incentives. The government should also leverage international partnerships (as envisioned in the AI Policy) to bring expertise to Pakistan. For instance, Centers of Excellence in AI (being planned for major cities) should collaborate closely with local universities and firms. By aligning academic outputs with the 20,000+ ICT jobs currently unfilled, Pakistan can produce a workforce that meets iv ndustry demand.
4. Regulatory and Financial Incentives:
To spur private sector adoption of AI, regulators can create sandboxes and standards for data sharing and privacy. The AI Policy suggests fiscal incentives for startups and AI-focused SMEs. Policymakers should consider tax breaks, subsidy schemes, or matching funds for companies investing in AI solutions (especially those addressing national priorities like healthcare, agriculture, or fintech). Public institutions (e.g. hospitals, schools, regulatory agencies) can pilot AI projects by partnering with tech firms, creating domestic use cases.
On the regulatory side, clarifying data governance and easing restrictions on computing infrastructure imports will lower barriers. Overall, combining “carrot and stick” – incentives for early adopters and a clear legal framework – will accelerate AI diffusion in both public and private sectors.
Conclusion
In conclusion, Pakistan stands at an inflection point. The government’s vision documents and policies recognize AI’s potential, but realizing that potential will require decisive action. By investing in data centers and compute power, by overhauling tech education, and by aligning industry with academia, Pakistan can bridge its infrastructure and skill divides. Failure to do so risks eroding competitiveness and losing talent abroad. The time to act is now: a cohesive strategy backed by funding and regulatory support could transform Pakistan’s tech ecosystem and ensure it benefits from the global AI wave.
