Data First, Robots Second

Imagine a twenty-year-old machine on a Pakistani factory floor. The operator knows from its sound when something is wrong. There are no vibration sensors, no digital maintenance history and no warning before failure. Quality is checked only after the batch is complete. Management discovers energy waste only when the electricity bill arrives.

Upstairs, however, the conversation is about artificial intelligence and robotics.

That gap reveals a weakness in Pakistan’s industrial-modernisation debate. The immediate challenge is not to wait for thousands of gleaming automated plants. It is to make existing factories measurable, maintainable and capable of learning from production data. Smart manufacturing should begin with understanding the process before automating it.

Industry is too important for digitalisation to become a technology-shopping exercise. The Pakistan Economic Survey 2025-26 reports that manufacturing and mining account for 13.5 percent of GDP. Large-scale manufacturing accounts for 67.4 percent of manufacturing and 8.2 percent of GDP. Manufacturing grew 6.6 percent in FY2026, and 16 of 22 large-scale industrial groups recorded positive growth during the July-March period. Yet, performance was uneven: automobiles expanded 61.7 percent and electrical equipment 11.9 percent, while iron and steel contracted 6.3 percent and pharmaceuticals 5.1 percent. Pakistan therefore needs factory-level diagnosis, not a single national shopping list for technology.

Textiles illustrate both an opportunity and a warning. During July-March FY2026, textiles and apparel accounted for 59.7 percent of national exports and 24.2 percent of industrial value addition. Yet, textile production grew only 0.7 percent, while textile and apparel exports edged down 0.5 percent to about $13.6 billion. At the same time, textile-machinery imports rose 20.9 percent. This does not prove that firms bought the wrong equipment; investment takes time to generate returns. But it does show that machinery acquisition and productivity are not the same thing.

A new machine delivers little value if breakdowns remain reactive, operators cannot interpret process data, quality problems are detected after production and managers cannot calculate energy or material loss per unit. In many factories, a temperature sensor, power metre, machine-vision camera, or condition-monitoring system may create more value than a trendy automation project.

Pakistan already has a digital foundation to build on. By March 2026, 34,420 IT and IT-enabled services companies were registered with the SECP. ICT export remittances reached $3.38 billion during July-March FY2026, up 19.7 percent, while freelancers earned $856.3 million. Broadband subscriptions reached 161 million. The paradox is clear: Pakistan has a rapidly expanding digital-services economy, yet much of that capability remains weakly connected to the factory floor.

WIPO’s Global Innovation Index 2025 reinforces this point. Pakistan ranks 99th overall, 124th in innovation inputs and 75th in innovation outputs. It ranks 17th in mobile-app creation and 18th in ICT services exports as a share of trade, yet 129th in domestic credit to the private sector and 128th in ICT access. WIPO’s latest underlying data show gross R&D expenditure at only 0.16 percent of GDP in 2023. Pakistan can produce digital output, but the financing, research intensity and adoption infrastructure needed to diffuse technology across industry remain limited.

This is also the central logic behind my work on intelligent and smart manufacturing. In Intelligent Manufacturing: Concepts and Beyond (Zaman, U.K.U., Kumar, A.A., & Baqai, A.A. (2025)), we treat sensors, intelligent control, production planning, machine learning and industrial decision-making as interconnected elements rather than isolated technologies. In the Handbook of Manufacturing Systems and Design (Zaman, U.K.U., Siadat, A., Baqai, A.A., Naveed, K., & Kumar, A.A. (Eds.). (2023)), we describe smart manufacturing across the product lifecycle and the factory of the future as connected, adaptive, intelligent and sustainable. The practical lesson is straightforward: factories become smarter when data moves reliably from the process to a decision and, where appropriate, back to the process.

Pakistan therefore needs a factory-upgrading program, not another technology showcase.

First, begin with diagnosis. Industrial clusters in textiles, engineering goods, food processing and selected process industries should receive structured digital-readiness and productivity assessments. Multi-disciplinary teams should work with plant personnel to measure downtime, rejections and rework, energy intensity, maintenance losses and production bottlenecks. The objective is to establish a baseline before any money is spent.

Second, co-finance retrofits based on measurable outcomes. Public support should be tied to verified improvements in output, quality, downtime, energy efficiency or export readiness, not merely to equipment purchases. Matching grants, concessional finance or tax incentives can help firms adopt sensors, industrial connectivity, machine vision, energy-management systems and production analytics, but independent verification should determine whether the intervention worked.

Third, Pakistan must deliberately build an industrial systems-integration market. Software companies, automation firms, engineering universities and manufacturers need joint projects on real production lines. Industry should define the problem and provide access to the production process. Technology firms should integrate hardware, software and analytics. Universities should contribute applied research, testing and talent. Public policy should set standards, reduce financing barriers and reward measurable productivity gains rather than memoranda and demonstrations.

CPEC 2.0 can provide an international partnership layer. Pakistan-China co-operation should move beyond equipment imports toward joint retrofitting, local systems integration, training and applied R&D. Technology agreements should embed local engineers in installation, commissioning and troubleshooting teams and create pathways for local firms to maintain, adapt and eventually replicate systems. The relevant question is not how many machines are imported, but how much technical capability remains in Pakistan after the foreign supplier departs.

Pakistan’s National AI Policy 2025 aims to train 200,000 people annually and provide 20,000 paid internships. Some of that talent should be deliberately channelled towards industrial AI, machine vision, predictive maintenance, energy optimisation and production analytics. Otherwise, Pakistan risks developing AI skills without connecting them to one of the economy’s most important productivity challenges.

Pakistan does need modern machinery, automation and robotics, especially where equipment is unsafe, inefficient or obsolete. But sequence matters. If we automate a wasteful process, we do not eliminate the waste; we produce it faster.

The future of Pakistani manufacturing will depend less on how much technology companies buy and more on how well they measure losses, interpret data, adapt technology and continuously improve. Before Pakistan’s factories need more robots, many of them need better data first.