Pakistan’s power sector is entering a paradox that policymakers have yet to fully confront: electricity demand from the grid is falling, yet the cost of supplying it is rising. While the rapid spread of rooftop solar has begun to hollow out the consumer base of the national grid, the fixed capacity payments and persistent inefficiencies keep total system costs elevated. At first glance, the shift toward solar appears to be a zero-sum outcome: households gain cheaper electricity with solar adoption, while the grid loses its most valuable customers. However, the framing remains incomplete and overlooks deeper structural issues.
The conventional narrative assumes that electricity demand rises as economies digitize. However, Pakistan is no longer a conventional case. High tariffs and unreliable supply have triggered a wave of self-generation, with households and firms installing solar panels to escape the grid. This is not just energy conservation; it is strategic exit as well.
The economics behind this exit are straightforward. Electricity systems are built on high fixed costs and long-term contracts. These costs do not fall when consumption declines. Instead, the cost is spread over fewer paying consumers. As higher-income consumers move off-grid, the system is left with a smaller, less stable base. The result is a negative feedback loop: rising tariffs induce further exit, which in turn pushes tariffs even higher. This resembles a coordination failure. Each household’s decision to adopt solar is individually rational; it reduces costs and increases reliability. But when many households do the same, the collective outcome undermines the financial viability of the shared system. What emerges is not a competitive equilibrium, but a suboptimal one, where private incentives diverge from social efficiency. Moreover, the process generates adverse selection: the most creditworthy and high-consuming users exit, making the system harder to sustain.
Within this fragile system, consider the arrival of AI. Globally, AI is energy-intensive. Data centres and machine learning systems require continuous, large-scale electricity. In most economies, this would strengthen the grid by increasing demand and improving the utility of existing capacity. In Pakistan, however, the effect is more complex. Faced with high tariffs and weak reliability, new demand, especially from technologically sophisticated users, may bypass the grid altogether through captive generation or dedicated renewable systems.
This creates a distorted outcome: total demand rises, but the grid does not capture the gains. The marginal unit of electricity is increasingly supplied outside the formal market, while the grid remains burdened with sunk costs. The result is a widening gap between system costs and recoverable revenue.
AI is not only a new player in the system. It is also a potential mechanism designer. Pakistan’s energy problem is not just one of supply, but of coordination, transmission losses, weak billing systems, lack of a competitive market for electricity and poor demand forecasting, which reflect deeper information asymmetries that prevent the system from operating efficiently.
AI can help reorganize the system. By improving forecasting, reducing losses, data streaming and enabling real-time monitoring, it strengthens the grid’s operational efficiency. More importantly, it allows solar and the grid to move from operating separately to functioning as part of the same system. In this context, AI also enables the emergence of retail electricity markets, where households can trade surplus solar power with one another over the grid. This is enabled by AI, which dynamically matches buyers and sellers, prices electricity in real time, and coordinates transactions across the network. However, in Pakistan, such a transition depends critically on regulatory reform, as existing frameworks do not yet support peer-to-peer electricity trading or dynamic pricing mechanisms.
At the same time, AI enables the efficient coordination and operation of virtual power plants (VPPs), where distributed energy resources such as rooftop solar and battery storage are aggregated and managed as a single coordinated unit. This allows the system to smooth fluctuations and potentially reduce peak demand by dispatching distributed supply when it is most needed.
AI can turn the grid into a platform for exchange rather than just supply, ensuring that solar users remain connected as active participants rather than exiting the system. Crucially, because all transactions take place through the grid, fixed system costs can be recovered more equitably across participants, rather than falling disproportionately on those who remain fully dependent on the grid.
This shift is not theoretical. Countries such as Australia and Germany have already demonstrated models where households with rooftop solar can trade electricity within local networks, turning consumers into active market participants. In Australia, blockchain and AI are used to enable transparent pricing and real-time matching between buyers and sellers of electricity. In the United States and the United Kingdom, pilot projects and flexibility markets are enabling peer-to-peer transactions and local energy exchanges, often supported by AI-driven platforms. Emerging economies such as India and Brazil are also moving in this direction through smart metering and distributed energy integration. Across these systems, households are no longer passive consumers; they become active participants in the energy network.
Pakistan is not facing a simple energy supply shortage. It is navigating a transition from a centralized electricity model to a decentralized and digitally mediated one. In such a transition, the outcome is not predetermined; it depends on how the rules of the system are structured. When solar wins, the grid may appear to lose. However, this is not inherently a zero-sum game. With the right coordination through pricing mechanisms and distributed generation integration, it can become something far more valuable: a system where everyone gains from participation rather than exit.
The Author is an Economist and Independent Researcher. Previously, he served at the Central Power Purchasing Agency (CPPA), where he worked on issues related to Pakistan’s power sector.