Pakistan’s 2035 Power Plan Must Be Built Around the Consumers Who Will Use It

Pakistan has approved a long-term electricity plan looking to 2035. Its headline is enormous: roughly US$47 billion for the new generation and about US$10.65 billion for transmission, or close to US$58 billion when the two are combined. The scale alone makes one question unavoidable: is Pakistan planning for the electricity consumers it will actually have in 2035?

This is not an argument against planning or investment. Pakistan needs reliable electricity, stronger transmission and a system that protects households and businesses from avoidable costs. It is an argument for a different standard of planning — one that recognises consumers are no longer merely buyers of electricity. Increasingly, they are producers, potential storage owners and flexible users of power.

NEPRA’s approval of the Integrated System Plan (ISP) 2025-35 is therefore more than a routine administrative event. The plan projects peak demand rising from 26,950 MW in 2025 to 35,521 MW in 2035 and anticipates 26,045 MW of new capacity. It also contains 8,120 MW of net-metering capacity in the reference case. That figure alone signals the structural change under way: consumers are reshaping the grid from the edges inward.

The demand curve is already moving

Rooftop solar changes the problem planners must solve. A household, farm or factory with solar may still use substantial electricity, but it buys fewer units from the grid during sunny hours. Add a battery, and the customer can shift when it imports or exports power. This means total electricity consumption and measured grid demand can move in different directions.

Recent analyses of Pakistan’s distributed-solar boom illustrate the scale of that challenge. One 2026 assessment maps roughly 38 GW of distributed solar across residential, industrial, agricultural and commercial users over FY2023-FY2025. Other analyses estimate distributed solar produced about 51 TWh in 2025—around 27-28% of the national electricity supply under their methodology—while daytime grid demand in areas such as LESCO has developed a much deeper midday trough. These estimates are not identical to registered net-metering capacity; that difference is itself a planning problem because a large share of consumer-owned generation can sit outside conventional utility statistics.

A forecast based mainly on historical grid sales can therefore understate underlying electricity use. It can also miss the changing shape of demand: a low midday draw followed by a sharper evening ramp as solar output falls.

“The grid of 2035 will not simply be today’s grid with more power stations. It may be a system in which millions of consumers have solar, some have batteries, EVs move demand between hours, and a few large new loads have outsized local effects.”

If planners overestimate grid-supplied demand, consumers may ultimately pay for unnecessary capacity through tariffs and capacity charges. If planners underestimate local effects, feeders and transformers may face reverse power flows, voltage-management problems or clustered new loads even while national demand appears weak.

Why the battery decision matters

NEPRA’s decision to exclude a proposed roughly US$900 million battery energy storage programme from the approved base case is important for what it says about planning discipline. The regulator’s concern was not that batteries have no value; it was that need, sizing, use and cost-effectiveness had not been adequately demonstrated through technical and economic evaluation and the optimisation model.

That is the correct test. A battery can be poor value if it is inserted into a capacity plan simply as another asset. The same battery can be valuable if it shifts midday solar into the evening peak, reduces renewable curtailment, provides reserves and fast frequency support, relieves a network constraint or avoids a more expensive reinforcement. Storage therefore needs to be co-optimised with generation, transmission, flexible demand and distributed resources rather than assessed in isolation.

The controversy over low-cost renewable projects reinforces the same principle. Reporting on NEPRA’s decision says competitively priced K-Electric renewable projects — including about 640 MW at around 3.09 US cents per kWh — had not been incorporated promptly into national planning despite regulatory concerns. Whatever the final treatment of individual projects, long-term planning is credible only when its inputs are transparent, current and reproducible.

EVs and digital loads make a single forecast even weaker

Pakistan’s New Energy Vehicles Policy 2025-30 targets 30% of new vehicle sales by 2030 and envisages about 2.2 million new-energy vehicles over the policy period, together with a national charging network. EVs will add electricity consumption, but their effect on peak demand depends heavily on when and where they charge.

Unmanaged evening charging can worsen peaks; smart charging can absorb daytime solar. The same technology can therefore be either a grid burden or a flexible resource.

Digital infrastructure is another uncertainty that should be modelled carefully. Pakistan is developing new public digital infrastructure, including the federal Sky 47 data centre, while AI and cloud computing are driving large electricity-demand revisions in several international markets. For Pakistan, however, a major AI-data-centre boom should be treated as a scenario, not as a guaranteed load forecast. Planners should test modest, accelerated and high-growth digital-load cases, with explicit assumptions about location, load factor, reliability and flexibility.

The solution: scenario-based, adaptive planning

The practical solution is not to abandon forecasting. It is to stop pretending that one forecast is enough.

Pakistan’s planning process should publish a small family of transparent demand futures: a central case; a high rooftop-solar and behind-the-meter battery case; a high electrification case covering EVs and industrial loads; a high digital-load case; and a combined disruption case. Each should be stress-tested against weather extremes, hydrology, fuel-price shocks and slower or faster economic growth.

Recent research supports this direction. Pakistan-specific studies show that stochastic adoption of solar PV and EVs can materially affect distribution networks, while long-term scenario modelling produces different investment pathways under business-as-usual, renewable, efficiency and zero-emission assumptions. Research on Pakistan’s 2026 prosumer-regulation shift also shows how compensation rules can change the economics of self-consumption, batteries and demand-side management. In other words, policy itself can change the demand forecast.

For policymakers, one of the most useful metrics should be net load — the demand the central grid must serve after rooftop generation, batteries, demand response and other embedded resources — alongside gross electricity consumption. Pakistan also needs much better feeder-level visibility of registered and unregistered distributed resources. A system cannot be optimised accurately if a growing share of consumer-owned assets is statistically invisible.

The ISP should therefore be treated as a living decision framework rather than a fixed construction list. Annual updates should ask whether demand is moving above or below the assumed trajectory, where rooftop solar is changing feeder profiles, whether batteries are shifting peaks, how EV charging is emerging, whether industrial electrification is accelerating and whether large digital loads are actually materialising.

Pakistan has spent decades worrying about electricity shortages. The next decade may present a more complicated challenge: too much contracted capacity in some hours, too little flexibility in others, falling grid sales alongside rising total electricity use, and consumers making investment decisions faster than institutions can update their models.

The country now has a 2035 power plan. The next test is whether its planning system can become as dynamic as the grid it is trying to build.

What policymakers should publish next

1. Hourly demand profiles for high, medium and low distributed-solar cases.
2. Explicit assumptions for behind-the-meter batteries and EV charging.
3. The requested technical/economic BESS study and its optimisation inputs.
4. Consumer-tariff impacts under each major scenario.
5. A feeder-level plan to measure distributed resources and local network constraints.
6. A transparent process for incorporating newly available low-cost generation into future plan updates.

About the Author

Dr Habib Ur Rahman Habib, SMIEEE, MIET, PEC P.Eng, is a UK-based Energy Innovation Scientist and power-systems researcher. His work spans future electricity grids, hybrid AC/DC systems, converter control, energy storage, EV integration, protection, AI/ML for power systems and hardware-/power-hardware-in-the-loop research. He has academic and research experience across the UK, Europe, China and Pakistan.