Renewable power is set to pass coal in 2026. The celebration may last one headline. The grid work will last decades.
The international energy agency expects renewable generation to grow by more than 8% this year after reaching near parity with coal in 2025. Its share of world electricity is projected to rise from 33% in 2025 to 37% in 2027. At the same time, global demand is forecast to grow 3.6% in 2026 and another 3.8% in 2027. Clean generation is winning the race against coal while the track itself is getting longer.
The crossover is historic and easy to misunderstand
Becoming the largest source of electricity generation is a major structural change. Coal has powered industrialisation for generations. Renewables overtaking it shows how quickly solar, wind and other low-emission sources have moved from marginal additions to the centre of power systems.
The crossover does not mean coal disappears. A source can rank second and still generate enormous quantities of electricity. Nor does it mean every country reaches the same point in 2026. Global totals combine systems with very different resources, demand patterns and investment capacity.
It also does not mean electricity is suddenly carbon-free. Gas, coal and other fossil sources remain in the mix, and grid construction has its own material footprint. The milestone should be described accurately. Renewables are becoming the largest source of generation, not the only source and not a finished transition.
Passing coal is a change of leader, not the end of the race.
Demand is accelerating at the same time
World electricity consumption is expected to reach 30,700 terawatt-hours in 2027, up from 28,600 in 2025. Growth is being driven by industry, appliances, cooling, data centres and the electrification of transport and heating.
This matters because new renewable generation must cover two jobs. It has to meet the increase in demand and displace existing fossil generation. If consumption rises faster than low-emission output, fossil plants can keep operating even as renewable capacity breaks records.
A data centre, electric vehicle or heat pump can reduce emissions elsewhere only if the power system can supply it reliably and increasingly cleanly. Electrification moves energy demand onto the grid. It does not remove energy demand.
Efficiency is therefore part of the power transition. Better buildings, motors, cooling and industrial processes reduce the amount of new generation and network equipment required. The cheapest grid upgrade is sometimes the load that never appears.
Solar is carrying the largest new load
Solar output is forecast to increase by around 600 terawatt-hours in 2026, roughly matching the record expansion in 2025. Its modular nature helps. A project can range from a household roof to a vast utility site, and construction is usually faster than for large thermal or nuclear plants.
Solar also has a predictable weakness: it produces according to daylight, not the evening arrival board. High output around noon can depress prices or exceed local demand. A system with a great deal of solar needs flexible consumption, storage, stronger connections or other generation that can adjust.
The useful metric is not installed capacity alone. It is how much electricity is delivered when customers need it, after curtailment and network constraints. A megawatt connected behind a congested line is not equal to a megawatt beside flexible demand.
Wind and solar need a larger network
Conventional plants were often built near fuel supply, cooling water or demand centres. The best wind and solar resources may be far from cities and factories. New transmission must connect them. Distribution networks must also handle rooftop exports, electric vehicles and heat pumps.
Grid projects can take longer to permit and build than generation. That mismatch creates queues. Developers secure land and equipment, then wait for a connection. Consumers see renewable capacity announcements while system operators see substations and transmission lines that do not yet exist.
Planning should move from one-project-at-a-time reactions to corridor forecasts. Where will generation grow? Where will data centres and factories locate? Which lines must be ready first? A cable built after the power plant is late infrastructure.
Storage is valuable and frequently oversold
Batteries can shift solar output into the evening, stabilise frequency and help manage local constraints. Pumped storage and other technologies can cover longer periods. Demand response can perform a similar function by moving consumption rather than electricity.
No storage device solves every time scale. A battery designed for a few hours cannot cover a week of low wind. Seasonal balancing requires different resources, larger networks, firm low-emission generation or retained thermal capacity. Cost comparisons must state duration, cycling and location.
Storage also needs a revenue model. If market rules do not pay for flexibility, investors may install too little. If subsidies reward capacity without performance, systems may install equipment that rarely helps. The grid needs services, not decorative containers.
Coal’s role changes before its volume vanishes
As renewable generation rises, coal plants may run fewer hours or move from steady output to balancing. Many were not designed for frequent ramping. Lower utilisation can raise the cost per unit while plants remain necessary for reliability during certain periods.
This creates a difficult policy sequence. Closing plants too early can threaten supply. Keeping all of them fully protected can block cleaner investment and burden consumers. Authorities need transparent reliability assessments, retirement schedules and support for affected workers and communities.
Capacity payments or strategic reserves may keep some plants available without letting them dominate daily generation. Such mechanisms should be time-limited and reviewed. Reliability cannot become a permanent blank cheque.
Power prices will not move in one direction
Renewables have no fuel bill, but electricity prices include networks, balancing, financing and backup. Hours with abundant solar or wind may have very low wholesale prices. Other hours can remain expensive, especially during fuel shocks or constrained supply.
The mid-year update noted that conflict in the Middle East had temporarily raised generation costs in some markets. This shows why reducing fuel exposure has strategic value. It also shows why the transition must maintain reliability while new infrastructure catches up.
Consumers need retail tariffs that encourage useful behaviour without becoming impossible to understand. Time-based pricing can move charging and industrial loads. Vulnerable households may need protection from volatility. A clever tariff that no one can interpret is not clever for long.
Data centres make location a power decision
Computing facilities can require large, continuous loads. Their growth is one reason electricity demand is accelerating. A site choice based only on land and fibre can create years of delay if the grid connection is weak.
Developers should compare available power, connection timing, generation mix, water use and the ability to reduce load during stress. Utilities need credible project schedules. Announced computing capacity is not the same as a signed, energised load.
There is room for flexibility. Some computing tasks can move in time or between locations. Backup systems may support the grid under carefully designed rules. Yet customers expect reliable service, and not every workload can wait for a windy afternoon.
Emerging systems face a double build
Many countries must expand electricity access and decarbonise at the same time. Their demand growth reflects rising living standards, industrialisation and cooling needs. Asking them to freeze consumption is neither realistic nor fair.
They need generation, grids and finance together. Renewable resources may be excellent, but high borrowing costs can make projects expensive. International capital often prefers established markets even where the physical resource is better elsewhere.
Policy debates should apply consistent standards. Wealthy economies built much of their infrastructure using fossil power and public support. Emerging economies should be judged on credible progress and system constraints, not expected to complete an instant transition with more expensive capital.
The 37% share will test market design
At a 37% renewable share in 2027, weather-dependent generation will influence more hours and more prices. Markets designed around fuel-burning plants may not reward flexibility, reserves and network services correctly.
Shorter trading intervals, better forecasting, wider regional markets and transparent congestion prices can help. So can contracts that finance new generation while exposing operators to performance. The aim is to make reliability and clean output valuable at the times they are useful.
Technical standards matter too. Inverters must support grid stability. Cybersecurity becomes more important as millions of devices communicate. Distributed generation needs safe connection and remote-control rules. A more digital grid gains flexibility and a larger attack surface.
What progress should be measured now
Permits and people can delay the hardware
Grid expansion is often discussed as a shortage of transformers, cables and capital. Projects also wait for route approval, land access and community consent. Transmission lines cross many properties and jurisdictions. A good national plan can still stall at each local boundary.
Faster permitting should mean clearer deadlines, early consultation and coordinated review. It should not mean ignoring communities or environmental effects. Projects delayed by late engagement often take longer than projects that address concerns before a route is fixed.
The workforce is another limit. Engineers, line workers, electricians, planners and control-room operators are needed across generation and networks. Training must expand before retirements and construction peaks collide. Safety cannot be compressed to match an installation target.
Supply chains need steady demand rather than brief procurement panics. Manufacturers are more likely to add transformer or cable capacity when grid plans are credible over several years. Publishing project pipelines, technical standards and connection needs can turn uncertain announcements into investable orders.
Capacity additions remain useful, but five other indicators deserve equal attention: renewable curtailment, connection-queue length, transmission delivery, storage duration and outage performance. Together they show whether the system is absorbing new generation.
Power-sector emissions must also be tracked against demand. Renewable output can rise while emissions stay flat if electricity use rises quickly. The climate result depends on the net change, not the ranking of sources.
Finally, affordability should be measured across households and industry. A system that is clean but unreliable will lose support. A system that is reliable but unaffordable will lose it too. The engineering and social tests arrive on the same flight.
Renewables overtaking coal in 2026 is a genuine global milestone. Demand growth of 3.6% this year and 3.8% next year makes the next stage harder, not less important. Generation has moved first. Networks, storage, markets and efficiency must now catch up.
The sequence matters. Build generation without networks and output is curtailed. Build networks without credible projects and consumers carry idle cost. Add storage without market rules and equipment sits underused. Coordination is slower than a slogan and cheaper than repairing three disconnected plans later.
That is the real power-shift timetable: generation, wires, flexibility and demand arriving in an order the system can actually use.
The power transition has changed the largest source on the board; success will be decided by whether the grid can change just as quickly without switching off the lights.