Global warming policies were expected to drive a rapid shift toward a renewables-based energy system dominated by wind and solar. While growth in these sources did occur, it has not matched the pace that was widely anticipated. In the United States, the rise of cheap and abundant shale natural gas significantly reshaped the energy mix, displacing coal and limiting the relative share of wind and solar in electricity generation. In China and India, the situation has been different.
Coal remains dominant because it is widely available domestically, while natural gas is more limited or expensive to secure at scale. As a result, coal has retained its central role in both countries’ power systems. Solar and wind always provide intermittent, variable power. It was widely assumed that a cost-effective, utility-scale electricity storage solution would emerge to solve this problem, but that has not yet happened at the scale originally expected. In the pre-AI era, solar and wind were typically integrated into power systems alongside more reliable sources such as coal, natural gas, and nuclear energy.
For example, if the sun was shining on a Monday, electricity demand could be met largely by solar power during the day. At night, coal, natural gas, or nuclear plants would supply the required electricity. If the following Tuesday was cloudy or gloomy, generation would shift back toward coal, gas, or nuclear to maintain supply. AI introduces a new and more demanding challenge. AI data centers require continuous, high-quality, always-on electricity, which solar and wind alone struggle to guarantee without large-scale storage or back-up systems. In addition, they require very large amounts of power.
As a result, the AI industry is now actively searching for new and expanded sources of reliable electricity. One of the major challenges in powering AI systems is electricity transmission. High-voltage transmission lines are expensive, slow to build, and often face regulatory and land-use constraints. As a result, some companies are exploring more localized power solutions, sometimes referred to as microgrids. These are self-contained energy systems that can operate independently from the main electricity grid. Technologies such as small modular nuclear reactors are an example of such microgrids.
In such isolated systems, the foccus is on highly reliable, always-available power generated close to the point of use. In this context, solar and wind are expected to play a limited role because their output is variable and depends on weather conditions, making them less suited as primary sources in fully self-contained AI-focused microgrids. The pace of AI infrastructure development is extremely rapid in both the United States and China. AI systems are widely seen as transformative technologies that promise significant new wealth creation, which is driving aggressive and sustained investment. As a result, development is moving quickly, without waiting for long-term solutions such as large-scale energy storage to mature alongside renewable energy systems.
In this environment, electricity demand is rising faster than new infrastructure can be built. In the United States, this reinforces the role of natural gas as the dominant source of reliable power. In China and India, where coal remains more established and readily available, it is likely to continue playing a central role in meeting growing demand. In India, AI data centers have not yet been built at the scale seen in the United States and China. When India does reach that stage, it will need to supply large amounts of reliable electricity. India has placed strong emphasis on solar energy in particular and has had some success in meeting the needs of ordinary consumers through renewable expansion. However, the key question is what choices will be made when large-scale AI data centers begin to arrive.
Will India rely more on coal generation, which is relatively cheap, widely available, and highly reliable, or on solar power, which is intermittent, variable, and often more expensive when reliability is taken into account? My view is that India is more likely to turn to coal to meet this demand, given its existing infrastructure and the need for dependable electricity supply. Then there is an overall question. Solar and wind were already struggling in the pre-AI days to displace coal and natural gas at the system level, despite strong expectations that they would become dominant sources of electricity. Now that AI is here and electricity demand is rising rapidly, will they push solar and wind further behind in the energy mix?
(The writer is an expert on energy and contributes regularly to publications in India and overseas.)