UK vs India: Two Power Grids Meet the AI Boom, With Opposite Problems
Every AI answer needs a data centre, and every data centre needs a power grid. Compare the United Kingdom and India and you see the two opposite ways a country’s grid can meet the AI boom: one has demand queued up for a grid that grew too slowly, the other has a grid growing at record speed but cities racing to keep up locally.
Size: a giant and a compact veteran
India’s grid is enormous: about 520 gigawatts of installed generation capacity as of early 2026, serving a record peak demand of around 270 gigawatts during the May 2026 heatwave. It is also the world’s largest single synchronised national grid, one connected machine from Kashmir to Kanyakumari. The UK’s peak demand is 45 gigawatts, roughly one sixth of India’s, on a grid largely built decades ago for a demand that has stayed broadly flat for years.
Momentum: the number that says everything
India added a record 65 gigawatts of new generation capacity in financial year 2025-26 alone, most of it solar. Read that again: India built more new capacity in one single year than the UK’s entire peak demand.That momentum is the fundamental difference between the two systems.
The AI guest at each table
In the UK, the guest is bigger than the host. Around 140 proposed data centre projects have requested about 50 gigawatts of power, more than the whole country uses at its busiest moment. The connection queue grew 460 percent in just six months to June 2025, some projects have been told to wait up to 15 years, and land with a ready power connection near London now sells for roughly double ordinary industrial land.
In 2022 the Greater London Authority even warned that parts of West London might not get new housing connections until 2035, partly because data centres had booked the capacity; the warning was later softened, but the message landed.
The government has since declared data centres critical national infrastructure and created AI Growth Zones, starting at Culham in Oxfordshire, with priority grid access of up to 500 megawatts.
In India, the guest is still small. Operational data centre capacity is only about 1.2 to 1.4 gigawatts, a rounding error on a 520 gigawatt system, using under 1 percent of the country’s electricity. But it is projected to reach 9 to 10 gigawatts by 2030 with over 200 billion dollars of investment, consuming about 3 percent of India’s electricity by then, according to energy think tank IEEFA and official reporting.
So India wins? Not quite
India’s risk is local, not national. Its data centres cluster in five city hubs, led by Mumbai, and one average data centre can draw as much power as 100,000 homes, which strains city level cables and substations even when the nation has surplus. India’s state distribution companies also remain financially weak, and heatwave driven demand sets new records every summer, leaving thinner safety margins each year.
The one line verdict
The UK’s problem is a queue: plenty of AI ambition waiting years for a grid that expanded too slowly. India’s problem is a race: a grid growing faster than almost any on earth, trying to stay ahead of both air conditioners and the AI boom now heading its way. For readers in both countries, the lesson is the same: the speed of AI in your country will be set less by software and more by how fast someone can build pylons, substations and cables to your nearest data centre.
Sources
India 520 GW capacity and world’s largest synchronous grid: pib.gov.in
India record peak demand and 65 GW added in FY 2025-26: pib.gov.in
India data centre capacity, growth and city clusters: ieefa.org
UK 140 projects, 50 GW vs 45 GW peak: computerweekly.com
UK queue growth and 15 year waits: gov.uk