In the recent Union budget, India recently announced a major tech incentive - no taxes until 2047 for global cloud providers that use Indian data centres. Google is investing $15 billion, Microsoft $17.5 billion, and Amazon has pledged $75 billion by 2030. On paper, India looks set to become the next AI superpower.
But there’s a problem that isn’t getting enough attention. It’s not mainly about money, policy, or even talent. The real issue is something more basic - electricity.
The Elephant in the Server Room
The hard truth that the Finance Minister’s budget speech didn’t address is that India’s data centre capacity must grow from ~1.5 gigawatts today to over 9 gigawatts by 2030. This will require about 40-50 terawatt-hours of extra electricity each year, which is enough to power a mid-sized European country. Right now, during peak summer months, India already faces power shortages of 10 gigawatts, and states often cut power to industry to keep homes and farms running.
India is asking the world’s biggest tech companies to invest in a country where a reliable 24/7 power supply is still out of reach in many areas. Reliance Industries is already building its own power plants for AI infrastructure because the grid isn’t dependable. This adds $0.03-0.08 per kilowatt-hour to their operating costs just to keep servers running.
The Economic Survey 2025-26, released before the budget, recognised that India needs to build its own AI computing infrastructure. But this won’t matter if there isn’t enough power. A 500-megawatt data centre needs a dedicated substation, and will cost hundreds of millions of dollars and take a few years to build. Microsoft’s campus in Thailand waited a few years for substation approval and ended up with a percentage of the needed capacity. Oracle’s facility in Malaysia also faced similar issues.
Why This Matters More Than Talent Gaps
Yes, India does have a talent shortage. The Economic Survey pointed out that India needs more people with real experience in algorithms and software engineering. The skills gap is clear. By 2027, India will need 1.25 million AI professionals, but there are only about 600000 today. That’s a huge gap.
But here’s the point: people can be trained. Microsoft, for instance, is already working with the government on large upskilling programs. Universities can update their courses. Online platforms can quickly expand training. Skills can be improved in three to five years.
But building power infrastructure takes at least ten years. Gas-fired plants need six years from planning to completion, and onshore wind projects take about the same. India’s move toward renewables, which is crucial for its 2070 net-zero goal, also means it needs large-scale battery storage, a technology still being tested worldwide. Up to 40% of a data centre’s energy is used just for cooling. AI workloads produce a lot of heat, and in India’s climate, the need for cooling is even greater.
The Budget’s Missing Link
The budget set aside ₹400 billion for the Electronics Components Manufacturing Scheme and announced a second phase of the India Semiconductor Mission. Reducing rare-earth minerals reliance on China has also been a strategic focus. These are smart steps that show awareness of supply chain risks.
But there was no mention of dedicated power generation for data centres. There was no special funding for renewable energy linked to computing infrastructure, no fast-track approvals for data centre substations, and no public-private partnerships for clean energy generation.
The contrast is clear: the budget raised electronics manufacturing spending by 74% after the program beat its targets, but there is no similar “power-first” plan for AI infrastructure. It’s like building a mansion but forgetting to connect it to the power grid.
The Path Forward
India should make reliable power the foundation of its AI strategy. This involves three steps - setting up renewable energy zones next to data centre parks, speeding up regulatory approvals for power infrastructure linked to key tech investments, and requiring sustainability standards that match our climate goals while ensuring nearly 100% uptime.
Tax breaks may attract attention, but reliable power is what truly matters. Without it, billions in foreign investment will go to places like Vietnam, Indonesia, or the Middle East, where energy infrastructure is being built together with data centres, not as an afterthought.
India really can become an AI hub, but only if it makes sure the servers stay on.