India's $100 Billion AI Bet: Zero Tax Holiday and the Infrastructure Race That Could Reshape Global Compute

India has rolled out one of the world’s boldest incentives to attract AI and cloud infrastructure. Under the Union Budget announced on February 1, foreign cloud companies can operate data centers in India and pay zero tax on global revenues until 2047. The policy is meant to pull large-scale AI computing into the country at a time when demand for data centers is surging worldwide.

What the policy actually says

Finance Minister Nirmala Sitharaman announced a 21-year tax holiday for export-oriented cloud and AI services.

Key points

• Zero tax on profits from global customers served from Indian data centers
• Indian customer revenue is still taxed through local reseller entities
• Applies mainly to hyperscale data centers and AI workloads, not consumer apps

In simple terms, companies can use India as a global AI “factory” without paying Indian taxes on overseas earnings.

Why big tech is lining up

The announcement builds on a wave of investment commitments made since late 2025 by global tech giants.

Recent commitments

• Google: $15 billion for AI-focused data centers
• Microsoft: $17.5 billion over four years
• Amazon: $35 billion by 2030

Together with Indian partners, announced investments now total nearly $100 billion.

Companies are attracted by a mix of low costs, long-term tax certainty, and India’s large engineering workforce.

Why India makes sense for AI infrastructure

India offers advantages that are hard to match elsewhere.

Structural strengths

• Lower costs: Data centers cost 30–40% less to build and run than in the US or Europe
• Talent scale: Over a million engineers graduate each year
• Geographic diversification: Firms want alternatives to concentrating compute in the US or China

With the tax holiday, India becomes one of the cheapest large-scale locations to run AI systems.

What India gets in return

Even without taxing export profits, the government expects broad economic benefits.

Expected gains

• Jobs in construction, operations, and engineering
• Skill development and technology transfer
• Growth of local suppliers (power, networking, hardware)
• Stronger position in the global AI infrastructure race

The strategy is to make India indispensable to how AI actually runs—not just where software is written.

The risks the government admits

India’s own Economic Survey has warned that infrastructure limits could slow the plan.

Main constraints

• Power: Data center electricity demand is rising much faster than grid expansion
• Water: AI data centers need large volumes for cooling, and many are planned in water-stressed cities

By 2030, data centers could consume hundreds of billions of liters of water annually, raising concerns about competition with households and agriculture.

What happens next

India currently has about 1.5 GW of data center capacity. Projections suggest this could grow three to five times by 2030 if power and water infrastructure keep pace.

Execution now matters more than incentives. If India can expand utilities fast enough, it could become a major global AI compute hub. If not, the tax holiday risks drawing investment that struggles to operate at full scale.

Bottom line

India’s zero-tax offer is a high-stakes bet to host the physical backbone of AI. It has already triggered massive investment interest. Whether it reshapes global compute—or runs into hard infrastructure limits—will depend on how quickly power and water challenges are solved.