- At least 1GW IT capacity mooted; OpenAI is seeking Indian partners and sites.
- Reports link the move to OpenAI’s wider Stargate plan, framed around multi-gigawatt AI campuses.
What happened: Reports flag a 1GW India build as part of ‘Stargate’
OpenAI is weighing a new Indian data centre with at least 1GW of capacity, according to multiple outlets. Data Centre Magazine first highlighted the plan, with local press noting that OpenAI is talking to potential partners as site options are assessed. Economic Times and DataCenterDynamics both attribute the story to Bloomberg’s reporting.
The project is presented as part of OpenAI’s global Stargate infrastructure push. Prior analysis has linked Stargate to about US$500bn of long-term investment across multiple regions, underscoring the scale of planned AI capacity. Bloomberg and FT coverage point to India’s importance as a fast-growing user base and potential hub for sovereign AI workloads.
Also Read: OpenAI clarifies Robinhood tokens are not real equity
Also Read: Meta targets OpenAI researchers with massive $100m bonuses
Why it’s important
A 1GW campus would be a landmark for India’s digital economy, anchoring supply chains from power and cooling to construction and fibre backhaul. It could also accelerate domestic AI adoption if paired with local talent pipelines and credible data-residency frameworks. But the numbers are daunting: grid connections, water use, and land approvals will determine whether the project advances at pace.
Investors and policymakers are likely to judge OpenAI on delivery rather than promises. Multi-gigawatt campuses demand huge amounts of capital and consume vast amounts of power. Without firm energy contracts, timely transmission upgrades and clear permits, deadlines slip and costs escalate quickly. If the company manages to lock in long-term power at predictable prices and show that the build fits India’s wider infrastructure plans, the proposed Stargate site could turn into a bankable project. If not, it may join a long list of mega-schemes delayed by red tape and grid bottlenecks.