AI Compute Infrastructure

5MW to 100MW+, brought online in phases.

Capacity scales in step with contracted demand, with each phase deploying on AI Factory partner sites already under development across Mumbai and Chennai.

5MW
Sept 2026 – Mar 2027

First capacity online.

10MW
2027

Air and liquid cooled fleet.

30MW
2028

Additional partner sites.

100MW+
2028 – 2029

National compute network.

Phase detail

Four phases, from first cluster to national network.

4 × SuperPOD clusters of NVIDIA Blackwell B300 GPUs (air-cooled), 576 GPUs per cluster across 72 nodes, hosted on a colocation model with a partner AI Factory for a five-year period.

Speed to market

Already partnered. Ready to deploy.

Curiosity AI has partnered with leading AI Factories and AI Data Centres in India that are building large-scale facilities, giving the platform the ability to move quickly from demand to deployment.

01

Speed to market

Partner sites are already under development, removing the multi-year lead time of greenfield AI Factory construction.

02

Co-investment

Curiosity AI invests and co-invests in the GPU and supporting compute infrastructure; partners don't finance the buildout alone.

03

Built for scale

Every site is designed to scale from an initial 5MW deployment toward 100MW+ over the following three years.

Execution Proof

Deployment Case Studies & Milestones

Real-world commissioning data, thermal engineering breakthroughs, and cluster handoffs.

Mumbai 5MW Phase 1

Mumbai Phase 1: 5MW Fast-Track Turnkey Commissioning

Liquid Cooling Retrofit

Tier-IV Direct-to-Chip Liquid Cooling System (PUE 1.22)

Neo-Cloud Offtake

Global Demand Translation: 100% Contracted Sovereign Capacity

The GPU infrastructure behind every phase.

NVIDIA Blackwell, AMD Instinct, and the workloads they're built for.