"Startups and academics can access H100-class compute at 65 rupees (about $0.72) per hour, the cheapest subsidized rate in the world." [$0.72]
The specific price point of $0.72/hour for H100-class compute, explicitly labeled the world's cheapest subsidized rate, directly evidences the public-utility compute model the signal describes.
Reasoning from this article
India's IndiaAI Mission has already deployed 38,000 GPUs — nearly 4x its initial target — and priced access below any private market rate, treating compute as a public good. This is structurally analogous to how electrification programs or broadband subsidies worked: state absorption of infrastructure cost to accelerate private-sector adoption. If this model proves effective at generating startup density and research output, other middle-income economies will replicate it, creating a new axis of AI competition that bypasses the hyperscaler cloud market entirely.