When OpenAI abruptly pulled the plug on its highly touted video-generation platform, Sora, last week, the tech world was jolted.

But the reason for its demise was a lesson in hard economics: the app was bleeding cash.

Sora’s daily inference costs — the raw computing power required to run the engine — were estimated at a staggering $15 million, driven overwhelmingly by energy consumption. Its lifetime revenue? A meager $2.1 million.

OpenAI may have shuttered Sora to pivot toward robotics or because it recognised that chasing ever-more-complex models yields diminishing returns.

Yet, the underlying truth remains: even as models become more efficient, the relentless demand for faster, more intense computing will invariably drive energy consumption higher.

This reality check arrives precisely as India is throwing its weight behind the Artificial Intelligence arms race.

India has boldly outlined a $200 billion pipeline for data centre investments.

This is on top of multi-billion-dollar pledges from tech titans like Google, Microsoft, and Amazon, alongside domestic giants Adani and Reliance, who are erecting sprawling new facilities across the western and eastern parts of India.

But beneath the techno-optimism lies a glaring vulnerability.

Fragile Digital Future

The math of data centres is measured not in square footage, but in power consumption. A 10-megawatt (MW) facility means it continuously draws 10 MW of electricity.

As of early 2026, India’s operational data centre capacity has crossed 1.5 gigawatts (GW).

Here is where the strain becomes evident. Over half of that 1.5 GW capacity is clustered in and around Mumbai.

That equates to roughly 750 MW of data centre infrastructure situated in a metropolis whose total power consumption hovers around 4 GW.

While the actual power draw may currently be closer to 500 MW, the projections are ambitious: reports suggest Mumbai could see its data centre capacity explode to 3.2 GW in the coming years.

Powering these massive server farms — whether through the Maharashtra state grid or new dedicated power plants — will require a mix of thermal, gas, and renewable energy, including solar. And perhaps supplemented by nuclear power.

India’s installed power capacity sits at a formidable 524 GW, but the demands of a rapidly growing economy are insatiable.

Incidentally, India has a total non-fossil fuel capacity of around 262 GW, of which around 132 GW is solar.

But as we know, solar power serves daytime needs, and the technology to store that electricity at scale for nighttime release remains elusive. Mumbai’s own generation, including Tata Power’s Trombay unit, is heavily reliant on coal, much of it imported.

Now comes the point.

If the geopolitical chaos of the last month has taught us anything, it is that energy flows cannot be taken for granted.

Gas Goals Meet Reality

India is currently facing a massive gas shortage, affecting both the Liquefied Petroleum Gas (LPG) produced by refineries and the Liquefied Natural Gas (LNG) extracted from oil wells.

With Qatar’s LNG production completely offline, it may be years before global supplies fully recover.

Incidentally, Qatar used to feed around 20% of the world and 80% of Asia’s LNG requirements.

India had deliberately steered its economy toward gas, achieving a 6% share in its energy mix with ambitious targets of 15% by 2030.

That goal now looks highly unlikely.

Energy is fungible; if the economy cannot run on gas, it must pivot back to oil, coal, or renewables.

Over the past month, possibly hundreds of thousands of Indian households and commercial eateries have switched to electric induction heaters or revived the use of kerosene — a fuel the government had spent years successfully phasing out.

Even if Middle Eastern energy supplies were to miraculously come back online in the short term, nations like India must fundamentally prioritise their energy needs differently.

Remember, we import 90% of our crude oil requirements, 60% of our LPG needs of which 90% came through the Strait of Hormuz and 50% of our LNG requirements, of which 40% came from just Qatar and thus the Strait.

Domestic Energy in Flux

As the current crisis took hold, the Indian government made the necessary, albeit painful, choices. It prioritised domestic consumers relying on piped LNG, abruptly cutting supplies to industries — like ceramics manufacturing — that had fully transitioned to gas-only operations.

Refineries were directed to alter their output, maximising the production of butane and propane for domestic LPG while restricting commercial LPG supplies. In a stressed environment, the basic needs of citizens (who also happen to be voters) will always be served first.

All of this has worked, up to an extent.

There will be painful readjustments in the near to medium term, but the economy will cope.

But this is also a good time to start thinking.

Should it abandon its ambitions to build data centres or relook at some of its energy-intensive growth ambitions?

Not necessarily. But it presents a critical moment for introspection regarding the nation’s energy planning for the next decade.

Policymakers must ask hard questions: How do we engineer a system of incentives and disincentives that prioritises our most critical applications for both consumer survival and industrial growth?

These are not simple equations, and they will require rigorous, unsentimental scenario mapping.

Failure to do so risks a dystopian outcome.

India cannot afford to funnel scarce electricity into, for example, hyperscale AI platforms while significant swaths of the country are forced to return to burning firewood just to cook their daily meals.