India's data centre boom needs a green growth plan, says Business Standard column

[JonMoxley19]

A Business Standard opinion column makes the case that the answer to the environmental cost of data centres is to build them better, not to stop building them. The author accepts that these facilities eat up huge amounts of electricity and water. But they are also the backbone of digital services that billions of people rely on every day. The argument is that sustainability has to be designed in from day one

The scale in India is striking. According to the column, cumulative data centre commitments in the country already exceed $90 billion to $100 billion, with another $20 billion to $30 billion expected by 2030. That puts India among the fastest growing markets in the world and a serious digital hub for Asia

Globally, the numbers cited are just as big. Power demand from data centres is projected to grow by more than 20 percent a year through 2030, and one consulting firm expects capacity to reach around 130 gigawatts by 2028. The column also mentions Sam Altman's claim that 38,000 ChatGPT queries use about as much water as growing a single California almond, which is one of the more memorable comparisons I have seen

The practical suggestions draw on a World Economic Forum report. Build sustainability into siting, design and procurement from the start, make facilities resilient to environmental shocks, and use resources more efficiently. None of that is especially radical, but it is easier said than done when everyone is racing for capacity

I agree with the general direction, though I think the almond comparison undersells the local impact. A data centre in a water stressed area can still cause real problems for the people nearby, however efficient it is per query. Should governments be stricter about where these get built? Or is it better to let the market sort it out and focus on efficiency?

Data centres need a green growth planwww.business-standard.com

[CrowSting09]

The almond line always gets quoted but it came from someone running an AI company. I would want to see independent measurement before accepting it

Per query figures also hide the total. Billions of queries a day adds up fast

[SpinState]

Over 20 percent annual growth in power demand is enormous. Grids in most countries were not built for anything like that. Local residents will end up competing with server farms for electricity and water at some point. That is when the politics will get interesting, and this kind of plan will look less optional. Nobody wants power cuts because of chatbots

[Erin91]

Location is everything with data centres. Put one next to cheap renewable power and a cool climate and the footprint drops a lot. Put it in a hot, dry region with a coal heavy grid and no amount of clever design will fix it. Planning rules should reflect that

[Jacob_64]

India has a lot of solar potential, which could make this a strength rather than a weakness. Pair new sites with dedicated renewables and storage and the story changes. Some of the big operators are already signing renewable deals

The grid there still leans heavily on coal though. That transition will take a long time

[Totally]

Stopping data centres was never realistic, so this column is at least being practical. The demand is not going away. Better to push for efficient builds than pretend we can hold it back

[Kev5]

Gas turbines seem to be the short term fix

[ExoplanetRouter]

Fair point on building them better rather than halting progress, but "better" needs teeth. What does that actually mean in practice? Mandatory PUE targets, renewable energy procurement requirements, waste heat recycling into district heating? Without specific standards, "build them better" risks becoming corporate greenwashing speak. Would love to see the column spell out what metrics matter.