In recent months, the Internet has been abuzz with fiery debates: Is AI, particularly ChatGPT, wrecking the environment?
Across forums, newsletters, speeches, and social media, people are sounding the alarm. TikTok influencers are urging followers to ditch AI tools altogether. One user, @nikitadumptruck, even claimed that sending 100 words via email uses 500 millilitres of water[1]. That’s half a bottle just to email your boss!
But not everyone’s buying the panic.
Another TikTok creator, @ktrivz, countered the narrative, warning that those who refuse to adopt AI may get left behind in the next decade. She pointed out that it’s unfair to single out AI, arguing that all internet usage contributes to environmental degradation, not just ChatGPT[2].
So, what’s really going on? Is ChatGPT the villain in our climate story… or are we just giving it main character syndrome?
Tech vs The Planet: An Unexpected Relationship
Before you rage-quit ChatGPT, let’s take a step back and unpack how tech in general affects the environment.
Technology has undeniably made life more convenient. Remember when SMS had a character limit and cost money? Now, you can WhatsApp a meme, a voice note, and a video to your entire family group – instantly and for free.
AI is no exception. It’s helping conservationists analyse satellite images to track deforestation or identify changes in water levels. Drones are being used to catch illegal loggers and poachers[3].
In fact, tech can actually reduce emissions. Switching to renewable-powered servers, for instance, helps shrink the carbon footprint of digital infrastructure. The key? Using tech responsibly and ethically to build a more sustainable future[3].
Still, it’s not all green and glowing.
So, How Bad Is It Really?
A 2024 study analysing data from 2020 revealed that the Information and Communication Technology (ICT) sector, think smartphones, data centres, internet networks, was responsible for about 1.4% of global greenhouse gas (GHG) emissions[4]. That’s not peanuts.
The sector also gulped down around 4% of the world’s electricity just to keep our screens glowing and our clouds running.
Surprisingly, user devices (our beloved phones and laptops) were the biggest contributors. Data centres and networks? Most of their emissions came from the energy it takes to run and cool them.
Why Is AI an Energy Hog
According to Canadian computer scientist Dr. Sasha Luccioni, generative AI uses up to 30 times more energy than your average Google search.
Why? Instead of just fetching answers, it creates them from scratch, powered by massive language models trained on billions of data points.
All of this requires heavy computing power and enormous amounts of electricity.
I find it particularly disappointing that generative AI is used to search the Internet. — Computer Scientist (AI & Climate) Dr. Sasha Luccioni[5]
To put things in perspective, the combined energy use of AI and cryptocurrency industries in 2022 was around 460 terawatt hours, roughly 2% of global electricity production[5].
Your Chatbot Uses More Water Than You Think
ChatGPT and AI systems like it don’t just run on data and code. They also require massive amounts of processing power, especially during training and usage. All this heavy lifting happens on servers inside data centres, which guzzle electricity not only to operate but also to stay cool.
To avoid overheating, which can fry hardware or throttle performance, many data centres, including those in Malaysia, use water-based cooling systems. Why water? Because it’s highly effective at absorbing and dissipating heat. That’s where the thirst factor comes in.
Before ChatGPT could fire off clever responses to your late-night existential crises, help you write that awkward email to your boss, or explain the plot of Dune, it underwent a long, energy-draining training process. That process involved billions of calculations across high-powered machines, generating intense heat that had to be cooled, often with water.
But let’s get one thing straight: calling this water “wasted” isn’t entirely fair. Many data centres recycle or treat the water they use. That said, there’s still some unavoidable loss, mainly through evaporation in cooling towers. And while training a model like ChatGPT is the most energy- and water-intensive phase, even everyday use draws electricity – and yes, a bit of water too[6].
Which begs the question: With Malaysia positioning itself as Southeast Asia’s next big AI data centre hub, are we quenching our thirst for tech at the planet’s expense?
Malaysia’s Data Centre Gold Rush — At What Cost?
Tech giants like Google, Nvidia, and Microsoft have been pouring billions into Malaysia’s data centre boom[7]. Over just the past three years, Johor has become a hotspot, attracting more than 50 projects, including major players like Microsoft and ByteDance.
Research firm DC Byte reports that this explosion has generated over 40,000 jobs and triggered a 100-fold increase in Johor’s total data capacity, including both active and upcoming sites[8].
It looks like in the space of a couple of years, (Johor Bahru) alone will overtake Singapore to become the largest market in Southeast Asia from a base of essentially zero just two years ago. — APAC managing director James Murphy[7]
So, why Malaysia?
For one, AI data centres are notoriously resource-hungry – they need space, electricity, and water to keep their servers cool. Malaysia offers all three, and at a much lower cost than compact, resource-constrained cities like Singapore or Hong Kong. Add in cheap electricity and competitive infrastructure, and Malaysia quickly becomes a prime destination[7,9].
According to DC Byte’s 2024 Global Data Centre Index, Malaysia’s data centres are on track to hit a combined energy capacity of 16 gigawatts – that’s enough to power up to a million homes[9].
Government support has added fuel to the fire. TNB’s Green Lane Pathway, launched in 2023, helps fast-track approvals and encourages sustainable practices among data centre operators[10]. Other perks include tax incentives, subsidies, and ample industrial land and water, especially crucial for water-cooled facilities trying to cut emissions[9].
But here’s the catch.
As exciting as this rapid growth is, there’s rising concern over water and energy shortages, particularly in Johor, the epicentre of this boom.
Even before factoring in the surge in data centres, Malaysia is already staring down potential water crises in the next five years, thanks to ageing infrastructure and climate change. The National Water Services Commission warns that the situation could get worse.
On the energy front, Tenaga Nasional Berhad projects that by 2035, electricity demand from data centres alone could exceed 5,000 MW – that’s more than 20% of Peninsular Malaysia’s current power capacity[8].
Even local leaders are hitting pause. Johor Bahru’s mayor, Mohd Noorazam Osman, has voiced serious concerns:
People are too hyped up about data centres nowadays, but the issue in Johor is that we do not have enough water and power. I believe that while promoting investments is important, it should not come at the expense of the local and domestic needs of the people. – Johor Bahru Mayor Mohd Noorazam Osman[8]
Reframing The Real Cost of Convenience
Yes, AI uses water. But how much, really? Consider this: boiling water for a cup of tea takes about a litre. Streaming a 10-minute video? Surprisingly, it consumes around the same amount, thanks to the water used to cool the data centres powering your content[6].
The Johor state government isn’t turning a blind eye. New sustainability guidelines for data centres are slated for release in June. Even Prime Minister Anwar Ibrahim has weighed in, stressing that Malaysia can’t rely on cheap energy and water forever.
Additionally, to ensure local resources are not affected, the Cabinet approved new data centre design guidelines Data Centre Planning Guidelines (GPP), back in October 2024. These guidelines, developed by the Housing and Local Government Ministry through PLANMalaysia, aim to make the planning and approval process easier and more consistent. They also support business growth and help guide those involved in building data centres.
With the launch of the Data Center GPP, we intend to strengthen the development of digital infrastructure and support the data centre development ecosystem through the construction of data centres in more strategic locations according to the designated land use zones. This is to ensure the distribution of resources for domestic needs is not affected. — YB Nga Kor Ming, Minister of Housing and Local Government[11]
Future developments must strike a balance, ensuring investments don’t come at the cost of local communities or natural resources[8].
In the end, the real question isn’t just how much water AI consumes. It’s how we, as users and decision-makers, choose to engage with this technology. Are we demanding smarter, more sustainable innovation?
AI isn’t necessarily the villain. But if we’re not careful, our thirst for convenience could come at a cost we didn’t bargain for.
Explore our sources:
- @nikitadumptruck. Why AI is destroying our environment!! Chat gpt is trained off YOUR BRAINS let’s go back to raw doggin thoughts. TikTok. Link
- @ktrivz. big oil watching us attack OpenAI and laughing 🤫 the consumers are ALWAYS to blame — please stay curious and open to new tech and pay attention to AI. TikTok. Link
- Y. Chalup. (2024). Technology and the environment: a battle between harm and benefit. Telefónica. Link
- J.Malmodin, N. Lövehagen, P. Bergmark & D. Lundén. (2024). ICT sector electricity consumption and greenhouse gas emissions–2020 outcome. Telecommunications Policy, 48(3), 102701. Link
- AFP (2024). AI is ‘accelerating the climate crisis,’ expert warns. Free Malaysia Today. Link
- T. Empire. (2024). Every time you use ChatGPT, half a litre of water goes to waste. Really??. Medium. Link
- D. Butts. (2024). Malaysia is emerging as a data center powerhouse amid booming demand from AI. CNBC. Link
- U.Daniele & K. Alam. (2024). Malaysia’s new data centers create thousands of jobs — and worries about power and water shortage. Rest of World. Link
- N. Muslim. (2024). Malaysia’s Data Centre Boom Could Leave People Dry – Experts. Bernama. Link
- Tenaga Nasional Berhad. (2023). TNB Establishes Exclusive Green Lane Pathway & Strategic Offerings for Malaysia’s Data Centre Market. Tenaga Nasional. Link
- Bernama. (2024). Cabinet approves guidelines for data centre planning – Nga. New Straits Times. Link
