Jakarta, INTI - As golden twilight settled over Memphis last May, Sharon Wilson aimed a thermal imaging camera at Elon Musk’s flagship data center to reveal a planetary threat invisible to the naked eye. Without pollution controls, gas-fired turbines powering the world’s largest AI supercomputer continued to pump unseen emissions into the Tennessee sky.  

“It was jaw dropping,” said Wilson, a former oil and gas worker from Texas who has spent more than a decade documenting methane releases and estimates that xAI’s Colossus data center emits more planet-warming gases than a large power plant. “Just an unbelievable amount of pollution.”

That same week, the facility’s flagship product became a hot topic on social media. Grok, Musk’s eccentric chatbot, sparked controversy after repeatedly echoing conspiracy theories about a supposed “white genocide” in South Africa, despite being asked unrelated questions ranging from baseball to scaffolding. Although the posts were quickly removed, Grok soon drew further attention for praising Hitler, spreading far-right extremist ideology, and making false claims.

“It’s a horrible, horrible waste,” said Wilson, director of the campaign group Oilfield Witness, pointing to a Nazi-themed Mickey Mouse image generated by Grok as an illustration of what is produced by burning fossil gas. “What useful purpose does this serve?”

Wilson is not alone in raising such questions. Scientists are increasingly uneasy about the rapid expansion of AI, which they say not only adds to the carbon burden on the natural environment but also pollutes the digital sphere with risks ranging from dubious health myths to deepfake pornography targeting children.

Some experts warn that the growth of data centers could undermine the transition to a clean economy, adding unnecessary strain to the already daunting task of keeping global warming below 1.5°C (2.7°F). Others remain more optimistic, arguing that AI’s energy costs are still far lower than those of heavily polluting industries and are proportionate to the technology’s power to reshape society.

Weighing AI’s Climate Risks Against Its Potential as a Solution

When Hannah Daly was running models for the International Energy Agency (IEA) in Paris and later became a professor of sustainable energy at University College Cork in Ireland, the carbon footprint of computers barely registered on her radar. At the time, the main pressures on the carbon budget came from complex sectors such as transport, livestock farming, and home heating, while emissions from digital services were relatively small and stable.

That situation has changed dramatically in Ireland, where the energy demand from computing has reached a scale that can no longer be ignored. Data centers now consume around one-fifth of the country’s electricity supply and are projected to approach one-third within the next few years. The rapid expansion of chip-filled facilities, outpacing the capacity of the power grid, even led to restrictions on new data-center connections as early as 2021.

It is this “very large exponential growth trajectory” that is most alarming, Daly said. She questioned whether Ireland is a unique case or an early warning of a global trend to come, stressing that the situation should be taken as a serious cautionary tale.

Globally, data centers currently account for about 1% of total electricity consumption, but that share could rise sharply. In the United States, BloombergNEF projects that data centers’ share of electricity use will more than double to 8.6% by 2035, while the IEA estimates they will account for at least 20% of electricity demand growth in wealthy countries by the end of this decade.

Part of this demand is met through long-term power purchase agreements for renewable energy, helping to expand clean power even as the electricity actually running the facilities remains carbon-intensive. Some technology companies have also signed deals to use nuclear power.

In the near term, however, fossil fuels are expected to continue dominating supply. In China, data centers are concentrated in eastern regions rich in coal. In the United States, where natural gas is expected to generate most of the electricity powering data centers over the next decade, the Trump administration has used this projection to justify increased coal burning. “Beautiful, clean coal will be essential to winning the AI race,” Energy Secretary Chris Wright said in September when announcing a $625 million (£467 million) investment package.

In Ireland, which is building liquefied natural gas (LNG) import terminals and gas-fired power plants, the data-center boom has effectively offset the climate benefits of renewable energy expansion in the power sector, according to analysis by Daly for Friends of the Earth Ireland last year. Developing countries may not be immune either. In Pakistan, cheap solar power has begun replacing coal at a remarkable pace, but idle backup generation capacity now risks being absorbed by data centers after the government announced plans to allocate 2 gigawatts of electricity to AI and bitcoin.

“The idea that lower renewable energy costs alone will drive decarbonization is not enough,” Daly said. “Because if there is a large source of energy demand that wants to grow, it will rely on these stranded fossil-fuel assets.” 

The Environmental Impact of Chatbots in Everyday Digital Use

Technology companies have so far resisted pressure to disclose detailed data on AI’s energy footprint. However, widely cited estimates suggest energy use ranges from 0.2 to 3 watt-hours (Wh) for a simple text query, rising significantly for “deep research” tasks and video production. In a blog post last July, OpenAI CEO Sam Altman claimed that a single ChatGPT query uses as little energy as turning on a light bulb for a few minutes, echoing Google’s recent report on the average energy demand of text queries to its AI assistant, Gemini.

These figures are small compared with activities such as flying, eating meat, or driving a car. Still, critics point to the immense scale of the technology, ChatGPT claims hundreds of millions of weekly users just three years after its launch, and the tech industry’s enthusiasm for embedding AI into every aspect of digital life. Google, which controls around 90% of the global search market, has pushed generative AI directly into its search results pages. The rise of AI agents and behind-the-scenes services is expected to add further invisible background energy use.

“What worries me is that we are deploying AI in ways where we don’t have a clear picture of its energy use,” said Sasha Luccioni, head of climate at AI company Hugging Face, who has grown increasingly frustrated with the “selective disclosure” practices of major companies that obscure the true climate impact of their products. “We are essentially operating under the assumption that it’s not a problem, or that if it is, it will somehow be solved, rather than anticipating it.”

“AI can accelerate the deployment of clean technologies by essentially speeding up their progress along the innovation and adoption curve,” said one of the authors, Roberta Pierfederici, a policy researcher at the Grantham Institute at the London School of Economics.

Projected carbon savings, however, remain highly uncertain. The IEA warns that greater efficiency can lead to higher overall consumption, while rebound effects may erode gains,such as autonomous vehicles undermining public transport. Still, concrete examples are emerging. Google says AI has reduced data-center cooling needs by 40%. Spain’s Iberdrola reports a 25% increase in operational efficiency of wind turbines through AI optimization, while France’s Engie has cut downtime at solar farms by using AI to detect faults.

Because other sectors are far more polluting, researchers argue that AI would only need to reduce emissions in those industries modestly to offset the carbon footprint of its own computation, estimated by recent studies at 0.1–0.2% of global emissions and rising. “In the energy sector, we are already seeing results,” Pierfederici said. “The meat sector has not reached that stage yet.”

Conclucion

AI can be both a tool and a threat to the climate. The growth of data centers risks increasing emissions if they rely on fossil fuels, especially without transparency and oversight. However, when implemented appropriately and supported by clean energy, AI can increase efficiency and help reduce emissions in the most polluting sectors.

Read more: Elon Musk Invests Rp300 Trillion in Large-Scale AI Data Center Development