No, AI is not destroying the planet

by Antoine Buéno | 8 Nov 2026 | Our experts in the media, Climate, Artificial intelligence

Reading time is .

Key points

  • The figures: around 0.2% of global electricity, less than 0.2% of water, 0.1 to 0.2% of emissions.
  • Local damage: real, but due to carbon-intensive electricity, unregulated mining and poor siting choices.
  • The remedies: decarbonise, regulate, recycle, as we did for the railways.
  • The gains: 4 to 10% fewer emissions by 2030 according to BCG, and faster discoveries.

🤖 AI, an ecological black hole? The orders of magnitude say otherwise.

In Le Point, futurist Antoine Buéno goes through the figures one by one, electricity, water and metals included.

📊 What AI accounts for today
• around 0.2% of global electricity (between 1 and 1.5% in 2030);
• less than 0.2% of the water used by humanity;
• 0.1 to 0.2% of greenhouse gas emissions.

⚠️ Local damage does exist (coal-fired electricity, unregulated mining, data centres in arid areas), but it stems from the framework more than from AI: decarbonised electricity and standards address it. Railways, too, polluted in their early days: we managed to correct those nuisances.

🌱 Its potential gains: mapping methane leaks, more precisely dosed fertiliser and irrigation, fourth-generation nuclear power, fusion, new materials. According to BCG, 4 to 10% fewer emissions by 2030.

💬 “The moral of the story: AI is far more fundamentally an environmental solution than an ecological problem.”

Should we still feel guilty with every query?

👉 The summary of the article and the link to Le Point are below.

#ArtificialIntelligence #climate #energy

🤖 Is artificial intelligence destroying the planet?

Giant data centres, metal mines, water pumped in arid areas: AI is seen as an ecological black hole. In Le Point, futurist Antoine Buéno goes through the figures one by one and reaches the opposite conclusion.

📊 A footprint that remains marginal
AI consumes around 0.2% of the world’s electricity, less than 0.2% of the water used by humanity and 0.0006% of the metals produced. Agriculture, by contrast, takes 70% of fresh water. AI emits 0.1 to 0.2% of greenhouse gases. Even if its consumption triples, it will account for only between 1 and 1.5% of the world’s electricity in 2030.

⚠️ A symptom rather than a culprit
The local damage is real: coal-fired electricity, unregulated mines, data centres in dry areas. But the problem lies in the framework: nearly 60% of the world’s electricity is still carbon-intensive. Decarbonisation, standards, cooling and recycling are enough to address it, as was done for trains.

🌱 The eyes and ears of the transition
AI has already made it possible to map methane leaks worldwide. It can fine-tune fertiliser and irrigation, and speed up fourth-generation nuclear power, fusion and the development of new materials. According to BCG, it could cut emissions by 4 to 10% by 2030.

💬 The author’s conclusion: “The moral of the story: AI is far more fundamentally an environmental solution than an ecological problem.”

An analysis that does not deny the local damage, but puts it in proportion.

Should we still feel guilty with every query?

👉 The three arguments in detail, with the link to the Le Point article, are below.

#ArtificialIntelligence #ecology #climate #energy #transition #digital

🤖 Is AI really destroying the planet?

In Le Point, futurist Antoine Buéno goes through the figures one by one.

1️⃣ A marginal footprint: around 0.2% of the world’s electricity and 0.1 to 0.2% of greenhouse gases.
2️⃣ A symptom rather than a culprit: the local damage comes from carbon-intensive electricity and unregulated mines, which we know how to fix.
3️⃣ Potential gains: 4 to 10% fewer emissions by 2030, according to BCG.

And you? Do you see AI as more of a problem or a solution for the climate?

📎 Full summary and link to the article: link in bio.

#ArtificialIntelligence #ecology #climate #energy #transition

by Antoine Buéno, Le Point, 8 November 2026

Electricity, metals, water, emissions: artificial intelligence absorbs only a tiny fraction of the world’s resources. Its local damage is real, but it reveals problems that go beyond AI, and its potential gains for the transition far outweigh that damage, the article argues.

In brief

  • AI consumes around 0.2% of the world’s electricity, less than 0.2% of the water used by humanity and 0.0006% of the metals produced; it emits 0.1 to 0.2% of greenhouse gases.
  • Locally, it can be a scourge (coal-fired electricity, unregulated mines, data centres in arid areas), but the cause lies in the framework, not in AI: we know how to fix these harms, as was done for trains.
  • Its gains could be considerable: 4 to 10% fewer emissions by 2030 according to BCG, and far more with the discoveries it accelerates and the personal agents now on their way.

The article’s three arguments

1. Orders of magnitude

A footprint that remains marginal

Even if its consumption triples, AI will absorb only between 1 and 1.5% of the world’s electricity in 2030, while heavy industry and transport will take two-thirds of it. Its emissions would reach around 1% of the total, far behind industry, agriculture or transport. Agriculture takes 70% of fresh water.

2. The real problems

A symptom rather than a culprit

The problem is not that AI consumes electricity, but that nearly 60% of the world’s electricity is still carbon-intensive; not that it uses metals or water, but that some mines escape all regulation and some data centres are set up in dry areas. Decarbonisation, standards, cooling and recycling are enough to address this.

3. The gains

The eyes and ears of the transition

AI has already made it possible to map methane leaks worldwide. It can fine-tune fertiliser and irrigation, and speed up fourth-generation nuclear power, fusion, new crop varieties and new materials. Tomorrow, personal agents could make our everyday choices greener and at last make ecological planning effective.

Worldwide, fossil fuels produced 65% of electricity in 2000 and still 57% in 2025. Renewables rose from 19% to 34%, and nuclear fell from 17% to 9%. The problem is not the electricity AI consumes, but the way it is produced.
In 2023, humanity emitted around 50 billion tonnes of CO₂ equivalent: 17 for electricity and heat, 8.5 for transport, 6.2 for agriculture and 6.2 for manufacturing and construction. AI, around 1% of the total according to the article, is included in the first category.
Worldwide, agriculture accounts for 71% of freshwater withdrawals (2022). The share varies widely: 90% in India, 32% in the European Union, 10% in France.
“The moral of the story: AI is far more fundamentally an environmental solution than an ecological problem.”Antoine Buéno, Le Point, 8 November 2026

Why this debate matters to us

APALA sees artificial intelligence as a tool for intellectual emancipation and an accelerator of the transitions, and prefers to judge a technology by orders of magnitude rather than by impressions. This article does exactly that: it compares AI’s footprint with that of other human activities, then weighs it against what AI can deliver. Its conclusion ties in with another of our areas of work: at the individual level, eating less meat matters far more than giving up AI.

Questions for debate

  1. Should a technology be judged on its footprint alone, or on its net balance, costs and gains included?
  2. According to Bpifrance, the announced gains range from 1.5 to 24% of emissions: how can they be verified, and how can we prevent a more efficient AI from mainly driving up usage?
  3. AI agents that gently nudge us towards greener choices: progress or an infringement of our autonomy?

To go further on apala.fr

Summary written by Claude (Opus 5.5 model), an artificial intelligence, for APALA, based on the article published in Le Point on 8 November 2026. The views expressed are those of the author. Spotted an error? Write to us.

Read the article in Le Point ↗︎

AI: is the apocalypse imminent?

In Le Point, Thomas Lepeltier looks back at the affair of the OpenAI agents that hacked Hugging Face: human negligence, not signs of an apocalypse. Our summary in three arguments.

Is the zombie chicken the future of the animal rights movement?

In L’Amorce, Thomas Lepeltier proposes reducing farmed animals’ capacity to suffer rather than waiting for the end of meat, going as far as imagining a “zombie chicken”. Our summary in three arguments.

What if the danger of AI were not that it exterminates us, but that it advises us too well?

In Le Point, Thomas Lepeltier presents Antoine Buéno’s ‘Guidance’: the danger of AI would be less that it exterminates us than that it advises us too well. Our summary in three arguments.

Is the world burning?

In Causeur, Thomas Lepeltier compares the exceptional wildfires of summer 2026 in France with global satellite data, which describe a rather quiet year. Our summary in three arguments.

Let’s have done with organic farming!

In Atlantico, Thomas Lepeltier challenges the founding principle of organic farming, the preference for the natural over the synthetic, and judges its record to be more nuanced than its image. Our summary in three arguments.

A future without suffering? A video to open the debate

With The Flares, APALA offers a short foresight video: what if reducing avoidable suffering, among humans and other animals alike, were one of the great undertakings of the century?

Tatiana Giraud: when biodiversity turns into a religion

In Causeur, Thomas Lepeltier criticised the way Tatiana Giraud presents biodiversity; she responded, and he replied. Our summary of the three texts: arguments, quotations and questions for debate.

The fanciful claims of biologist Olivier Hamant against performance

In Le Point, Thomas Lepeltier challenges the opposition between performance and robustness championed by the biologist Olivier Hamant: in his view, the resilience of societies comes from innovation and efficiency. Our summary in three arguments.

Organic: beware, toxic!

In Causeur, Thomas Lepeltier draws a lesson from Anses’s decision to withdraw copper fungicides: judge each product by its effects, not by its natural or synthetic origin. Our summary in three arguments.

Will artificial intelligence end up exterminating us?

In Le Point, Thomas Lepeltier examines the thesis of Eliezer Yudkowsky and Nate Soares that a superintelligent AI would exterminate humanity. Our summary in three arguments.

0 comments

make a donation!

Public-interest association: €10 for you means nearly €30 for us, after tax relief.

Understand how it works