How to reduce the environmental impact of AI in an eCommerce context

In 2024, digital technologies accounted for approximately 4% of global greenhouse gas emissions, more than the aviation industry (1), and this figure could double by 2025 if no action is taken (2). Every click, every streaming video, every AI-generated text, every email or newsletter sent carries an energy cost. And it’s a cost that often remains invisible. In reality, digital technologies now consume about 10% of the world’s electricity (3). That share is growing fast with the rise of AI tools, data centers, and high-demand networks. We’ve grown used to thinking that digital = intangible. Yet every digital interaction depends on very real physical servers, running non-stop in highly energy-intensive data centers often powered by fossil fuels.

AI: A driver of impact

With the rise of generative AI (ChatGPT, Midjourney, Copilot, etc.), digital energy consumption is skyrocketing.

• Training a single model like GPT-4 consumes millions of kWh (4);
• Simply generating a text or image activates hundreds of processors remotely;
• We multiply the use cases “just to try”, “out of curiosity” or “automatically” without always considering the actual need.

The issue? This ease of access leads to unchecked overconsumption. And contrary to popular belief, AI doesn’t replace human impact, it adds to it.

Why act now?

AI adoption is still relatively new and as with any emerging technology, habits haven’t fully formed. That’s precisely why now is the time to lay the groundwork. Every automatic behavior we develop today, helpful or not, risks becoming standard tomorrow. In other words, it’s better to instill responsible habits now than to have to undo them later.

For eCommerce brands: How to use AI responsibly

In a commercial context, AI can certainly boost performance, but at scale, its impact grows fast. Here’s how to take a more sustainable approach:

• Eliminate unnecessary AI automations (e.g. dynamically generated product recommendations with no added value);
• Evaluate the energy usage of AI tools in your stack (chatbots, search engines, personalization, etc.);
• Choose optimized or lightweight AI models that meet your needs without overwhelming servers;
• Time AI usage around site traffic. For instance, avoid triggering real-time predictive analysis across your full catalog if it’s not business-critical;
• Work with tech partners who are transparent about their infrastructure and energy policy (cloud services, AI providers);
• Prioritize eCommerce platforms committed to reducing the carbon footprint of their algorithms. Some SaaS providers now integrate digital eco-design criteria. It’s wise to choose partners who align with those values (e.g. Shopify);
• Train your teams on thoughtful AI usage. In marketing and customer service, offering training on best practices (e.g. avoiding redundant content variations without a clear objective) ensures more intentional and efficient use;
• Measure the environmental impact of AI features, just like you would for technical performance or conversion metrics.

For users: How to minimize AI’s environmental impact daily

AI is powerful, but energy-hungry. Here’s how to make your usage more intentional:

• Reduce non-essential prompts by grouping queries and avoiding aimless testing;
• Focus on high-value use cases instead of repetitive or low-utility content generation;
• Opt for AI tools that offer transparency around their energy usage or share their sustainability goals;
• Be mindful with multimodal AI (image, audio, video), which often consume more power than text-based tools;
• Adjust usage frequency or disable auto-generation when possible (e.g. turn off auto-generated AI replies in some apps);
• Rethink using AI for simple tasks that could be done locally or manually.

How to be transparent about AI usage and its environmental & ethical impact

Clearly communicate to your customers when features are powered by AI. Recommendation engines, virtual assistants, or interactive try-ons are great, but labeling them encourages more conscious use. Some platforms even disclose the energy cost associated with user actions. This kind of transparency not only raises awareness of environmental impact, but also encourages deeper reflection on the ethics of these technologies: algorithmic bias, data collection, automated decisions. By making these elements visible, merchants demonstrate a real commitment to building trust and offering a responsible digital experience.

Sustainable AI development isn’t just about reducing negative externalities. It’s also about building from a proactive foundation, designing responsible systems from the ground up, optimizing training practices, choosing eco-conscious providers, and including sustainability as a core performance metric. By applying these principles, the AI industry can both reduce its footprint and reinforce its legitimacy in a market increasingly driven by ethical and sustainable practices.

Sources :

1. The Shift Project – Rapport « Pour une sobriété numérique » (2019)
2. The Shift Project, IEA (International Energy Agency), Capgemini Digital Sobriety report
3. International Energy Agency (IEA) – Rapport 2023
4. OpenAI, Lambda Labs, MIT Technology Review
5. GreenIT.fr, ADEME, Carbon Literacy Project

Would you like to improve the sustainable practices of your eCommerce business?

Pierre-Olivier Brassard

Vice President - Products and Technology, Partner

contact@novatize.com+1 844 932-6682

Contact us