AI has made itself at home in the marketing industry in just a few years. And why not, it is Jack of all trades, yet master of all. From writing copies, generating images to helping analyse audiences and giving teams the ability to produce dozens of campaign variations all much before a traditional creative process might have just produced one. For brands, the attraction is obvious: more speed, more iterations and, potentially, more efficiency.
But technology has another side that is harder to see. Every AI interaction ultimately depends on computing infrastructure, and that infrastructure requires electricity, cooling and hence, significant quantities of water. The International Energy Agency estimates that electricity consumption from data centres worldwide rose sharply in 2025, with AI among the drivers of demand. Google, meanwhile, has published its own methodology for measuring the environmental impact of AI inference, estimating that a median Gemini Apps text prompt consumed 0.24 watt-hours of energy and 0.26 millilitres of water in its May 2025 assessment.
The numbers attached to an individual prompt may appear insignificant. The larger issue is the scale of usage. In advertising, AI is already finding its way into daily work. A marketing team is rarely generating a single prompt; across these functions, the volume can quickly run into hundreds or thousands of requests.
In the case of brands that have spent years making sustainability a part of their corporate and marketing vocabulary, one question becomes unavoidable and interesting to learn about. If the materials, packaging, suppliers and logistics behind a business are scrutinised for their environmental impact, should the technology behind its marketing be treated differently?
Ashok Lalla, independent brand and digital advisor, expects the answer to enter corporate conversations sooner rather than later. The infrastructure behind AI has moved beyond a technology-industry discussion and into public discourse already. “Responsible use of AI in a sustainable manner that does not impact the environment or precious natural resources will soon headline the narratives of even the largest corporations that pride themselves on their sustainable practices,” says Lalla.
Greener AI alternatives, however, are still far from being widely available or deployed. The associated costs could also remain a deterrent to quick and widespread adoption. Lalla expects brands and corporations at the vanguard of sustainable business practices to be among the first to consider such costs an essential part of staying true to their sustainability beliefs and practices.
The question is not arising out of conjecture. Platforms such as GreenPT and Viro have built environmental considerations into their propositions rather than treating them as an afterthought. GreenPT says its workloads run on data centres powered by 100% renewable energy and publishes power and water-efficiency metrics. Viro positions its own AI around clean-energy infrastructure and model routing, sending simpler requests to smaller models while reserving larger models for more complex work. So, if an alternative already exists, at what point does a brand’s sustainability commitment become strong enough to influence the technology it chooses?
The problem with switching
The obvious answer might be: if the environmental cost matters, brands can switch. But businesses rarely replace technology on principle alone. AI is now woven into workflows, APIs, employee training, governance and agency processes.
Nisha Sampath, Managing Partner, Bright Angles Consulting, sees procurement as the point where brands can bring environmental efficiency into the decision. “The change can begin during the procurement of technology, when companies evaluate AI platforms. At this stage, environmental efficiency should become one of the criteria alongside capability, security, cost, reliability and ease of integration,” says Sampath.
But the scrutiny should not end with the choice of platform. Sampath also believes brands need to examine how efficiently they use AI. “Ultimately, the very usage of AI consumes natural resources. Hence a sustainable organisation also needs to look within to ask how efficiently and intelligently AI is being used.”
AI providers also do not have a common way of reporting energy, water and emissions per workload. Google, Viro and GreenPT use different methods, and some of their estimates do not cover areas such as hardware manufacturing, data-centre construction or model training. For a procurement team, that matters. A green label by itself does not create a meaningful comparison. Sampath argues that greater transparency from technology providers is therefore the more immediate demand brands can make. “The digital world has enjoyed an invisibility in sustainability conversations. Consumers only see plastic packaging, factories and delivery trucks, and don’t see the data centres and water consumption happening at the back end,” she elaborates.
The gap between values and workflows 
There is a huge gap between what sounds responsible and what changes a business decision. Amyn Ghadiali, Country Head, Gozoop Creative, explains that gap bluntly. “Let’s be honest: brands aren’t choosing AI platforms in a boardroom debate about groundwater tables. They’re choosing whatever gets the campaign or work out the door by Friday.”
After years of client conversations around AI adoption, he says environmental impact has rarely been the question driving the decision. Performance, integration, cost are much more immediate concerns. For a marketing team working against a campaign deadline, the environmental footprint of a prompt is several steps removed from the problem in front of it.
That does not mean sustainability has no influence. Ghadiali expects it to matter when regulation, client mandates or formal ESG requirements make it part of the business case. Until then, the environmental argument has to compete with something considerably more tangible: whether the technology works, whether the team knows how to use it and whether it fits into the existing stack.
He also points to the difficulty of actually moving away from an established platform. “Should it? Sure. Will it? Rarely. The uncomfortable data point is that switching cost isn’t just tooling; it includes retrained workflows, fine-tuned prompts, integrated APIs and a team’s muscle memory. That’s not a Tuesday decision.”
Going greener may not mean replacing the old one
There may be a middle ground between staying with an existing AI platform and switching to a greener alternative. Yasin Hamidani, Director, Media Care Brand Solutions, believes the more realistic approach “will be multi-platform adoption: brands may keep their core AI stack while routing suitable, lower-complexity tasks to more energy-efficient models or platforms.”
Not every task needs the largest model available. Routine copy, summarisation and classification could be handled by smaller, more energy-efficient models, while more complex work stays with heavier systems.
For brands, however, sustainability will still have to compete with more immediate considerations such as accuracy, security, cost, integration and ease of use. Alternatives such as GreenPT and Viro are positioning renewable-powered infrastructure and visible energy-use reporting as product differentiators. “A greener alternative must therefore offer comparable performance and reliability, not sustainability alone,” Hamidani says.
Should sustainability influence the AI stack?
Brands are already used to making environmental trade-offs. They may pay more for different materials, redesign packaging, change suppliers or alter logistics because the environmental benefit is considered worth the additional cost or complexity. Hamidani argues that sustainability cannot remain outside the AI conversation for long.
Sampath, in turn, believes technology should face the same scrutiny. “Sustainability has always involved trade-offs. If a business is prepared to redesign packaging, change suppliers or absorb higher material costs for environmental reasons, there is no reason its technology stack should not face the same scrutiny,” she explains. Since sustainability alone cannot dictate a platform choice, brands must take the initiative to ask harder questions about the technology they use, from its environmental impact to how that impact is measured.
Ghadiali’s assessment captures the challenge. “Greener alternatives like GreenPT will win the values argument long before they win the workflow argument. Until ‘green’ also means ‘as good, as fast, as embedded,’ it stays a talking point in trend decks, not a line item in tech stacks.”
That may be the real test for responsible AI in marketing. Not simply whether brands can talk about its environmental footprint, but whether sustainability can influence an actual technology decision. For an industry that has learnt to measure almost every stage of the consumer journey, the next test may be whether it is ready to measure the resources consumed before that journey even begins.
.png)
