Every time a programmatic auction fires — and it fires billions of times a day — a sliver of compute power somewhere in a data centre burns energy to process it. Multiply that sliver across the full stack of modern advertising: the real-time bidding infrastructure, the AI-driven creative optimisation tools, the large language models generating copy variations, the audience prediction engines, the fraud detection layers running in parallel. What you get is not a sliver anymore. What you get is a carbon tab that the advertising industry is only now beginning to read in full.
The numbers, while still contested in their precision, are directionally alarming. Training a large language model of the kind increasingly embedded in advertising platforms consumes roughly the same energy as driving a petrol car across the United States more than three hundred times. Inference — the act of actually running that model at scale, millions of times a day — adds significantly to that figure over the model’s lifetime. When the Global Alliance for Responsible Media released its framework for sustainable media, it was responding to an industry that had quietly built one of the most computationally intensive commercial ecosystems on earth without ever being asked to account for the kilowatts underneath it. AI has accelerated that energy demand at precisely the moment when the climate stakes leave no room for unchecked growth.
The advertising industry built one of the most computationally intensive commercial ecosystems on earth without ever being asked to account for the kilowatts underneath it.
This is not purely a story about technology companies. Brands, agencies, and media owners are all participants in a supply chain that consumes energy at every node. A chief marketing officer approving a campaign that runs across twenty DSPs, refreshes creative in real time using generative AI, and targets audiences through several layers of data enrichment is making, whether consciously or not, an environmental decision. The adtech industry’s carbon footprint sits somewhere between aviation and shipping in some estimates — sectors that have faced mounting regulatory and reputational pressure for years. The difference is that advertising has, until recently, been largely invisible in the sustainability conversation, treated as a soft industry with soft impact. The rise of AI has made that position untenable.
The response from within the industry has been uneven but accelerating. On the infrastructure side, the major cloud providers powering most adtech stacks — Amazon Web Services, Google Cloud, Microsoft Azure — have made high-profile commitments to run on renewable energy, with varying timelines and varying degrees of third-party verification. Google, which operates both the largest search advertising business and one of the most powerful AI research labs in the world, has published detailed sustainability reports and invested heavily in geothermal and wind power contracts. These commitments matter, but they also obscure complexity: energy grids are not uniform, renewable contracts do not always mean real-time clean power delivery, and the pace of AI adoption is straining even the most ambitious green infrastructure plans. Microsoft’s own sustainability report in 2024 acknowledged that its emissions had risen significantly since its AI pivot — a candid admission that growth and green targets are, for now, in tension.
Within the advertising ecosystem specifically, a cluster of initiatives has emerged to build accountability infrastructure. Ad Net Zero, the industry body launched in the United Kingdom and now operating across markets including the United States and Australia, has developed a carbon framework specifically for advertising production and media operations. Its five-point action plan asks advertisers, agencies, and media owners to measure their emissions, reduce them through operational changes, include sustainability criteria in media planning decisions, support lower-carbon creative production, and invest in climate solutions. Signatories include some of the largest holding companies and brand advertisers in the world, though critics note that commitments remain largely voluntary and reporting standards inconsistent.
The measurement problem is particularly acute for AI-driven advertising. Traditional carbon accounting in media focused on relatively tractable categories: electricity consumption for offices and studios, travel for shoots, paper for out-of-home production. AI introduces a layer of computational intensity that is harder to isolate and attribute. When a campaign uses an AI tool to generate hundreds of creative variants, who owns the emissions of that inference workload — the brand, the agency, the platform provider, or the technology vendor? Current frameworks have not fully resolved this attribution question, which means AI-specific emissions risk falling into accounting gaps even as the industry claims progress.
AI introduces a layer of computational intensity that is harder to isolate and attribute. Current frameworks have not fully resolved this attribution question.
Some technology companies are approaching the problem from the model architecture side. Smaller, more efficient models that can accomplish advertising tasks — audience classification, copy generation, bid optimisation — with a fraction of the compute required by frontier models represent a genuine path to lower emissions without sacrificing capability. The shift toward edge computing and on-device inference for certain advertising functions also reduces the data centre load that characterises centralised AI processing. These are technical optimisations, and they are happening, but they are happening in the background, largely invisible to the brands and agencies that are the demand side of the equation.
What will ultimately move the needle is whether sustainability becomes a genuine criterion in procurement and platform selection. There are early signs this is beginning. Some large advertisers have started including carbon reporting requirements in agency pitches. Certain programmatic buyers have begun applying sustainability filters in media planning, routing spend toward publishers with verified lower-carbon supply paths. The Scope3 platform, which measures supply chain emissions for programmatic advertising, has gained traction precisely because it translates the abstract concept of adtech carbon into actionable data at the campaign level. When a media planner can see that one inventory source carries three times the carbon cost of another for equivalent reach, the sustainability conversation stops being philosophical and starts being operational.
India’s advertising market, growing rapidly and increasingly AI-powered, sits at a meaningful inflection point in this conversation. The programmatic ecosystem here is maturing fast, with domestic DSPs and international platforms competing for a market that is simultaneously hungry for AI-driven efficiency and navigating its own energy transition. The brands investing in connected TV, digital-out-of-home, and AI-optimised performance campaigns are making choices today about vendors and platforms whose sustainability credentials vary enormously. As global holding companies sharpen their sustainability reporting under pressure from institutional investors and regulators in Europe and the US, those standards will eventually cascade into how campaigns are planned and bought here.
The carbon cost of AI in advertising is real, measurable, and growing. The industry’s response is real too — patchy, sometimes self-serving, but also producing genuine infrastructure and frameworks that did not exist five years ago. The honest reckoning comes when those frameworks move from voluntary reporting exercises to hard constraints on how campaigns are built and bought. The technology exists to make advertising meaningfully less carbon-intensive. The question is whether the industry’s appetite for AI-driven scale will slow long enough to let accountability catch up.
