Only one-third of marketers claim to fully measure carbon impact of AI usage
Brand marketers are overwhelmingly aware that AI use is increasing carbon emissions and energy costs — but most aren’t measuring them.
That is according to a survey of 200 senior brand marketers working in in-house marketing departments at large US and UK brands, commissioned by climate technology and advisory company 51toCarbonZero and conducted by market research firm Censuswide.
The vast majority of respondents (88%) acknowledged that AI is increasing both their businesses’ carbon footprint and operational costs. More than two-fifths (42%) and more than one-third (35%) said AI is “significantly” increasingly emissions and operational costs, respectively.
Despite this near-universal awareness, only 36% of marketers surveyed said they have fully measured the carbon impact of their AI usage. A further 56% said they had “partially” measured this, with 8% admitting they had not measured it at all.
According to 51toCarbonZero co-founder and CEO Richard Davis, these figures may even represent an overestimation.
“It’s likely that some marketers are relying on rough estimates and optimistically assuming they have fully measured the impact of AI usage, when in reality they’re only seeing part of the picture,” he told The Media Leader. “Marketers clearly want to understand AI’s environmental impact, but they can only measure what suppliers are willing to disclose. That’s why greater transparency from cloud providers and AI platforms is so important.”
An investigation by The Media Leader in March found that media businesses have largely been unable to attain accurate, fully transparent carbon emissions data from AI suppliers such as Amazon Web Services (AWS), Microsoft or Google. One agency holding group told The Media Leader it has been working from “guesstimates”.
The Media Leader investigation: AI push outpaces carbon transparency in advertising
Daniel Schien, associate professor for sustainable information and communication technology at the University of Bristol, described that the providers “have been very slow” and noted that at best they have shared average carbon emissions numbers associated with their cloud services.
This is problematic because, as Schien explained, “there’s sufficient variability between an individual prompt for just one type of query versus other media types, such as video or images.”
Davis described the issue as a “governance challenge”, noting that businesses “cannot effectively reduce what they are not measuring” and that greater visibility into AI’s environmental impact is needed.
Cost controls?
A failure to measure the increased financial costs associated with AI use has also begun biting businesses’ bottom lines.
AI use, it turns out, is proving to be more expensive than the human labour it was meant to replace, even as tech giants like Meta have laid off thousands of workers to fund their AI intiaitives.
Nvidia executive Bryan Catanzaro told Axios in April: “The cost of compute is far beyond the costs of the employees.”
Uber’s chief technology officer acknowledged in May the company’s rush to embrace AI development burned through its annual AI coding budget within just four months. Yet Uber president and chief operating officer Andrew Macdonald noted that Uber’s rapid increase in token usage didn’t correlate with providing more useful features to users.
Sustainability remains important
In a sign of optimism, the survey also found that sustainability credentials are still commercially significant in the marketing industry.
More than nine in 10 respondents (91%) said such credentials were an important factor when considering agency, publisher, platform or technology partners. 41% said they were “extremely important.”
OpenAI wants $100bn of ad revenue. But it can’t ‘guarantee’ brand safety
Commitment to improving sustainability efforts also remains high in the business community, despite the issue generally falling out of favour among current US political leadership. Half of survey respondents said their company is increasing its commitment to sustainability this year; in contrast, just 4% said sustainability commitments have been reduced.
For Davis, brands “have more influence than they sometimes realise” over their relationship with AI giants. He recommended brands collectively “start asking tougher questions of their technology partners” to ensure carbon cost is factored into their AI usage.
“Companies like Google and Meta may seem untouchable, but their business models ultimately rely on advertisers,” he said. “If brands ask for better data, the industry will have to respond.”