How AI’s energy needs conflict with media agencies’ carbon reduction efforts

As if AI didn’t present enough existential challenges to the world in general — with fears of hallucinations, misinformation and SkyNet taking over, among other worries — its massive energy needs represent the latest stumbling block. And it’s a big block.

Over the last few months, it’s been widely chronicled that generative AI already requires massive amounts of energy to run, leading the International Energy Agency to recently project that global data center electricity demand will more than double by 2026, including in the U.S. 

This realization directly counters the efforts media agencies have made in recent years to reduce carbon emissions for themselves, their clients and even the media companies whose ad time and space they buy. 

Which means the agency world needs to tackle yet another challenging issue. Is the agency world ready to do that? So far the answer seems to be, not quite.

“I don’t think we recognize that AI is really an existential threat to the collective industry goals of trying to decarbonize media and advertising,” said Anne Coghlan, co-founder and COO of Scope3, an ad-tech firm helping brands, agencies and media companies reduce their carbon emissions. “The unconsidered use of AI is on a collision course with us trying to achieve this collective industry goal of reducing the environmental impact of our industry.”

It’s early days for agencies to recognize this, much less do something about it. More than one holding company passed on commenting for this story.

But Dentsu’s global head of brand assurance Deva Bronson, who did talk with Digiday, said the holdco is aware of the issue, and working on how to react and fold substantive change into its actions taken. 

“It’s starting to be a thought but it’s not necessarily being translated into ESG goals at this stage,” explained Bronson, referring to environmental, social and governance goals. “Any solutions that any of us are currently thinking about or entertaining are meant to be bespoke solutions, so we’re not at a point yet where we can accurately even project what the effect might be of leveraging various solutions — everything’s still quite new. But it is absolutely something that’s in our minds, but I don’t know that we have any hard and fast rules or directives or even recommendation yet.”

At this early point in AI’s evolution and effect on the advertising business, it’s not really on the minds of clients, said Bronson, unless they’re the type of client that puts ESG goals at the forefront of their mission. (Not many do.) And even the industry organizations dedicated to reducing carbon emissions and seeking out better behavior as corporate citizens, such as AdNetZero or the Global Alliance for Responsible Media, aren’t yet incorporating AI best practices into their recommendations or models. 

Court Avenue, a privately held mini-holding company of sorts, which is aggressively building out its own AI strategy (more on that in an upcoming story), believes the cost of AI will get the industry to adjust in a more green direction. David Dettmer, Court Avenue’s global CTO, explained that a client recently had launched a generative AI solution on its website, but was shocked when it got a massive bill that month from OpenAI — a seven-figure tab that was much larger than prior months.

That led Court Avenue to help the client figure out how to mitigate those costs. “That’s an opportunity for us to come in and do what I call token optimization,” said Dettmer. “Because that cost is directly related to how much content you’re sending over to a large language model and how much content you’re receiving back. We spent a lot of time on optimizing that.”

It’s the creative and content side of the agency world that is feeling the most immediate impact of AI — but also the side that can most be adapted to reduct the impact. 

One could say the industry’s collision course with generative AI’s energy needs and the industry’s commitment to reducing carbon footprints is the Titanic.  

“A CMO needs to know how the use of this technology is going to impact the net zero targets that have been set within their organizations, but it’s still relatively nascent — we’re almost at the Titanic leaving the Belfast docks,” noted Scope3’s Coghlan. “If we could provide their captain with a really detailed map of the icebergs, we could avoid that.”

One expectation, said Court Avenue’s Dettmer, is that technology will catch up with AI’s ability to work differently. 

“We are using [creative] technology today that was not built for AI — it was built for generating graphics,” he said. “Over the next few years, we will see new microchips coming out that will be less power intensive and will be optimized for AI. We’ll start seeing the reduction of that power surge.”

So goes the hope, if not actual expectation.

More in Media Buying

Omnicom and The Trade Desk co-develop new data solutions to optimize CTV budgets

The first-of-its-kind partnership with The Trade Desk will co-develop custom investment solutions for Omnicom clients on the major DSP platform. 

Updated: DoubleVerify report, Ad fraud schemes using generative AI will increase in scale, sophistication

Double Verify says generative AI contributed to 23% more new fraud schemes and a 58% increase in ad fraud on streaming platforms.

Amazon unveils Ads Relevance, claiming it is no longer reliant on single IDs

The retail giant’s DSP is using AI to process browsing data to help advertisers nudge shoppers toward a purchase without the need for identifiers.