This is Hamamoto from TIMEWELL.
The Environmental Cost No One Measures
Digital advertising has a carbon problem that almost no one tracks. Every ad served, every impression delivered, every programmatic auction triggers computation that consumes electricity — and most of that energy consumption is invisible to the companies paying for it.
At CES 2025, Scope3 stood out among sustainability-focused technology companies for addressing this problem with a platform that actually measures the cost. This article covers the interview with Scope3 CEO Brian O'Kelley on how the company approaches digital advertising sustainability and what the AI era means for the equation.
- The scale of data center energy consumption
- What Scope3's platform measures and what it delivers
- AI: efficiency or additional burden?
- Summary
Looking for AI training and consulting?
Learn about WARP training programs and consulting services in our materials.
The Scale of the Problem
McKinsey projects that US data centers will consume 580 terawatt-hours of electricity annually by 2028. For reference: this is roughly equivalent to the combined electricity consumption of California, Texas, and New Jersey.
Digital advertising contributes meaningfully to this total. Every programmatic ad delivery involves multiple computation steps: auction processing, targeting evaluation, creative serving, tracking, and measurement. Across billions of impressions per day, across every publisher and advertiser using programmatic infrastructure, the aggregate energy cost is substantial.
Until recently, there was no standardized way for companies to measure how much of that energy was attributable to their advertising activity — or to connect their media spend to its associated carbon output.
What Scope3 Measures and What It Delivers
Scope3's platform addresses the measurement gap directly. It quantifies the carbon emissions generated by a company's advertising spend, attributing emissions to specific placements, formats, publishers, and targeting approaches.
Beyond measurement, the platform evaluates the balance between advertising performance and carbon cost — identifying inventory where the performance return per unit of carbon is high, and inventory where the ad spend is generating carbon with poor performance outcomes.
The reported results from companies that have implemented the platform: 10–15% performance improvement and 30–50% carbon reduction compared to prior campaigns.
The mechanism for both outcomes is similar: reducing waste. Advertising that reaches the wrong audience, appears in low-attention placements, or generates no measurable business outcome is both financially inefficient and environmentally costly. Eliminating that waste improves both metrics simultaneously.
O'Kelley noted that sustainability in digital advertising is still at an early stage industry-wide. Most advertisers do not measure the carbon impact of their campaigns. The reporting frameworks and procurement standards that drive sustainability decisions in supply chain management — where ESG metrics are increasingly required by enterprise procurement — have not yet been consistently applied to media buying.
Scope3's position is that this will change, and that the companies building measurement infrastructure now will have an advantage when sustainability reporting requirements extend to digital advertising.
AI: Efficiency Gain or Energy Burden?
The obvious question in 2025: can AI make digital advertising more efficient, and does that efficiency reduce carbon emissions, or does the compute cost of AI models offset any gains?
O'Kelley's view is measured: "AI can improve advertising effectiveness — but using excessive energy to do so is counterproductive. AI adoption needs to be balanced between performance outcomes and efficiency."
The concern is real. Running large AI models for targeting optimization, creative generation, or audience analysis requires substantial compute. If AI-optimized advertising consumes significantly more energy per impression to achieve a modest performance lift, the net environmental impact may be negative despite the efficiency framing.
O'Kelley also pointed to a structural issue: users and companies currently bear no cost for the energy consumed by AI inference. If that cost were visible — if companies saw the electricity bill associated with AI-powered ad targeting, the way they see media spend and platform fees — purchasing decisions might change.
His view on policy: government engagement in energy and demand-side policy for AI will eventually be necessary. The technology sector's energy consumption is growing faster than the renewable energy supply coming online. Without either demand management or supply-side expansion that keeps pace, the sustainability math does not close.
Summary
| Element | Detail |
|---|---|
| US data center energy projection | 580 TWh by 2028 (equivalent to CA + TX + NJ) |
| Scope3 platform function | Measures carbon emissions per advertising spend; identifies waste |
| Reported client outcomes | 10–15% performance improvement; 30–50% carbon reduction |
| AI sustainability tradeoff | Efficiency gains possible, but compute energy cost must be weighed |
| O'Kelley's policy view | Government energy and demand-side policy involvement needed |
Scope3's approach is notable because it converts an abstract sustainability claim into a measurable business metric. The 30–50% carbon reduction figures are the kind of outcome that can appear in sustainability reports — but they are achieved by the same mechanism that improves performance: reducing waste. When advertising dollars spend more efficiently, both the financial and carbon outcomes improve.
The harder question O'Kelley raises — about AI's energy cost relative to its performance benefit — is one the industry has not yet resolved at the measurement level. As AI becomes more central to advertising infrastructure, the companies that can quantify that tradeoff will have a basis for making decisions that others will be making in the dark.
Reference: https://www.youtube.com/watch?v=YixAhLhtzHY https://scope3.com/