As the generative artificial intelligence boom reshapes the global economy, a silent, invisible cost is mounting: the energy required to fuel the massive data centers underpinning models from OpenAI, Google, and Anthropic. For sustainability professionals tasked with reporting Scope 3 emissions—the indirect greenhouse gas emissions that occur in a company’s value chain—the rapid adoption of AI has introduced a new, complex variable.
Until now, measuring the carbon footprint of "black box" proprietary AI models has been akin to guesswork. Sustainable AI Group, a consultancy founded by former Salesforce sustainability manager Boris Gamazaychikov, is looking to change that with the launch of CLEER (Closed-model Latent Energy Estimation Range). This new open-source resource provides a standardized methodology for converting AI token usage into concrete energy and greenhouse gas emissions estimates, offering a critical step toward corporate climate accountability.
The Core Methodology: Turning Tokens into Terajoules
At the heart of the CLEER framework is a sophisticated translation mechanism. While cloud providers often report the total electricity consumption of their data centers, they rarely provide granular data on the energy intensity of specific AI queries.
CLEER bridges this gap by focusing on the "token"—the fundamental unit of data processed by Large Language Models (LLMs). Whether a company is summarizing meeting transcripts, generating code, or automating customer support, the CLEER methodology maps these interactions against an established database of energy consumption patterns.
By leveraging data from more than 20 open-source models as proxies, Sustainable AI Group has created a robust estimation model for closed-source systems. As the industry evolves and more AI developers disclose actual power consumption metrics, the framework is designed to be modular. If a provider releases verified, model-specific energy data, that figure can be swapped into the CLEER calculation without requiring companies to overhaul their existing accounting workflows.
A Chronology of AI Accountability
The development of CLEER did not happen in a vacuum; it is the culmination of years of industry-wide frustration regarding the "black box" nature of AI energy data.
- Pre-2023: As generative AI gained mainstream traction, sustainability officers were largely left to rely on industry averages or opaque data provided by cloud vendors, neither of which accurately captured the surge in energy demand from LLM inference.
- Early 2023: Gamazaychikov and his team began foundational research into AI energy intensity, leading to the creation of the AI Energy Score, a project hosted on Hugging Face that sought to rank models by their environmental impact.
- Late 2023: Recognizing that the AI Energy Score was a proof-of-concept, the team began developing a more scalable, business-ready framework. They partnered with industry leaders, most notably the e-commerce giant Etsy, to pressure-test the methodology.
- 2024: After extensive validation, the Sustainable AI Group officially launched CLEER as a free, open-source resource. Simultaneously, the company began the process of integrating the tool into major sustainability accounting platforms like Watershed.
Supporting Data: The Hidden Costs of Optimization
One of the most startling revelations from the CLEER technical report is the massive variance in energy efficiency between different AI models. When choosing an engine for a specific task, the choice of model can result in a more than 30% difference in energy consumption.
Counterintuitively, the technical analysis suggests that "cheaper" models—often those marketed as cost-effective entry points for developers—can frequently be more energy-intensive than their more sophisticated counterparts. This creates a "sustainability paradox" for procurement teams: cost-cutting measures may inadvertently spike a company’s carbon footprint.
Furthermore, the data underscores the sheer volume of "invisible" energy use. For a sustainability professional, every interaction—from a simple chatbot inquiry to complex data analysis—now constitutes a measurable climate impact. By quantifying these impacts, CLEER allows organizations to make data-driven decisions that balance performance requirements with environmental responsibility.
Official Responses and Industry Collaboration
The launch of CLEER has been met with enthusiasm from the sustainability community, particularly among companies that have already made commitments to carbon neutrality.
The Etsy Perspective
Etsy, a long-standing advocate for corporate transparency in digital emissions, was a primary partner in the development of the tool. Chelsea Mozen, head of impact and sustainability at Etsy, noted that the company’s philosophy is to "test, learn, and share." Having previously open-sourced their own "Cloud Jewels" framework for measuring cloud computing emissions, Etsy’s involvement in CLEER signals a broader shift toward collaborative climate action in the tech sector.
"When new environmental challenges emerge, we don’t wait for the industry to give us new answers," Mozen stated. "We test new approaches, learn what works and share those insights so others can move forward with us."
Integration with Existing Platforms
A critical factor for the success of any new methodology is its ease of use. Sustainable AI Group is currently collaborating with Watershed, a platform used by approximately 800 companies to track their environmental impact. This partnership aims to make AI emission tracking a seamless part of the standard ESG (Environmental, Social, and Governance) reporting process.
"We want this data to go into the places where sustainability professionals are already doing their accounting," said Gamazaychikov. By embedding CLEER into existing software stacks, the barrier to entry is significantly lowered, encouraging companies that might otherwise avoid the complexity of calculating AI emissions to start measuring them.
Implications: The Future of AI and Climate Policy
The emergence of the CLEER framework has profound implications for how the tech industry will be regulated and how companies will disclose their environmental impact.
1. Standardization of Disclosure
As AI continues to integrate into every facet of business, regulators are beginning to pay closer attention to the energy demands of high-compute technologies. By providing a common language and methodology, frameworks like CLEER may become the "gold standard" for voluntary reporting, potentially informing future government requirements for energy efficiency disclosures.
2. The Procurement Shift
Procurement departments are now forced to consider "energy-per-token" as a key performance indicator. As organizations begin to account for the carbon cost of their software vendors, providers who prioritize energy efficiency in their model architecture may gain a competitive advantage. This could trigger a "green race" among AI developers to optimize their models not just for accuracy and speed, but for efficiency.
3. Closing the Data Gap
The ultimate goal of the Sustainable AI Group is to make their own estimations obsolete. By creating a framework that is "ready for truth," they are effectively encouraging AI model providers to step into the light. If OpenAI, Google, or Microsoft begin providing verified energy consumption data, the users of the CLEER framework can simply plug in those numbers, creating a feedback loop that rewards transparency with a more accurate, and potentially lower, emission profile.
4. Broadening Scope 3 Reporting
Scope 3 emissions—which include everything from the production of laptops to the electricity used by third-party cloud servers—are notoriously difficult to calculate. The inclusion of AI-related emissions in these reports has been a glaring gap in modern corporate reporting. With tools like CLEER, organizations can now provide a more complete picture of their climate footprint, reducing the risk of "greenwashing" accusations while enabling more precise climate action.
Conclusion: A Step Toward a Sustainable Future
The rapid rise of AI is one of the most transformative events in the history of the digital age, yet its environmental trajectory remains uncertain. The work being done by Boris Gamazaychikov and the Sustainable AI Group, alongside partners like Etsy, represents a vital intervention.
By turning the abstract, high-energy operations of artificial intelligence into tangible, reportable data, CLEER provides the necessary tools for companies to reconcile their innovation goals with their planetary responsibilities. While the methodology will continue to evolve alongside the technology it tracks, the message is clear: the era of "invisible" AI emissions is coming to an end. Transparency is no longer a luxury; it is the prerequisite for the responsible adoption of the technology that will define the next century.
