SCI for AI Bootcamp: How Four Organizations Are Measuring Their AI Emissions
Amadeus, AVEVA, Schneider Electric, and Siemens joined our first cohort to apply the SCI for AI methodology to their AI systems and work through common implementation challenges.
Developed through consensus among more than 20 member organizations, Software Carbon Intensity (SCI) for AI builds on the ISO SCI methodology to provide practitioners and organizations with a standardized way to measure the carbon emissions of AI systems throughout their lifecycle.
SCI for AI was ratified in late 2025, and through conversations with our members, it quickly became clear that organizations needed structured support to put it into practice, which then led to the creation of the bootcamp.
During the 12-week program, each participating organization applies the specification to a selected AI use case from its own environment. Working through common implementation patterns and measurement challenges, participants develop a measurement approach tailored to each system.
The first cohort brings together Angel Cataron (Siemens), Gregoire Vilde (Schneider Electric), Jacques Kluska (Schneider Electric), Laxmi Narayana Bingi (AVEVA), Lisa Zambujinho (Siemens), Olivier Gabriel (Schneider Electric), Pratham Deepak Rao (AVEVA), and Sreenivasan Lakshminarayanan (Amadeus) for six bootcamp sessions running from June through September.
We’ve already completed four sessions, and two more are ahead. Here, we’re sharing what we’ve been working on, how the cohort model helps organizations turn their sustainability goals into a system-specific measurement approach, and what comes next.
Why the SCI for AI Bootcamp Matters
With AI increasingly embedded across products, systems, and workflows, the need for standardized measurement approaches that can be incorporated into governance and decision-making is growing fast. At the same time, frameworks such as the EU AI Act are starting to introduce environmental disclosure, making consistent measurement even more important.
SCI for AI provides a foundation for accurate measurement and credible reporting of AI emissions.
In practice, applying the specification means navigating questions about the system boundary, data access, and measurement approaches while considering individual characteristics of AI systems. For instance, one organization encountered a common problem: its software ran on hardware owned by someone else, so the team couldn’t directly measure how much energy it used. They knew how much data moved through the system, but not how much power the underlying hardware consumed, and it wasn’t clear whether the software provider or the infrastructure owner should be responsible for providing the data.
The cohort model introduces a structured way for organizations to work through these decisions together, compare approaches, and build internal capability around a common methodology—while also generating feedback that can strengthen the current guidance, inform resources, and support adoption.
As part of a wider review process, the program helps connect the specification with real organizational practice, feeding into SCI for AI’s progress toward ISO submission later this year.
“The bootcamp turned what felt like an abstract sustainability goal into a practical engineering problem. That shift in framing from ‘we should measure this’ to ‘here’s how we measure this’ is what made it worthwhile.”
Laxmi Narayana Bingi, AI Developer and Pratham Deepak Rao, AI Engineer, AVEVA
A Closer Look at the SCI for AI Bootcamp
Participating organizations entered the program with different levels of AI sustainability measurement maturity, ranging from initial exploration to organizations that had already established preliminary measurement baselines.
The four use cases span both AI personas—provider and consumer—and a range of architectures, from custom, self-hosted systems to managed cloud services. This gives participants a chance to compare how the methodology works across different systems and operating environments.
The program is organized as a sequence of structured decisions, with each session building on the previous one and leading toward a tailored measurement approach. Starting from understanding the use case, participants move through defining boundaries, mapping the system and data availability, arriving at the approach needed to calculate an SCI for AI score, and translating it into a working blueprint.
“What we appreciate most about the SCI for AI Bootcamp is how practical it has been. Instead of discussing AI sustainability in theory, we have worked through real-world use cases, challenged assumptions, and exchanged experiences with organizations facing similar challenges. The peer exchange has been especially valuable in helping translate an emerging methodology into a practical measurement approach.”
Angel Cataron, Research & Technology Manager, Siemens and Lisa Zambujinho, Sustainable IT Manager, Siemens
During each session and between them, the project team and technical leads support participants through individual implementation questions.
By bringing together professionals across different roles and responsibilities, the bootcamp also helps connect technical teams, sustainability teams, and data owners around the same methodology, making AI impact easier to understand and manage consistently across the entire organization.
What’s Next
Over the summer, participants have continued to develop and validate their individual SCI for AI blueprints, filling gaps around components, data access, and ownership. In September, the cohort will reconvene to review that work together before moving into the final methodology decisions needed to define the calculation approach.
The experience and participant feedback will help refine supporting materials we’ve been developing, including an FAQ and a Getting Started Guide, while we also explore opportunities to share learnings and case studies from the program.
Get Involved
Building on our first bootcamp, we plan to translate these learnings into a repeatable implementation pathway to support more organizations in measuring and reducing the carbon impact of their AI systems.
If your organization is interested in exploring future cohort-based learning opportunities, we’d love to hear from you. Get in touch with Russ Trow at [email protected]
Not a member yet?
To join leading technology organizations already working to measure and manage the environmental impact of their AI systems, visit our membership page.
Thank you to Gosia Fricze, Kirsty Du Plessis, and Russ Trow (GSF) for designing and facilitating these sessions, and to Navveen Balani (GSF) for leading technical support.