Sluicebox and TrustedParts Introduce a Novel Solution for Component-Level Carbon Emissions Data: Generative LCAs

Published: July 7, 2025

A just-published, peer-reviewed white paper explains this new approach for computing carbon emissions and compares it to methodologies currently in circulation, including spend-based, process-based, and hybrid approaches.

“Many approaches focus on top-level energy allocation because it’s easy to verify, but that doesn’t add real value to customers. Such allocations are a good first starting point but often provide false pictures of CO₂ impact as they are highly margin impacted. The real challenge is tackling emissions from the bottom up—optimizing in-house production and chemical use, rather than relying on broad, inaccurate assumptions."

- Martin Hachenburg, Director of Sustainability, Infineon Technologies

One AI use case that has shown considerable promise is the consolidation and analysis of messy and missing data to compute carbon emission statistics, enabling sustainable electronics manufacturing. Having an automated and accurate tool to replace the manual data collection that takes place now to generate a "green" bill of materials (BOM) is something OEMs, like Apple, Google, Cisco, and Microsoft, have been demanding to meet their sustainability goals. Unfortunately, just applying AI to the traditional lifecycle assessment (LCA) process to calculate a product design’s carbon footprint is fraught with challenges. TrustedParts.com has partnered with Sluicebox to employ a novel approach that uses AI to scale expert reasoning, while strictly adhering to ISO 14040/44 standards, building software that truly addresses the issue. They have released a detailed whitepaper explaining the reasoning and research behind their novel approach.

This well-researched and peer-reviewed whitepaper was authored by electronics industry experts and a team of researchers from the fields of AI, electronics, and sustainability, including Dr. Sandeep Chinchali of the University of Texas at Austin; Sridhar Nagarajan and Mrinalini Iyer at Western Digital; Susan Monroe at Vishay Intertechnology; John Archer at TTI (a Berkshire Hathaway Company); and Elmar Kert, Emi Ayada, Piriya Sugumaar, and Sarah Tang from Sluicebox AI.

The paper carefully sets the stage by first comparing methodologies currently in circulation for computing carbon emissions, including spend-based, process-based, and hybrid approaches.

The Five Myths about Carbon Emissions

The paper debunks five myths about emissions in the electronics industry. Clarifying these misconceptions is essential for executives and stakeholders, they assert.

• Myth #1: Spend-based calculations are "good enough."

• Myth #2: Process-based LCAs are too expensive and time-intensive.

• Myth #3: Suppliers can readily provide comprehensive data.

• Myth #4: Primary data is always the most accurate.

• Myth #5: Carbon data is only for regulatory compliance.

The bulk of the paper outlines how Sluicebox’s proprietary AI model, which has been fine-tuned for the electronics industry using ECIA’s subject matter expertise, outperforms generic models.

To verify the accuracy of its dynamic real-time LCAs, Sluicebox has benchmarked its accuracy against human expertise. They used a widely adopted LCA quality evaluation system that assesses data accuracy based on technology, geography, and data recency.

The researchers then present results from three case studies implementing Sluicebox’s real-time LCAs for carbon intelligence at Western Digital, TTI, and Vishay. In each case study, a comparison is made of how LCAs were generated before and after the implementation of the Sluicebox approach.

Conclusion

“Primary data is the ideal North Star. However, not all companies have the mechanisms and processes to apportion resource intensity to every single widget that is produced. Activity-Based Costing was pioneered decades ago. This is the time for Activity-Based Carbon. Generative LCAs can deliver this at scale.”

– Sridhar Nagarajan, Sr. Director of Sustainability, Western Digital

Generative LCAs significantly reduce traditional barriers to accurate emissions data for components, thereby allowing emissions analysis to "scale efficiently across millions of electronic products," the paper concludes. However, the battle isn’t over. Because this challenge requires a cross-industry effort, robust collaboration between LCA platforms and all the players in the complex electronics manufacturing supply chain must occur to ensure success. Diverse industry players, standards, and data sources must be integrated into a cohesive ecosystem.

Unlike typical generative AI models, which routinely generate inaccurate or unsupported "hallucinations," generative LCAs can maintain complete transparency and verifiability because they only use AI to fill in gaps in known data sources. The results come from peer-reviewed literature, primary data, fabs, or validated supplier disclosures.

The authors contend that early adoption of environmental product declarations (EPDs) will be the key to both meeting customer requirements and achieving real transparency about carbon emissions. Generative LCAs can automate the production of EPDs within minutes in accordance with EN 50693. With accurate and reliable EPDs, companies can swiftly improve carbon visibility from minimal baseline levels to near-complete coverage, making informed sustainability interventions possible. Generative LCAs thereby help organizations go from mere compliance to proactive customer disclosures.

The paper’s authors conclude that "generative LCAs signal a new era of environmental intelligence: one that is fast, actionable, and built for scale. For electronics companies navigating fragmented supply chains, customer audits, and intensifying design pressures, this approach provides clarity in the complexity."

Access the entire white paper here: "Precision and Scale in Electronic Carbon Data: Generative LCAs for Green BOMs"