Helping SEMI Address the Emissions Data Gap for Semiconductors

The elephant in the room

A fab can account for what it consumes on site. Electricity, process gases, abatement performance, water. What it cannot account for with the same confidence is what arrived at the loading dock: the photoresist, the CMP slurry, the high-purity deposition precursor, the sputtering target, the etch gas blend that a supplier will not describe in more detail than a product code and a purity grade. This gap leaves companies with an incomplete picture of their emissions. 

The SEMI Semiconductor Climate Consortium and Boston Consulting Group found in the industry's first baselining study that semiconductor devices produced in 2021 carry a lifetime footprint of 500 megatonnes of CO2e, with 16% from the supply chain and 21% from manufacturing. The same report is direct about which part is hard: emissions from the supply chain and from manufacturing process gases "will require considerable research and development to address."

Existing approaches are not meeting the scale or specificity needed

General-purpose life cycle inventory databases handle commodity products well, but lack the coverage of exotic, high purity, and unique inputs necessary for semiconductor manufacturing.

Traditionally, when a specific or unique factor is missing, there were few options available: 

• Spend-based proxies. Multiplying purchase value by an industry-average intensity produces a number that moves when price moves and stays flat when the supplier changes its synthesis route. It reports. It does not inform.
• Nearest-analog substitution. Standing in a generic organic solvent for a formulated resist imports the wrong feedstock, the wrong synthesis energy, and the wrong purification burden, which for electronic-grade materials is often the dominant term.
• Waiting for the supplier. Formulations are proprietary and lifecycle inventories for them are rarely published. A materials supplier that has not built a cradle-to-gate model cannot hand one over unequivocally. Instead, they need to build bespoke reports and obscure confidential data, often taking months or even years before they can share the results with confidence. 
• Exclusion. Dropping uncharacterized inputs from the boundary makes a footprint look complete while quietly removing the part with the most reduction potential.

Each of these produces a reportable figure. None of them produces a clear map to decarbonizing the supply chain.

SEMI is filling the gap 

SEMI connects 4,000 member companies and 1.5 million professionals across electronics design and manufacturing, and it has put that convening power behind decarbonization. SEMI launched the Semiconductor Climate Consortium (SCC) in 2022 with more than 60 founding members to speed decarbonization of the semiconductor value chain. The SCC was founded on the commitment to collaborate on common approaches, build transparency into the ecosystem through annual emissions reporting, and build towards net zero GHG emissions no later than 2050. 

In January 2024, the SCC’s Scope 3 Emissions Working Group released the first alignment document for calculating Scope 3 Category 1, Purchased Goods and Services, covering accounting methodologies, boundaries, quantification methods and data selection. Guidance for Category 11, Use of Sold Products, developed with ERM, followed in June 2025. Before these resources existed, a supplier serving a dozen customers could be asked to account for the same emissions a dozen different ways.

It’s worth noting that included in the Scope 3 Category 1 Guidelines is a specific call-out that given the complexity of the semiconductor value chain, it may not always be feasible to utilize supplier-specific methodologies to allocate emissions. And that the SCC can address this gap by “promot[ing] collective actions, such as engaging with industry efforts to develop life cycle assessments (LCAs) for selected key products covering hot-spot emissions within the SCC ecosystem...” 

In July of this year, SEMI commissioned Muir AI (Muir) to do just that. Using a consortium-led approach to build a list of high-priority materials used across the value chain, Muir will build out  a database of custom emission factors for hotspot materials across a range of geographies. The database will provide a cradle-to-gate product carbon footprint built to ISO 14067, the international standard for quantifying and reporting product carbon footprints. Each hotspot material will carry its provenance with it, so SEMI members can see what was declared in source data, what was modeled, and where a value came from before deciding how much weight to put on it.

"SEMI has already done the hard part of this," said Harris Chalat, CEO of Muir. "The consortium aligned an entire industry on how to account for Category 1 emissions, and then said plainly that the data underneath is missing for the materials that matter most. We're here to help fill that gap, and allow manufacturers and suppliers to focus on decarbonization."

How Muir builds a factor where none exists

The reason a material has no emission factor is almost never that the material is unknowable. It is that characterizing it by hand takes a trained LCA practitioner weeks, and it is impossible to manually characterize when there are thousands of materials in scope.

Using its proprietary methodology, Muir builds a full model of the material. Where a formulation, a synthesis route, or a supplier disclosure is missing, the Platform builds a product twin from its material and manufacturing databases and reconstructs the production path: feedstocks, reaction steps, purification to electronic grade, the energy intensity of each stage, and the grid and geography where it happens. By recursively analyzing the product or material, Muir builds the complete picture of the material’s upstream processes. 

And the results are meant to be shared. Muir is a PACT-conformant solution, so a footprint calculated in the database can be exchanged in a standard format with any of the conformant solutions already in that ecosystem, and our methodology carries third-party assurance to ISO 14067:2018. A factor is only useful to a consortium if it can move between the systems that consortium members already run.

From a footprint to a decarbonization lever

The reason to model a material, rather than proxy it becomes clear at the point of action.

A spend-based figure for a precursor tells a member what to report. A modeled figure for the precursor tells them that most of the burden sits in a purification stage, or in the synthesis of one upstream feedstock, or in the grid serving a single plant. Those are three different decisions: a process change, a supplier substitution, a geographic shift. The insight is inherent in the decomposition, as that is where the reduction opportunity is visible; a single blended number does not have that, leaving users without the data they need to build decarbonization strategies.

It also changes what a supplier conversation can be. A member who can show a supplier a modeled cradle-to-gate breakdown is negotiating about a specific stage of a specific route, not asking an open-ended question about carbon and waiting a quarter (or more) for a spreadsheet.

The industry's reporting obligations are going to keep tightening, and the materials that are hardest to characterize are disproportionately the ones with the highest intensity per kilogram. SEMI decided not to wait for that data to appear on its own, and is investing in building out usable datasets for its consortium. Muir is excited about the opportunity to build the data needed to help leading companies achieve meaningful decarbonization opportunities. 

Talk to us about materials-level carbon data

If your team needs product carbon footprints for inputs that standard databases don’t cover, reach out to Muir and we’ll show you what’s possible. Book a demo.