Chip manufacturing is very emission- and energy-intensive. The main driver is the complexity of front-end fabrication — involving more than 50 equipment types, around 300 chemicals, and over 1000 process steps. The more advanced the manufacturing process, the more water, energy, and chemicals it requires.
To point out pressing challenges – fabs depend on fluorinated process gases with high global warming potentials, which account for up to 90% of direct emissions, and on water- and resource-heavy upstream supply chains. Furthermore, our previous research showed that front-end fabrication can be 4 to 13 times more energy intensive than back-end processes, on average around (insert number on energy) (Hess, 2024).
Against this backdrop, tracking industry reporting and performance is crucial. Semiconductor firms are obligated to publish sustainability disclosures under frameworks such as corporate sustainability reporting frameworks or national stock-exchange requirements, while others report voluntarily to meet investor expectations. Many companies choose to align their Corporate Social Responsibility (CSR) reporting with global norms like the GHG Protocol, an international standard for measuring and managing greenhouse gas emissions. This is an important point of reference as it constitutes the most standardised publicly available source of information on companies' emissions and energy use.
CSR reports typically measure GHG emissions in metric tons of carbon dioxide equivalents (MTCO2E), differentiating between three categories or scopes defined in the GHG Protocol, (Hess 2024; The Greenhouse Gas Protocol Revised Edition):
It is important to note that CSR reporting comes with data that is often inconsistent and difficult to compare. This lack of transparency makes it harder to understand the real footprint, which proves particularly relevant in the context of Europe's plans to scale its AI compute infrastructure, a move that is likely to cause emissions from chip manufacturing to significantly increase in the upcoming years. Comprehensive reporting is therefore essential to anticipate environmental risks, guide policy initiatives and ensure that expanding supply security does not increase an already significant ecological burden.
Nevertheless, limitations lead to different reporting boundaries across companies, methodologies vary, and some upstream or process-level details remain undisclosed.
This microsite aims to increase transparency, make data gaps visible, and support better-informed discussions on how the semiconductor industry can reduce its climate and environmental impact. By gathering and updating information directly from companies' annual CSR reports, the tracker offers an accessible way to see how disclosures are evolving and where improvements are still needed. The microsite is updated annually and aims to serve as a publicly available resource presenting the most up-to-date overview of semiconductor emissions trends.
The Emission Tracker section presents all available data extracted from CSR reports published by semiconductor manufacturers in the form of interactive charts:
This microsite was developed as a continuation of the Semiconductor Emission Explorer, an interface quantitative analysis published in spring 2025 that visualizes the evolution of greenhouse gas (GHG) emissions and energy consumption from global chip production starting from 2015. The study also examined key problems of collecting data from publicly available CSR Reports; showing that CSR data is often inconsistent, incomplete, or difficult to compare across companies. In addition, the previous work also highlighted the importance of contextualizing CSR trends with broader industry dynamics. By adding this second analytical dimension, our study helped identify underlying drivers and reasons for observed developments.
The Semiconductor Emission Explorer study, published in spring 2025, provides contextual guidance for the data available from 2015 to 2023. While this microsite is based on an updated dataset - incorporating the most recent CSR disclosures from the last calendar year - it does not include the market data featured in the original study. Therefore, the explanations and analytical framework presented in the initial study remain highly relevant. They help clarify how companies report their emissions, factor in market dynamics, explain why data gaps occur, and what limitations affect comparability across the sector. Users who want this additional background may find the study a useful companion resource.
Our analysis is based on data published in the annual CSR reports of the largest global semiconductor manufacturers (based on production capacity) from 2015 to 2024. Prior to 2015, only a handful of companies provided information on CSR. We focus on gathering data published on the following emission indicators: direct (Scope 1), indirect (Scope 2) and upstream and downstream emissions (Scope 3) according to the GHG Protocol. To maintain consistency, we excluded chip manufacturers focused primarily on optoelectronics, such as LEDs, because their manufacturing processes differ significantly from those for logic, memory, analog and other chip types (OECD 2024).
We analysed chip manufacturers' reporting based on the GHG Protocol. The GHG Protocol is a standardised framework for global carbon disclosure and serves as guidance for companies and other organisations to measure and manage GHG emissions (Greenhouse Gas Protocol 2024). It covers the accounting and reporting of seven GHGs according to the Kyoto Protocol (UN Climate Change 2025): carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), sulphur hexafluoride (SF6) and nitrogen trifluoride (NF3) (Ranganathan et al. 2004). The initiative is rooted in a multi-stakeholder partnership of businesses, non-governmental organisations and others. It was founded by the World Resources Institute (WRI) and the World Business Council for Sustainable Development (WBSCD) in 1998 (World Business Council for Sustainable Development 2004). Even though the GHG Protocol itself is not a binding regulatory framework, it has been widely adopted across several industries as part of their voluntary sustainability and emissions management efforts. Furthermore, emission-heavy industries, such as aluminum and cement, partnered with the initiative to develop industry-specific calculation tools (World Business Council for Sustainable Development 2004). Additionally, some regulatory bodies, such as the European Union Emissions Trading System or the United Kingdom's GHG reporting programme, have already referenced and incorporated elements of the GHG Protocol into their regulations or standards (Department for Energy Security; European Commission).
In our original dataset used for the spring 2025 publication, we considered all companies with more than 100,000 wafer shipments per month, as identified in the SEMI World Fab Forecast 2023 (SEMI World Fab Forecast). This resulted in an initial dataset of 59 companies, covering 82% of the global manufacturing capacity in 2023. For the purposes of the original study, usable data were available for 28 of these companies. However, for the purpose of this site, we were able to incorporate the necessary data from one additional company compared to the original study - covering 68% of the total global manufacturing capacity in 2024. The microsite therefore draws solely on data from 29 relevant players for the following reasons:
We manually calculated the data due to missing details for specific years in Samsung's CSR reports, as excluding these gaps would have significantly affected overall trends (see table below). In any other case, we did not estimate values for missing data points over the years. The interactive chart transparently depicts missing data points for specific companies. Please refer to the attached Excel sheet for further details.
Samsung, the largest manufacturer by global capacity, operates in two divisions: Device eXperience (DX), which produces finished products (e.g., TVs, washing machines and smartphones), and Device Solutions (DS), representing the chip production segment (Samsung CSR Report 2024). Before 2020, Samsung reported aggregated CSR data for both divisions. Since 2020, they have provided separate figures for the DS division for Scope 1 and Scope 2, as well as energy consumption. Since 2022, they have also provided separate figures for Scope 3. To estimate DS-specific data for 2015–2020 for Scope 1, Scope 2 and energy consumption as well as DS-specific data for 2015–2021 for Scope 3, we used ratio-based allocation, deriving proportions from 2020–2024 division-specific data.
On average (2020–2024), DS accounted for 95,40% of direct emissions (S1), 92,07% of indirect emissions (S2) and 87,15% of energy consumption. For upstream and downstream (S3) emissions, the average for the years 2022 and 2024 was 14,29%. Using these ratios, we applied the mean values to the aggregated CSR data for direct emissions, indirect emissions and energy for the years 2015–2020 and Scope 3 emissions for the years 2015–2021:
| Electricity | Energy | Scope 1 | Scope 2 | Scope 3 | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| DS | Total | % | DS | Total | % | DS | Total | % | DS | Total | % | DS | Total | % | |
| 2015 | 13.629 | 15.368 | 88.68% | 16.826 | 19.478 | 87.38% | 2.33 | 2.44 | 95.36% | 7.19 | 7.75 | 92.84% | 3.00 | 18.53 | 16.21% |
| 2016 | 14.709 | 16.587 | 88.68% | 18.203 | 21.073 | 87.38% | 2.43 | 2.55 | 95.36% | 8.40 | 9.05 | 92.84% | 1.24 | 7.65 | 16.21% |
| 2017 | 16.362 | 18.450 | 88.68% | 20.230 | 23.419 | 87.38% | 3.49 | 3.67 | 95.36% | 9.20 | 9.91 | 92.84% | 2.40 | 14.78 | 16.21% |
| 2018 | 18.231 | 20.558 | 88.68% | 22.484 | 26.028 | 87.38% | 4.63 | 4.86 | 95.36% | 9.56 | 10.30 | 92.84% | 2.58 | 15.91 | 16.21% |
| 2019 | 18.765 | 21.160 | 88.68% | 23.236 | 26.899 | 87.38% | 4.83 | 5.07 | 95.36% | 8.11 | 8.73 | 92.84% | 2.69 | 16.61 | 16.21% |
| 2020 | 19.654 | 22.916 | 85.77% | 24.556 | 29.024 | 84.61% | 5.45 | 5.73 | 95.14% | 7.55 | 9.08 | 83.11% | 2.39 | 14.73 | 16.21% |
| 2021 | 22.624 | 25.767 | 87.80% | 27.926 | 32.322 | 86.40% | 7.34 | 7.60 | 96.54% | 8.27 | 9.80 | 84.41% | 19.98 | 123.23 | 16.21% |
| 2022 | 25.249 | 28.316 | 89.17% | 30.850 | 35.177 | 87.70% | 5.72 | 5.97 | 95.75% | 8.97 | 9.08 | 98.77% | 14.76 | 124.72 | 11.84% |
| 2023 | 27.042 | 29.956 | 90.27% | 32.384 | 36.399 | 88.97% | 3.52 | 3.73 | 94.35% | 9.46 | 9.56 | 98.93% | 18.09 | 119.73 | 15.11% |
| 2024 | 28.996 | 32.083 | 90.38% | 34.592 | 38.772 | 89.22% | 4.49 | 4.72 | 95.01% | 10.06 | 10.16 | 98.96% | 22.90 | 105.61 | 21.68% |
| Avg | 88.68% | 87.38% | 95.36% | 92.84% | 16.21% | ||||||||||
| Previously | 88.25% | 86.92% | 95.45% | 91.31% | 13.47% | ||||||||||
For full details on the methodology we used, please refer to the Semiconductor Emission Explorer publication.
Our data brief, TITLE HERE, explores how the global AI boom driving up chip demand may be causing an increase in direct emissions from semiconductor manufacturing.