This chart presents the aggregated energy consumption of 29 semiconductor manufacturers from 2015 to 2024, highlighting how total demand has evolved over time. When users hover their mouse over any reported year, the exact energy-consumption value appears.
In our sample of 29 chip manufacturers, energy consumption grew about 2.5 times over the past 9 years, from 58,326 gigawatt-hours (GWh) in 2015 to 145,595 GWh in 2024.
Semiconductor manufacturing is energy-intensive largely because modern front-end fabrication relies on extremely complex, tightly controlled production steps: more than 1000 process steps, over 50 types of equipment and hundreds of chemicals all operating under strict cleanroom conditions. Equipment for lithography, etching and deposition dominates electricity use. Particularly advanced lithography machines like extreme ultraviolet (EUV) lithography are a main driver – using approximately 10 times as much electricity as conventional 193 nm immersion lithography (Hess, 2024).
As manufacturing advances to smaller nodes, the number of process steps and in turn the electricity needed per wafer increases, further raising energy demand per chip.
In addition to direct energy usage driven by tools used in wafer fabrication which is around 56% of all energy consumption, facility operations and utilities play an important role in shaping the overall energy consumption in chip manufacturing processes: ventilation, HVAC, chillers, vacuum systems, water purification and gas-abatement systems - all run 24/7 and account for ca. 44% of total consumption.
Consequently, electricity – whether generated directly on-site or purchased externally – accounts for the largest share of energy use and becomes one of the biggest challenges in chip production (Hess, 2024).
This chart shows aggregated direct emissions (scope 1) of 29 semiconductor manufacturers from 2015 to 2024, illustrating how total reported values evolved over time. Hovering over the reported year displays the exact emissions.
Direct emissions from our sample of 29 chip manufacturers have more than tripled over the past nine years, rising from 15.4 MMTCO2E in 2015 to nearly 64 MMTCO2E in 2024.
Scope 1 emissions in semiconductor manufacturing come from sources directly controlled by fabs. The primary drivers are process gases – particularly perfluorocarbons (PFCs) and nitrogen trifluoride (NF₃) – used e.g. in plasma etching, chemical vapor disposition, and chamber cleaning. It is important to note that substitution of process chemicals with alternatives characterised by a different environmental footprint has proven successful in lowering emissions. For instance, sulphur hexafluoride (SF₆), the process gas commonly used in plasma etching and chamber cleaning, is often substituted for NF₃ – a gas with similarly high global warming potential, but significantly shorter lifetime in the atmosphere, making it much less harmful for the environment in the long-term perspective (Hess, 2024).
Other Scope 1 emission drivers include on-site fuel combustion and chemical reactions needed for wafer processing, heating and gas handling. Other emission sources, such as fugitive emissions, fixed combustion or mobile combustion, are minor and not always rigorously measured (Nanya, 2024).
As chip manufacturing moves to smaller nodes, the quantity of gases used per wafer and the frequency of on-site chemical operations rise, driving higher Scope 1 emissions per chip.
This chart shows aggregated indirect emissions (scope 2) of 29 semiconductor manufacturers from 2015 to 2024, illustrating how total market-based values evolved over time. Hovering over the reported year displays the exact emissions.
Total market-based indirect emissions show an increase of nearly 60% between 2015 (22.7 MMTCO₂e) and 2024 (35.9 MMTCO₂e).
Scope 2 emissions in semiconductor manufacturing arise from the electricity and purchased energy that fabs consume but do not generate on-site (accounted for in Scope 1). The largest share is attributable to energy (mostly electricity) sourced from external energy suppliers by the fab. Externally sourced electricity can be reported as either location- or market-based.
According to the GHG Protocol, companies should apply dual reporting if they have "any operations in markets providing product or supplier-specific data in the form of contractual instruments" – clearly differentiating between market- and location-based reporting (GHG Protocol).
Market-based emissions are based on companies' procurement strategies, including electricity purchases, supplier offerings, Renewable Energy Certificates (RECs), Power Purchase Agreements (PPAs), etc. These solutions allow companies to source a specific energy mix, even if it is not aligned with or cannot be matched by local grid resources (Rick and Luo 2023). In summary, reporting market-based emissions leads to an underestimation of actual emissions, as companies could report zero emissions in corporate GHG inventories if they matched their power consumption with RECs (Bjørn et al. 2024). Although some RECs – such as unbundled ones from different geographies or years – face criticism for not representing true renewable energy use or adding capacity, companies that focus on PPAs or source RECs close to where they are produced adopt more credible approaches to reducing emissions from electricity consumption. This is because PPAs directly support the generation of additional renewable energy in the relevant grid, ensuring that the claimed emissions reductions correspond to actual decreases in fossil-based electricity consumption (Bjørn et al. 2024).
As manufacturing moves to smaller nodes and more advanced manufacturing processes, both tool operation and facility support systems demand more electricity per wafer. In our sample of 29 chip manufacturers, total Scope 2 emissions have grown – although not to the same extent than energy consumption. This indicates that some companies are offsetting part of their electricity use through Renewable Energy Certificates (RECs).
Because Scope 2 emissions depend on purchased electricity, the carbon intensity of the local power grid also strongly affects their magnitude, evidenced by location-based emissions.
This chart shows aggregated indirect emissions of 12 semiconductor manufacturers from 2015 to 2024, illustrating location-based scope 2 emissions evolved over time. Hovering over the reported year displays the exact emissions.
Please note that this chart focuses solely on location-based emissions, which are reported less consistently than market-based values. In the most recent reporting period, 12 out of 29 semiconductor manufacturers did not provide location-based emissions data, making market-based figures a more reliable way to compare indirect emissions across the industry.
Scope 2 emissions in semiconductor manufacturing arise from the electricity and purchased energy that fabs consume but do not generate on-site (accounted for in Scope 1). The largest share is attributable to energy (mostly electricity) sourced from external energy suppliers by the fab. Externally sourced electricity can be reported as either location- or market-based.
Location-based emissions reflect the average carbon intensity of the electricity grid where the facility is physically located, independent of the company's specific procurement choices (GHG Protocol). They are determined by the mix of power sources feeding the local grid, including fossil fuels, nuclear, and renewables, and capture the actual environmental burden imposed by the fab on its regional electricity system (Hess, 2025). Unlike market-based reporting, which can reflect lower emissions through contractual sourcing of renewable electricity, location-based emissions cannot be offset in this way and therefore represent the real emissions occurring in the local grid supplying the fab (Hess, 2025).
In semiconductor manufacturing, location-based emissions are strongly influenced by the energy intensity of advanced fabrication processes, the efficiency of facility support systems, and the carbon profile of the local grid (Hess, 2024). As fabs move to smaller nodes and more complex processes, electricity consumption per wafer increases, amplifying the impact of grid carbon intensity on total Scope 2 emissions. Because location-based emissions depend on the local grid, reductions require either improvements in energy efficiency, on-site generation of low-carbon power, or decarbonization of the regional grid itself (Hess, 2024).
These values provide a complementary perspective to market-based reporting, offering a more direct measure of the environmental impact of semiconductor operations on their surrounding energy systems. Interface's Semiconductor Emission Explorer shows that location-based Scope 2 emissions for 29 chip manufacturers are consistently higher than market-based values, reflecting both rising electricity demand and the carbon intensity of the grids supplying fabs (Hess, 2025).
This chart shows aggregated up- and downstream emissions of 29 semiconductor manufacturers from 2015 to 2024, illustrating how total reported values evolved over time. Hovering over the reported year displays the exact emissions, enabling users to track precise figures and observe year-to-year trends.
In our sample, Scope 3 emissions reporting rose sharply - from 9 companies in 2015 to 23 in 2024. Reported values also increased, driven by higher production volumes and more complex chip designs.
Scope 3 emissions in semiconductor manufacturing cover all indirect emissions outside the reporting company's operational control, including the entire semiconductor value chain. Reliable data for Scope 3 emissions is still limited: many companies do not report these emissions at all, and among those that do, reporting often varies in granularity and scope, making comparisons across manufacturers challenging. Based on the available CSR reporting, the largest Scope 3 contributors are the production and transport of raw materials as well as outsourced wafer and component manufacturing (Hess, 2025).
Other important drivers include logistics and distribution, the energy consumed by products during their end phase, and end-of-life treatment. Because semiconductor manufacturing relies on highly specialised inputs and global supply chains, Scope 3 emissions often exceed the direct (Scope 1) and electricity-related (Scope 2) emissions of the fab itself.
Our latest data update highlights a shift in the Scope 1 emissions trend, reversing the previous slowdown and signaling renewed growth in direct emissions within the semiconductor industry. Our full data brief explaining the drivers behind this trend can be consulted here.
Comparing Scope 1, 2, and 3 trends, we observe the following:
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.