Transparency, Ambition,
and Collaboration:
Advancing the Climate Agenda of the
Semiconductor Value Chain
Table of Contents
03
05
07
09
11
14
16
Why this report is critical
Baseline of value chain emissions
Electricity as the largest lever
Investment and innovation to address
the remaining 16%
Future manufacturing emissions scenarios
Dilemma of value chain emissions
Semiconductor Climate Consortium aspirations
Introduction
Takeaway 1
Takeaway 2
Takeaway 3
Takeaway 4
Takeaway 5
Conclusion
03
Why this report is critical
Global organizations such as the United Nations are increasingly concerned
that the world is not on a path toward net zero by 2050, resulting in increased
urgency for virtually all industries to forge concrete commitments that result in
tangible actions to reduce greenhouse gas emissions. Not surprisingly, then, as
the semiconductor industry’s value chain activities expand rapidly, the industry’s
carbon footprint is drawing more attention.
Semiconductor manufacturing, including electronic design automation and
intellectual property (EDA & IP), wafer fabrication, chip design, and package,
assembly, and test, is directly responsible for 0.3% of global carbon emissions
today and induces another 1% in upstream and downstream suppliers and users.
Companies in the industry recognize the need to come together to address the
largest and most difcult sources of emissions and set the industry on a path
toward net zero by 2050 within the 1.5°C pathway.
Past studies have attempted to provide visibility into the sources and quantities
of emissions from semiconductor production. However, these studies have
not been sufciently comprehensive across the entire value chain, failed to
examine each source of emissions in depth, and did not offer credible forecasts
of how emissions may shift in the future. To address these shortcomings, BCG,
SEMI, and the Semiconductor Climate Consortium (SCC) partnered to examine
greenhouse gas (GHG) emissions from every source and from the supply chain,
manufacturing processes, and device usage (see Exhibit 1).
Our goal is to distinguish this report by its comprehensiveness and the co-
authorship of climate and sustainability leaders across the semiconductor
value chain. Working with industry experts, we undertook the most expansive
analysis to date of the semiconductor industry’s current emissions prole and the
possible future scenarios. This analysis is built upon 1 million data points across
200 companies in the semiconductor value chain that account for 80% of the
emissions we studied; in all, we analyzed emissions involving fteen types of
greenhouse gases and nearly three dozen types of greenhouse gas sources. The
authors deeply appreciate the support and partnership of CDP (www.cdp.net),
the primary source of data for this effort.
Introduction:
Why this report is critical
04
Exhibit 1: Semiconductor value chain activities
Five takeaways from the research:
1. Baseline of value chain emissions: Semiconductor devices produced in 2021
have a lifetime CO
2
e footprint of 500 megatonne (MT), 16% from supply
chain, 21% from manufacturing, and 63% from device use.
2. Electricity as the largest lever: Low-carbon energy sources can address
>80% of industry emissions primarily by reducing electricity used to
manufacture and then use devices, which could be achieved with bold and
decisive investments in low-carbon energy.
3. Investment and innovation to solve remaining 16%: Emissions from
the supply chain and from manufacturing process gases will require
considerable research and development to address, necessitating
investments now.
4. Future manufacturing emissions scenarios: Current government and
company commitments will substantially reduce manufacturing emissions,
but they are still forecasted to overshoot the carbon budget for
the 1.5°C pathway.
5. Dilemma of value chain emissions: Digital technologies which require
semiconductors play a big and crucial role in reducing energy use and
emissions across industries.
Manufacturing
N-tier Supply
Chain
Parts and
Components
Direct Inputs
Research and
Design
Fabrication
and Assembly
Processing
Sold
Products
Use of Sold
Products
End
of Life
Supply Chain Device Use
Not exhaustive
Metals
Raw
Materials
Chemical
Extraction
Subsystems
Purified Silicon
and Metals
Commodity
Chemicals
Equipment
Wafers
Process
Chemicals
IP Block /
Core
EDA
Software
Circuit
Design
Wafer
Fabrication
Package /
Assembly
Product
Testing
Devices
Infrastructure
Datacenters
Consumers
Companies
Governments
Recycling
Disposal
Source: BCG analysis
Exhibit 1.
Semiconductor value chain activities
Introduction:
Why this report is critical
05
Takeaway 1:
Baseline of value chain emissions
Baseline of value chain emissions
The most fundamental ndings from the study – indeed, the necessary starting
point for decarbonization roadmaps – involve the magnitude and sources of
greenhouse gas emissions (see Exhibit 2). The deep interdependencies of
this industry to produce such a complex product mean that most emissions
reduction solutions will not come from within one company or activity but from
collaboration across many.
Exhibit 2: Lifecycle emissions of a semiconductor or device
(Megatonnes CO
2
e, 2021)
Supply chain emissions are calculated using wafer fabrication company’s reported
upstream scope 3 emissions as these companies have the best visibility into the
equipment, materials, and services required during manufacturing. The supply
chain accounts for 16% of a semiconductor’s lifetime emissions, over half of which
are from purchased goods and services used in manufacturing processes.
Semiconductor manufacturing is calculated directly from company reported
emissions performing these activities. Over 20% of semiconductor emissions are
produced during design, fabrication, and packaging and testing. Of manufacturing
emissions, 65% is from electricity to power equipment and buildings. The
remaining 30% is a result of direct company actions such as using process
!"#$%&'()*+*,-,.*/$*0"1)"2*!"2134)&2'*5%"3#6*744*%&'()1*%818%98:6
74
100
299
Supply Chain Manufacturing Device Use
99%
from use of
sold products
from purchased
goods and services
16% 21% 63%
65%
Source: CDP, BCG analysis
Exhibit 2.
Lifecycle emissions of a semiconductor
(Megatonnes CO2e, 2021)
Share
of total
06
chemicals that enter the atmosphere during manufacturing.
Device use is calculated using chip designers’ reported scope 3 downstream
emissions, as they interface directly with end customers and have the most
visibility. Device use accounts for 63% of a semiconductor’s lifetime emissions
due to collective energy consumption of billions of semiconductors in products
worldwide. Chips consume electricity in phones, computers, cars, networks,
datacenters, and just about any other electronic device you can think of.
I t ’ s i m p o r t a n t t o n o t e t h a t a l l g u r e s
in this report use
location-based energy emissions,
which relies on average grid energy
availability and does not consider any
market-based tools to achieve lower
carbon intensity from energy use.
Typical market-based tools chosen by
some companies are power purchase
agreements to accelerate their and the
entire grid’s adoption of low-carbon
energy sources. These tools have a
signicant impact on emissions across
the entire semiconductor value chain.
In fact, the largest purchasers of low-
carbon energy are large hyperscalers,
which operate datacenters powered
by semiconductors. Therefore,
a semiconductor deployed in a
datacenter will likely have much
smaller-than-average market-based
emissions when compared to location-
based emissions for device use
(see Exhibit 3).
The SCC acknowledges that measuring, tracking, and inuencing the trajectory
of emissions from device use are exceptionally challenging because the emissions
output is driven by global consumer behavior and low-carbon energy availability.
Further, as is true for many industries, there is limited company reported data
quantifying scope 3 semiconductor emissions. While 96% and 92% of the
companies we studied reported scope 1 and 2 emissions to CDP respectively,
only 70% reported any scope 3. And the reported data varies in granularity and
completeness, sometimes utilizing different calculation methodologies. Overall,
these factors tend to result in underreporting. As a result, the industry recognizes
that scope 3 data has the greatest opportunity and need for improvement – and
as that data is captured more complete emissions estimations will likely increase.
Exhibit 3:
Impact of market-based tools on use
phase emissions
!"#$%&'()*+*,-,.*/$*0"1)"2*!"2134)&2'*5%"3#6*744*%&'()1*%818%98:6
16%
Ma
nufacturing
Device Use
Average:
Location-based
63%
21%
Supply Chain 39%
53%
Datacenter:
Market-based
8%
Source: CDP, BCG analysis
Exhibit 3.
Emissions from device use
diminished with market-based tools
Takeaway 1:
Baseline of value chain emissions
07
Electricity as the largest lever
Reaching net zero by 2050 and minimizing carbon expenditure should start by
pulling the largest levers with the most available solutions. For this value chain,
as for many others, the largest single lever is low-carbon electricity. Eighty-three
percent of semiconductor device emissions are tied to generating electricity
consumed by activities across its lifecycle, reported as scope 2 by individual
companies (see Exhibit 4).
Exhibit 4:
Emissions by Relation to Electricity Generation (Megatonnes CO
2
e, 2021)
There are three primary actions semiconductor companies have been taking and
must accelerate to eliminate emissions from electricity generation. First, building
and designing more efcient manufacturing facilities and ofces which consume
less electricity. Second, working with suppliers to use less electricity and to
manufacture more energy-efcient equipment and materials. Finally, in partnership
with device users, designing and manufacturing more energy-efcient devices.
These three actions are critical to addressing >80% of value chain emissions,
but progress on energy efciency has historically been outpaced by increased
consumption and utilization of devices. That is, as semiconductors become
more economical to purchase and operate, consumers demand more. If energy
consumed by semiconductors continues to increase, reducing energy-related
!"#$%&'()*+*,-,.*/$*0"1)"2*!"2134)&2'*5%"3#6*744*%&'()1*%818%98:6
46
65
296
31
35
0
100
200
300
Supply Chain Manufacturing
MT
77
100
Device Use
Non-Electricity Related Emissions Emissions from Electricity Generation
Note: Figures use location-based data Source: CDP, BCG analysis
Exhibit 4.
Emissions by Relation to Electricity Generation
(Megatonnes CO2e, 2021)
Total
3
391
(83%)
82
(17%)
Takeaway 2:
Electricity as the largest lever
08
emissions will require investments to accelerate transitioning the global electrical
grid to low-carbon energy. This process has already started, but the transition
rate varies greatly by market. For instance, it is especially difcult to source low-
carbon energy in markets where semiconductor manufacturing emissions are
greatest, such as Mainland China, Taiwan, South Korea, and Japan (see Exhibit 5).
Further, reducing supply chain and device use emissions from electricity requires
access to low-carbon energy for consumers and companies worldwide.
To speed up the transition to low-carbon energy, semiconductor producers
and their suppliers and customers must invest in low-carbon onsite energy
production, purchase available low-carbon energy, and advocate for a faster grid
transition, especially where semiconductors are manufactured and used. All while
continuing to reduce energy consumption.
The full global transition to low-carbon energy will require robust and
collaborative efforts among multiple parties to address the foremost challenges
in attaining net zero status. This calls for the private sector, public advocates,
government bodies, and low-carbon energy providers and energy investors to
unite to forge cooperative partnerships, make courageous decisions, allocate
assets, and take bold, decisive actions. This collective endeavor can alter the
trajectory of carbon emissions, leading to the realization of ambitious
net zero goals.
USA
7
21%
Europe
5
3300%%
Japan
10
1122%%
12
1100%%
Mainland China
15
1122%%
South Korea
11
1177%%
Other
2
5
2021 energy
emissions
Mt CO
2
e
% low-carbon
energy of
region grid
1As of 2020
2Other regions include, but is not limited to, Malaysia, Israel, Singapore, and the Philippines
Source: CDP, 2021 IEA, BCG analysis
Taiwan
1
Takeaway 2:
Electricity as the largest lever
Exhibit 5: Sourcing electricity to reduce manufacturing emissions from
energy generation requires investing to increase low-carbon energy
availability
09
Investment and innovation to address
the remaining 16%
While emissions from electricity generation can be addressed through known
and viable solutions, this is not the case for emissions from many other sources.
Emissions from extracting and rening materials used in semiconductor
manufacturing are difcult to eliminate, especially for lower-margin activities and
when materials renement requires high heat or unique chemical processes.
Companies in the semiconductor supply chain are pursuing and achieving their
own reduction targets to address their emissions. Still, as major customers,
semiconductor manufacturers can collaborate with their existing suppliers in
developing equipment and materials with smaller carbon footprints. Additionally,
they can engage with new suppliers to explore ways of further
reducing emissions.
Direct manufacturing emissions, which an individual company would report
as scope 1, account for about 7% of value chain emissions and eliminating
them is very challenging; this requires altering ne-tuned, highly complex
production methodologies and recipes. Of the process gases used, peruorinated
compounds (PFCs) have the greatest global warming potential (GWP), 100 to
20,000 times greater than CO
2
due to their unique chemical properties which
also make them so extraordinarily useful (see Exhibit 6). Alternatives for critical
gases such as PFCs are now being evaluated by leading research
institutions worldwide.
Exhibit 6: Direct semiconductor manufacturing emissions (Mt CO
2
e, 2021)
09
!"#$%&'()*+*,-,.*/$*0"1)"2*!"2134)&2'*5%"3#6*744*%&'()1*%818%98:6
PFCs are fundamental to semiconductor
manufacturing due to unique properties
NF3 PFCs
1
SF6N2O CO2HFCsHTF
1.4
Other
1.0
34.7
2.2
2.5
2.7
4.6
6.1
14.2
Total
Chemical
Stability
Electrical
Properties
Non-Reactive &
Non-Flammable
Composed of
strong carbon-
fluorine bonds
Exhibit high
resistance to
degradation in
use and in
environment
Used in etching
process due to
low polarize-
ability, resulting
in weak
intermolecular
interactions
Non-reactive
nature makes it
safe for use
Forms reactive
species in
plasma which is
used for etching
in manufacturing
Acronyms: Heat Transfer Fluids, Hydrofluorocarbons, Nitrous Oxide, Sulfur Hexafluoride, Carbon Dioxide, Nitrogen Trifluoride, Perfluorinated Compounds. 1 PFCs include carbon
tetrafluoride (CF4), hexafluoroethane (C2F6),fluoroform (CHF3), octofuoropropane (C3F8), octafluorocyclobutane (C4F8).
!
Exhibit 6.
Direct semiconductor manufacturing emissions (Mt CO2e, 2021)
Source: CDP, ECHA, BCG analysis
Takeaway 3:
Investment and innovation to address the remaining 16%
10
Companies are also actively exploring various technologies to convert or capture
high GWP gases to reduce emissions. Abatement technologies are a key part
of this solution, and electrication is driving the development of more efcient
solutions. However, we anticipate this to be a long journey for the industry, with
many deeply technical challenges remaining to be solved.
Investment in research and development to decarbonize the supply chain
and manufacturing processes is critical to accelerate now and will require
coordination across many players of the value chain to realize return.
International research institutions are working with companies to address these
sources of emissions and help companies surpass their emission
reduction targets.
Takeaway 3:
Investment and innovation to address the remaining 16%
11
Future manufacturing emissions scenarios
There is a viable pathway to dramatically reduce emissions from semiconductors
over the following years and decades. To map out this pathway and possible
obstacles to limiting global warming to 1.5°C by 2050, we forecasted how
emissions from semiconductor manufacturing activities could shift under
different conditions. Given the current difculties in quantifying supply chain and
device usage emissions, we did not include them in our forecasts.
First, viewing global greenhouse gas emissions benchmarks broadly, the 1.5°C
pathway scenario requires a 43% reduction in GHG emissions from 2019 to 2030,
an additional 50% reduction to 2040, and then net zero emissions by 2050,
according to the 2022 Intergovernmental Panel on Climate Change (IPCC) Sixth
Assessment Report. This analysis allocates 1.0 gigaton (Gt) of emissions to the
whole semiconductor industry during this period. However, from 2019 to 2050,
even if company CO
2
reduction commitments are met, the industry is unlikely to
meet this goal, instead emitting 3.5 Gt. (see Exhibit 7).
Takeaway 4:
Future manufacturing emissions scenarios
2018 2020 2022 2024 2026 2028 2030 2032 2034 2036 2038 2040 2042 2044 2046 2048 2050
50
100
150
200
0
Manufacturing Emissions
(Mt CO
2
e)
2030:
Emissions peak
according to announced
pledges
2045:
Industry peak with no
company action
Emissions if company
commitments are realize
d
Trajectory with IEA STEP
s
low-carbon energy scena
rio
Total Carbon
expenditure
4.8 Gt
Total Carbon
expenditure
3.5 Gt
Total Carbon
expenditure
1.0 Gt
60 Mt
168 Mt
Low-carbon Energy Announced Pledges 1.5C˚ IPCC Pathway
1 Emissions growth based on projected capacity growth (3.25%) and average intensity growth (1.01%), all else constant
Note: Low-carbon energy scenarios use IEA STEPS for North America, Europe, and Asia-Pacific. Source: BCG analyses on data from: CDP, imec, SEMI, IEA
Exhibit 7:
Announced sustainability pledges reduce total 2019–2050 emissions
by 30% and are within 60 Mt CO
2
e of 2050 net zero
12
Takeaway 4:
Future manufacturing emissions scenarios
Looking more closely at the scenarios, in the low-carbon energy analysis,
semiconductor industry emissions will atten and stabilize in the next 20-25 years
if the IEA Stated Policies Scenario (STEPS) – based on policymaker low-carbon
energy goals in North America, Asia Pacic, and Europe – is met. In this scenario,
the only action taken by the semiconductor industry would be to help accelerate
the adoption of low-carbon energy. But this outcome leaves chip manufacturing
emissions at 168 MT, well above net zero and carbon expenditure targets.
In the announced pledges scenario, company commitments by the forty highest
emitters in semiconductor manufacturing further reduce emissions in 2050 by
70%. However, this still falls short of reaching net zero emissions in 2050 by
about 60 Mt and results in 3.5x the carbon budget.
Drilling down into company commitments to further identify contributors
to the carbon expenditure, we anticipate that directly controlled emissions,
or the emissions produced from the manufacturing process itself, will
experience an average annual increase of 5% through 2035 (see Exhibit 8)
before starting to decline as signicant net zero milestone years of 2040
and 2050 are approached. Low-carbon energy adoption will cause
electricity-related emissions to begin declining at 9% on average annually
from 2030 to 2050, becoming less than direct emissions by 2041.
2018 2020 2022 2024 2026 2028 2030 2032 2034 2036 2038 2040 2042 2044 2046 2048 2050
0
50
100
150
Mt CO
2
e
2030:
Emissions peak according to announced pledges
2041:
Direct manufacturing emissions pass
emissions tied to energy use
Directly controlled Electricity consumption Total Manufacturing
Note: Forecast uses IEA STEPS for North America, Europe, and Asia-Pacific
Source: BCG analyses on data from: CDP, imec, SEMI, IEA
Exhibit 8:
With current announced pledges by governments and companies,
direct emissions from manufacturing will surpass those from electricity use
by 2041
13
Takeaway 4:
Future manufacturing emissions scenarios
These projections clearly show that more companies must take more aggressive
steps to reduce emissions over the next 25 years to reach net zero and stay
within the carbon budget. Further, the global transition to renewable energy will
not solve the problem itself. Proactive company action to increase and accelerate
clean energy adoption must be paired with aggressive efforts to reduce direct
manufacturing emissions.
It is important to note that the industry is cyclical and emissions will uctuate
with shifts in economic conditions that impact utilization rates year-to-year.
The forecasted scenarios represent an average view over 30 years rather than
precise emissions estimates in any particular year. Further, since nearly no public
commitments provide annual-level targets, the slope of the emissions forecast is
an estimation that will be highly impacted by the timing of substantial actions,
affecting the estimate of the total carbon expenditure over time.
14
Dilemma of value chain emissions
While the industry contends with its emissions, manufacturing expansion and
advancement are required to, among other things, support the global economy
and limit the impacts of climate change on all industries, adding an intriguing
level of nuance to the semiconductor emissions story.
Semiconductors are required for climate change solutions such as
electrication, low-carbon energy, electric vehicles, and digitalization, which
have led to a reduction in global emissions of 1 to 2 gigatons in 2020 with
potential to enable annual emissions avoidance of 2 to 5 Gt by 2025
2
.
In addition to more chips, the world demands greater computational power for
applications, such as articial intelligence, which could aid in climate change
solutions. Advanced semiconductors require more steps to manufacture,
and these steps utilize more electricity and chemicals. Growing emissions
intensity per unit of production compounds the emissions impact of increasing
manufacturing output.
However, companies in the semiconductor value chain are not accepting the
status quo. Many are already taking steps to reduce emissions in alignment
with the 1.5°C pathway and are working toward roadmaps to reach net zero by
switching to low-carbon energy, collaborating with suppliers and customers,
and changing manufacturing processes.
Semiconductor Climate Consortium Aspirations
2
Goldman Sachs. (2023). GS SUSTAIN: Avoided Emissions: How quantifying Avoided Emissions can broaden the
decarbonization investment universe.
Takeaway 5:
Dilemma of value chain emissions
15
Conclusion:
Semiconductor Climate Consortium aspirations
There is still a lot of work to be done, but companies in our industry
are committed to doing it together, and we see tangible signs of
progress and actions. With the understanding that the quality of this
report and ongoing tracking of emissions are a function of reported
data, the SCC continues to encourage full transparency from all industry
companies. Over the past few years, companies across the value chain
have increased their efforts to report GHG emissions more accurately.
Since the inception of SCC, working groups have been established within
scopes 1, 2, and 3 to measure and track the industry’s performance in
each area, share best practices with peers, and drive concrete actions.
The climate consortium has three key focus areas – (i) Evolving toward full
transparency within the value chain; (ii) Enabling a swift transition to low-
carbon energy, impacting their own operations, decarbonizing upstream
suppliers, and addressing product use energy; (iii) Collaborating with
research institutions to nd solutions for high GWP process gases.
The scope 1 working group is focused on sharing mitigation strategies
for high GWP greenhouse gases. In parallel, the scope 2 working group
is focused on three areas: rst, accelerating the adoption of low-carbon
energy; second, improving access to low-carbon energy in Asian operations;
and third, discovering novel methods for low-carbon energy. In the scope
3 working group, members are studying how to increase transparency
into upstream emissions and identify actions to decarbonize them.
These actions aim to steer the industry’s commitments and actions toward
a 1.5°C pathway. The Baseline, Ambition Setting, and Roadmapping
working group, which was responsible for this report with BCG, is now
developing short and long-term targets that the value chain can use to
hold itself accountable in the journey to net zero. Although substantial
work lies ahead, the industry is highly motivated to pursue the appropriate
and practical measures required to reach the collective ambition.
16
Gaurav Tembey
Managing Director and Partner
Trey Sexton
Project Leader
Chris Richard
Managing Director and Partner
Ramiro Palma
Managing Director and Partner
Jan Hinnerk Mohr
Managing Director and Partner
Mousumi Bhat
Vice President of Sustainability
Programs, SEMI
Marijn Vervoorn
Director Sustainability Strategy, ASML
Chris Jones
Environmental Solutions Business
Development Manager, Edwards
Chris Librie
Senior Director ESG, Applied Materials
Jim Larsen
Supply Chain Sustainability Program
Manager, Intel
John Golightly
Corporate Director Global Head
Sustainability, ASM
Bruce Gall and Joe Palazzo
Strategic Partnerships, Google
Kevin Martins
Carbon Strategy & Development,
Microsoft
Young Bae
Global Business Director, Advanced
Cleans Technologies, DuPont
This paper would not have been
possible without the contribution of
BCG colleagues Carlos Licona, Omid
Zandi, Arpita Mahajan, and our BCG
Design Studios team
Authors
