Research and analysis

The wider economic impacts of emerging technologies in the UK (HTML)

Published 9 April 2025

The wider economic impacts of emerging technologies in the UK

A report by PwC for the Government Office for Science

This research was conducted in 2023, under the 2022 to 2024 Sunak Conservative government.

This report has been prepared for the Government Office for Science and solely for the purpose and on the terms agreed with the Government Office for Science and cannot be relied on by anyone else. It does not constitute professional advice, and anyone other than the Government Office for Science should not act upon the information contained in this report without obtaining specific professional advice. No representation or warranty (express or implied) is given as to the accuracy or completeness of the information contained in this document, and, to the extent permitted by law, PricewaterhouseCoopers LLP, its members, employees and agents do not accept or assume any liability, responsibility or duty of care for any consequences of anyone acting, or refraining to act, in reliance on the information contained in this report or for any decision based on it.

© 2024 PricewaterhouseCoopers LLP. All rights reserved. ‘PwC’ refers to the UK member firm and may sometimes refer to the PwC network. Each member firm is a separate legal entity. Please see www.pwc.com/structure for further details.

© Crown copyright 2024

This information is licensed under the Open Government Licence v3.0. To view this licence, visit nationalarchives.gov.uk/doc/open-government-licence/version/3.

Any enquiries regarding this publication should be sent to: contact@go-science.gov.uk.

Executive summary

In May 2023, the Government Office for Science (GO-Science) commissioned PwC UK to undertake an assessment of the impacts of emerging technologies on the UK Economy. The main goal of this study is to quantify the economic impacts of AI and other emerging technologies on the UK economy via productivity gains. More specifically, how the potential productivity gains will impact output within the broader economic landscape of the country until the year 2035.

Understanding the economic impacts of emerging technologies is a key policy area for the UK Government given the large potential impact on the future of the country across each sector and throughout society (PwC, 2024)^{[footnote 1]}. It also aligns with a wider inter-governmental mission to ‘unleash innovation and accelerate science and technology throughout the country’ with a long-term view^{[footnote 2]}. The UK is uniquely positioned to meet this mission via emerging technologies - being home to Europe’s largest capital market, with the value of R&D investment to reach £39.8 billion for 2022 to 2025 (GOV.UK, 2022).

This study builds upon the ‘Emerging Technologies with UK Commercialisation Potential’ report undertaken by the Department for Business, Energy & Industrial Strategy (BEIS) and UK Research and Innovation (UKRI) in 2021. GO-Science has chosen to focus on 15 key technologies for the purpose of this analysis.

These technologies include a wide range of fields and are as follows:

• Artificial Intelligence (AI) / Machine Learning (ML)
• Synthetic Biology/Engineering Biology
• Therapeutics
• Augmented Reality (AR) / Virtual Reality (VR) / Extended Reality (ER)
• Future Telecoms
• Digital Twins
• Advanced Sensing
• Semiconductors
• Future Computing
• Robotics and Autonomous Systems
• Autonomous Vehicles
• Advanced Materials
• Quantum Technology
• Agritech
• Photonics

Approach

The approach taken by PwC to quantify the wider economic impacts of emerging technologies is built around the following 3 broad steps:

1. A business survey conducted to understand the current levels of investment in emerging technologies within the UK, as well as the projected investment trends through 2035. Our approach involved targeting respondents by industry, taking into account the potential for adoption of emerging technologies as well as the industry’s significance in transmitting economic shocks to other sectors. We collected information on a set of key characteristics from a representative sample of UK businesses, including their employment numbers, and investment in emerging technologies. In total, 504 interviews were completed. We targeted executives at general manager and above in UK-based organisations with 15 or more employees.
2. Created scenarios of future adoption based on current technology investments. Forecasting the level of adoption of emerging technologies is a complicated task, influenced by various factors. To consider the uncertainty surrounding technology adoption, we have developed 3 distinct scenarios: a baseline scenario, an optimistic (upside) scenario, and a pessimistic (downside) scenario.
3. Estimated the wider economic impacts associated with the adoption of emerging technologies stemming from increased productivity. To do this, we employed a Computable General Equilibrium (CGE) model to analyse the broader economic effects of emerging technologies on various industry sectors. Our CGE model captures the interlinkages of various sectors in the economy, producing consistent market projections for selected emerging technologies that can be meaningfully compared with each other.

How to interpret the findings

In this study, we seek to quantify the economic impacts of 15 emerging technologies on the UK economy. While these technologies show promise, it’s important to recognise that they may not necessarily represent the most crucial innovations for the future, and there is no guarantee that they will fulfil their potential. There could be other emerging technologies not covered in this report that might attract substantial investments or significantly enhance productivity over the coming years. Investing in these alternative technologies might offer even more advantages.

We seek to quantify the economic impacts via productivity gains. This perspective aligns with conventional economic theory, which views emerging technologies as drivers of economic growth through enhancing productivity. The economic literature indicates that there are other channels through which emerging technologies may contribute to the economy. For example, it has been argued that emerging technologies (such as AI) can impact the economy through product enhancements resulting from AI (PwC, 2017). Since 1 of this study’s prerequisites is to ground all projections in realistic assumptions supported by historical evidence and economic theory, we exclusively examined the productivity channel. Consequently, our projections may appear smaller in comparison to some of the estimates found elsewhere in existing literature.

It is also important to acknowledge that accuracy of our projections hinges on the reliability of the survey responses about business investment and adoption plans for emerging technologies. Although survey data provides valuable insight, it is crucial to be aware of the dynamic nature of investment decisions in the face of evolving circumstances. Economic fluctuations, technological advancements, and changing business landscapes due to environmental and geopolitical disruptions can all lead to significant shifts in firms’ actual future behaviour versus their stated future intentions, and hence our projections.

Additionally, in our analysis, we have primarily leaned on academic literature to quantify the productivity gains from the emerging technologies. Nevertheless, it’s important to acknowledge that these estimates based on historical data carry a degree of uncertainty.

Finally, predicting the economic impacts of emerging technologies involves a significant degree of uncertainty. The extent and speed of their adoption into various sectors depend on a number of factors, including technical feasibility, the cost required for the development and implementation of technologies, labour market dynamics such as the availability and skills of the workforce, as well as the regulatory and societal acceptance landscape. Given these uncertainties, we have constructed 3 distinct scenarios: A baseline scenario in which firms follow through with their investment plans, an optimistic (upside) scenario in which the barriers to adoption are removed and the UK becomes a leader in adoption of these technologies, and a pessimistic (downside) scenarios where the adoption of these technologies experiences delays.

Key findings

• We estimate that businesses will invest approximately £76 billion in 15 emerging technologies from 2023 to 2028. This projection is based on the survey results. However, this projection must be considered within a range, with a potential downside estimate of approximately £60 billion and a potential upside estimate of £87 billion, given the uncertainties inherent in the adoption of emerging technologies. Our survey results also show that the largest future investments could be in AI and Machine Learning, Synthetic/ Engineering Biology, Augmented Reality, Therapeutics, and Robotics and autonomous systems.
• In our baseline projections, we estimate that the increased productivity from the adoption of emerging technologies will contribute approximately 8.4% to real GDP by the year 2035 (Table 1). This statistic represents cumulative real GDP growth relative to the year 2023, which is equivalent to a real GDP increase of £223.4 billion. When comparing this projection to the average annual growth rate of 1.6% in the UK over the 2013-2023 period (ONS, 2023), and assuming a consistent growth rate up to 2035^{[footnote 3]}, our estimates indicate that approximately half of the projected GDP growth in the coming years can be attributed to the adoption of these emerging technologies.
• There is significant uncertainty involved in predicting the uptake of emerging technologies. We project that the enhanced productivity resulting from emerging technologies adoption could potentially contribute 11.9% to real GDP in our upside scenario, equivalent to £317.0 billion in 2023 prices. In the downside scenario, the contribution to real GDP could potentially be as low as 2.65%, equivalent to £70.5 billion.
• The impacts of adopting emerging technologies depend on both the investment magnitude and the sector’s significance in adopting the technology. The economic impacts resulting from the adoption of a technology in our model hinge on the level of investment in that technology. A larger investment leads to higher adoption and greater economic impact. However, the extent of this impact also depends on how effectively the sector adopting the technology propagates the impacts throughout the broader economy. In essence, it is not just about the money invested but also about the sector’s importance and its ability to drive change throughout the broader ecosystem.
• We estimate that a significant portion of this cumulative growth across all technologies is expected to come from 3 emerging technologies. In line with planned investment figures, we expect that the largest economic impacts result from the adoption of Artificial Intelligence / Machine Learning, Synthetic Biology / Engineering Biology technologies and Therapeutics. We project that these 3 technologies will contribute to a 5.50% increase in real GDP by 2035, which is an additional £146.5 billion to the UK’s real GDP.
• The adoption of AI/ML technologies alone is projected to contribute to a 2.98% overall increase in the UK’s real GDP by the year 2035, which is equivalent to £79.3 billion. This growth is due to increased productivity and does not factor in other channels, such as consumption channels through which AI may affect the economy.

Table 1: Real GDP growth by 2035 (%)

Year	2035	2035	2035
Scenario	Upside	Baseline	Downside
Artificial Intelligence (AI)/Machine Learning (ML)	3.88%	2.98%	0.75%
Synthetic Biology/Engineering Biology	2.32%	1.55%	0.63%
Therapeutics	1.45%	0.97%	0.32%
Augmented Reality (AR)/Virtual Reality (VR)/Extended Reality (ER)	0.80%	0.60%	0.22%
Future Telecoms	0.73%	0.55%	0.15%
Digital Twins	0.74%	0.46%	0.14%
Advanced Sensing	0.33%	0.24%	0.09%
Semiconductors	0.46%	0.24%	0.07%
Future Computing	0.28%	0.23%	0.10%
Robotics and Autonomous Systems	0.40%	0.23%	0.06%
Autonomous Vehicles	0.18%	0.11%	0.03%
Advanced Materials	0.09%	0.09%	0.04%
Quantum Technology	0.11%	0.08%	0.03%
Agritech	0.07%	0.04%	0.01%
Photonics	0.05%	0.04%	0.01%
Total	11.39%	8.39%	2.65%

Note: Percentages represent the overall GDP growth relative to 2023 GDP. Source: PwC analysis

Emerging Technologies Taxonomy

Here we provide the list of emerging technologies relevant to this study. The definitions adopted are based on Government Office for Science taxonomies and PwC expert guidance. These definitions consist of descriptions and a set of keywords designed to encompass all subsidiary technologies that collectively define the full technology family.

1. Artificial Intelligence and Machine Learning: The ability of machines to demonstrate intelligence analogous to humans and the ability to perform decisions and tasks that typically require human execution. This includes the use of algorithms to find patterns, learn, and extract knowledge from data; as well as improving execution or inference over time akin to human learning.
2. Extended Reality, Augmented Reality and Virtual Reality: The real time use of visual information integrated with real-world objects and presented using head-mounted displays or projected graphic overlays. A computer-generated 3D environment that surrounds a user and responds to an individual’s actions through those displays.
3. Digital Twins: The virtual representation of a real-world entity such as an asset, person, organisation or process, that serves as an indistinguishable digital counterpart for practical purposes. Digital twins rely on sensor data to transmit information between the physical and digital object.
4. Quantum Technology: Systems that utilise properties of quantum mechanics to acquire, encode, manipulate or process information, run algorithms or sense changes in motion, and electric and magnetic fields through collecting atomic level data.
5. Photonics: The use of light for a variety of purposes. Encompassing the generation, guidance, manipulation, amplification and detection of light.
6. Advanced Sensing: Systems that measure a physical input (temperature, pressure, etc.) with an output signal.
7. Robotics and Autonomous Systems: Machinery and physical systems that can act independently of human control, by sensing, reasoning and adapting to a given situation or environment.
8. Autonomous Vehicles: Vehicles able to operate and perform necessary functions without any human intervention, through the ability to sense the surroundings.
9. Synthetic Biology / Engineering Biology: The application of engineering methods and principles to aid the design of biological systems such as food, chemicals, energy and health etc.
10. Agritech: An agricultural technology that utilises 3 or more defined emerging technologies.
11. Semiconductors: A small electrical component that can act as both a conductor and insulator of electrons; a fundamental component to computing, and various other technologies.
12. Future Telecoms: An umbrella term that encompasses a range of technologies that enable the transportation and sharing of digitised data and information.
13. Future Computing: Technologies that improve a machine’s ability to acquire, encode, manipulate or process information, or run algorithms, such as graphene, DNA, and optical computing
14. Therapeutics: Technologies that aim to treat, alleviate or mitigate disease such as gene and cell therapy, and vaccines.
15. Advanced Materials: The development of materials that possess enhanced or novel properties that enable enhanced performance in application.

1. Introduction

Emerging technologies have significant and complex impacts on the economy. Emerging technologies are those that are in the early stages of development or have only recently been introduced. These technologies are often innovative, presenting new approaches and solutions that were previously not possible. They are characterised by rapid change, high levels of uncertainty, and potential for significant societal and economic impacts. Their integration is the main driving force behind their enhancement of productivity and economic growth.

However, it is also important to recognise that they may also lead to disruptions^{[footnote 4]}. 1 significant concern is job displacement, as technologies potentially replace tasks that humans used to perform. There is a growing concern among economists and politicians that technological progress may take jobs from more and more workers (e.g. Brynjolfsson and McAfee 2011; Acemoglu and Restrepo 2017; Bessen 2017; Autor and Salomons 2017; PwC 2018). This not only affects blue-collar jobs but also white-collar professions, such as data analysis, and customer support, all of which are vulnerable to automation and the utilisation of AI. As such, governments and industries must work together to ensure that the impacts of these technologies are well understood.

This study seeks to contribute to that evidence base by estimating the wider economic impacts of adopting such technologies across different industries in the UK. This study builds on prior work by BEIS and UKRI, which identified emerging technologies along with their commercialisation potential (GOV.UK, 2021). This study builds on that study by measuring the productivity benefits of using these technologies and projecting the wider benefits to the UK economy. Through economic analysis, we aim to measure the extent to which these emerging technologies can make contributions to the UK economy. Our goal is to provide policymakers and stakeholders with a deeper understanding of how these innovations can drive economic growth and development in the UK.

1.1. Research objectives

The primary objective of this study is to assess the wider economic impacts, such as Gross Value Added (GVA) and employment, resulting from the adoption of emerging technologies in various industries. Additionally, this research seeks to develop projections regarding the potential impacts of these emerging technologies in the UK economy, up to the year 2035.

Our overall approach is structured around these objectives, and will complement and build upon previous research by developing an economic model that captures the interlinkages of various sectors in the economy and producing consistent market projections for selected emerging technologies that can be meaningfully compared with each other.

Our study makes 4 distinct contributions, building on prior research in the following 4 ways:

1. We adopt a consistent approach to estimate the impacts of emerging technologies that are otherwise different in nature and stage of maturity on the UK economy, in contrast to prior studies that often focused on specific technologies, used different models, and had varying timeframes. This ensures a more accurate understanding of how emerging technologies shape the UK economy.
2. We ground all projections in realistic assumptions supported by historical evidence and economic theory, setting us apart from previous studies often characterised by overly optimistic or speculative assumptions. This will provide policymakers and other stakeholders with more precise insights into how emerging technologies affect the UK economy, particularly through productivity gains.
3. Our methodology captures the interdependencies between UK sectors, enabling us to measure the impact of changes in 1 sector of the economy on other sectors and the economy as a whole. This allows us to account for general equilibrium effects of technology adoption, even when the direct impact is confined to 1 sector.
4. We recognise the uncertainties surrounding the adoption of emerging technologies and their impact on the economy. The extent and speed of their adoption into various sectors depend on a number of factors, including technical feasibility, the cost required for the development and implementation of technologies, labour market dynamics such as the availability and skills of the workforce, as well as the regulatory and societal acceptance landscape. To address these uncertainties, our study involves a benchmarking analysis to determine upward scenarios that could be achieved by the UK, if it implements similar policies to those of the leading industries in selected OECD countries. Similarly, we construct a downward scenario in which firms adopt technologies at a slower pace than anticipated.

1.2. Structure of the report

The purpose of this report is to set out the research questions, define the scope of the research and describe the approach and methodologies which will be used to assess and quantify both the market for emerging technologies and the impacts of their use.

The rest of this report is structured as follows:

• Section 2 provides an overview of the approach, describing how each of the data sources and modelling elements connect. This section also summarises the goal and expected outcome of each of the modelling elements.
• Section 3 provides a comprehensive literature review that explores the potential economic impacts of emerging technologies.
• Section 4 provides an overview of the business survey and our findings, including their financial performance, employment numbers and investment strategies including current and planned investment in emerging technologies.
• Section 5 describes our methodology required to compute the wider economic impacts realised by the use of emerging technologies across UK industries and presents our empirical findings.
• The Appendix presents a detailed overview of our CGE model, outlines our approach to estimating adoption curves, detailing the method by which new technologies are adopted across the UK economy. We also provide details regarding the optimistic (upside) and pessimistic (downside) scenarios to assess the uncertainties involved in the adoption of emerging technologies.

2. Overview of the Study Approach

This section sets out the high level approach undertaken to deliver this project. The overall approach is underpinned by a specific focus on producing consistent market projections for emerging technologies that allow for meaningful comparisons.

The project was delivered through 2 main workstreams. The first workstream centred on a survey of business executives across different sectors to identify:

• business use of emerging technologies
• the expected impact of those technologies across business performance metrics (particularly productivity)
• information on planned investment in those technologies up to 2035
• factors (internal and external to organisations) enabling and disrupting the adoption of emerging technologies.

The second workstream developed a methodology to estimate the wider economic impacts of these emerging technologies. In particular, we combine the information on planned investment from the survey with the input-output tables published by ONS to estimate wider economic impacts of those technologies.

Table 2.1 explains in more detail the purpose of each of the modelling elements of the overall approach and describes the expected outcome of each. To simplify matters, we have moved the technical aspect of the report to the appendix.

Table 2.1: Overview of the approach

Approach Element	Role in the overall approach (and key assumptions)	Expected Outcome
Business Survey on Emerging Technology Adoption	To understand how companies are currently utilising emerging technology and to understand the levels of investment across sectors. In addition, it will assist us in calibrating S-Curves by determining a UK baseline for leaders and laggards.	A consistent data set to understand the reach and use of emerging technologies in different sectors. These technology use categories are defined under the GO Science taxonomy. Using these inputs we will obtain the level of existing adoption that can be used as an input into our CGE model.
Calibrating Technology Adoption S-Curves	Determine the parameters of the S-Curve, for each industry group. This provides the profile of technology adoption out to 2035 and beyond. We determine the adoption of technologies by leaders compared to laggards, and the hypothetical case of all laggards becoming leaders.	Adoption S-Curves
Developing Upside Scenarios	To construct the upside scenarios, we calculate the plausible maximum adoption potential for the UK by comparing indicators of technological advancement against leading OECD countries, for each industry. This potential is applied to the UK adoption to model the potential upsides given leading countries’ policy choices.	Adoption potential (%) for the UK compared to OECD leaders
Developing Downside Scenarios	To construct the downside scenario we assume that in the downside scenario firms do not invest as reported in our survey in 2023-2028. This reflects a situation where firms will delay future investments. Several factors could contribute to this scenario, including economic uncertainty, changing market dynamics, regulatory changes, or external shocks such as geopolitical tensions.	Adoption S-Curves (%) when future investments are delayed and the market has imperfect competition.
Efficiency S-Curve	Relying on the literature on the impact of emerging technologies on productivity, we combine the business investment from our survey, and the resulting S-Curve of adoption to form the expected productivity gains overtime.	S-Curve of the productivity gains across time to be imputed into the CGE model
CGE Model	We employ a Computable General Equilibrium (CGE) model to analyse the broader economic effects of emerging technologies on various industry sectors. Our CGE model captures a number of complexities of the real-world economy including, but not limited to: household expectations about the economy and its development, government policy, trade flows between sectors both within the UK and abroad (based on historical data).	Wider economic impacts

Source: PwC Analysis

3. Literature Review

Emerging technologies play a crucial role in driving economic growth, primarily through enhancing productivity. Several studies have investigated how emerging technologies can enhance firm productivity through the automation of tasks or generation of new tasks (see Acemoglu and Restrepo, 2019; Agrawal et al. 2019; and Brynjolfsson et al. 2019).

Throughout history, successful technologies have had a significant impact on the economy. Nonetheless, even with these technologies, the economic impacts may unfold gradually or even result in negative outcomes^{[footnote 5]}. However, the prevailing consensus is that the adoption of emerging technologies leads to increased productivity and economic growth.

Among the most successful general-purpose technologies (GPTs) known to date, 3 prominent examples are steam engines, electricity, and information technology (IT)^{[footnote 6]}. Crafts and Harley (1992) found that after the introduction of steam engines, for example, productivity remained sluggish for decades. The introduction of steam engines increased labour productivity by an estimated 0.1% p.a. during 1760–1800, 0.35% p.a. during 1800–1830, and 0.8% p.a. during 1831–1860^{[footnote 7]}. Other general purpose technologies have contributed more to economic growth soon after their introduction. Fiszbein et. al. (2020) found that labour productivity gains from electricity were relatively rapid and sustained. These gains were attributed partially to cheaper energy prices and the quick adaptation of firms in altering their production processes. O’Mahony and Timmer (2009) estimate ICT’s contribution to EU and US labour productivity growth from 1995-2005, at 0.6 and 1.0%, respectively. Crafts (2004b) found that ICT contributed 0.77% per year in Britain during the 1996-2001 period.

Considering the potential impacts of emerging technologies on economies and the labour market, there has been a significant focus on Artificial Intelligence (AI), automation, and robotics. Over the past decades, there have been significant developments in AI and robotics. In the literature, there are at least 3 views on how AI may positively contribute to the economy. Firstly, economists have traditionally regarded emerging technologies as catalysts for economic growth due to their capacity to increase productivity. For instance, AI can also improve capital efficiency through optimising transportation costs and minimising unplanned production downtime (Tantzen, 2015).

Others have argued that emerging technologies (such as AI) can impact the economy through product enhancements resulting from the availability of personalised and/or higher-quality AI-enhanced products and services. In particular, these studies argue that AI might enhance available consumer products through increasing their quality, increasing consumer choice through more personalised and varieties of goods, and saving consumers time from being able to multitask better and delegate to AI technologies. While empirical evidence regarding the relative significance of these channels is limited, some studies suggest that the impacts from the consumption channel might outweigh those of the productivity channel. For instance, a recent PwC report finds that by 2030, AI has the potential to increase the UK’s GDP by as much as 10.3%. However, this growth will primarily be driven by enhancements in product offerings on the consumption side and stimulate demand, contributing 8.4%, and only 1.9% through the productivity channel
(PwC, 2017).

Finally, other studies have speculated about how AI as a new factor of production can propel economic growth, rather than just enhance productivity. Traditionally, economists have regarded labour and capital as the primary factors of production. Some studies have argued that emerging technologies as a new factor of production can lead to significant growth opportunities.

Since 1 of the study’s prerequisites is to ground all projections in realistic assumptions supported by historical evidence and economic theory, we exclusively examined the productivity channel. Consequently, our projections may appear smaller in comparison to some of the estimates found in the existing literature.

In addition to the varying predictions regarding the role of AI adoption, empirical evidence on the impact of AI (as well as other emerging technologies) on productivity is limited. This is primarily due to lack of data on technology adoption at the firm level. Given this limitation, some studies have analysed the impact of AI by examining patent applications and scientific papers related to AI as their primary measures of interest (Cockburn et al., 2019; Van Roy et al., 2020). However, it is important to note that many companies adopt AI technologies developed by other companies. In such instances, relying on patent applications may lead to an underestimation of AI’s impact on productivity.

Other studies use data on specific components of AI technologies and generally find that adopting firms experience higher labour productivity growth following technology adoption. Using firm-level data with isolated automation events, Bessen et. al. (2020) show that automating firms experience faster employment and revenue growth than do non-automating firms. A McKinsey report finds that automation could raise productivity growth globally by 0.8 to 1.4 percent annually (McKinsey & Company, 2017), and Generative AI has the potential to increase labour productivity by 0.1 to 0.6 percent annually until 2040 (McKinsey & Company, 2023).

However, interpreting these results can be challenging due to variations in the definitions of AI employed across different studies. For example, early empirical research examining the potential productivity impacts of AI predominantly relied on data related to robots, as robots have been increasingly using AI technology. However, it is important to note that not all forms of automation involving robots are necessarily AI-driven^{[footnote 8]}.

Additionally, the measurement of adoption varies across studies, with some using a binary metric and others assessing the level of investment per worker. This diversity in measurement methods can lead to varied interpretations of the effectiveness of technology integration in boosting productivity. Lastly, the metrics used to measure productivity also vary significantly among these studies. This inconsistency in measuring productivity – whether it’s output per hour worked, revenue per employee, or other metrics – further complicates the task of drawing clear conclusions about the impact of AI and robotics on labour productivity.

Finally, many studies focus on comparing adopters and non-adopters, which may not indicate a direct causal link between technology adoption and productivity. For instance, firms adopting these technologies already had higher productivity levels or were subject to various external shocks before adopting these technologies. In such cases, the estimated effects may be overestimated.

4. Business Survey on Emerging Technology Adoption

To gather data on current and planned investments in emerging technologies by UK businesses, we carried out a business survey across various sectors. Our approach involved targeting respondents by industry, taking into account the potential for adoption of emerging technologies as well as the industry’s significance in transmitting economic shocks to other sectors.

The rationale behind this approach is to ensure that we do not miss out on valuable insights from companies in industries that play a crucial role in the adoption of particular technologies or dissemination of shocks to other sectors. To illustrate, let’s consider the adoption of Therapeutics. Our analysis indicates that Therapeutics is most likely to be adopted primarily within the Human Health Activities industry. However, it’s worth noting that firms within this sector constitute a relatively small portion of the overall landscape of UK businesses. If we were to employ a random sampling method without this targeted approach, we would run the risk of ending up with a dataset that either lacks responses from this sector entirely or contains an insufficient amount of data to draw meaningful conclusions regarding the current and planned adoption of Therapeutics.

To mitigate these risks, we use information from our desk research on use cases of technologies, and the UK Supply-Use Input Output Tables to inform a more targeted survey sampling scheme. In particular, we take into account the economic centrality of each industry group in UK production networks, as well as the potential for adoption of emerging technologies within each industry group.

The results of this analysis are presented in Table 4.1^{[footnote 9]}. In total, 504 interviews were completed. The survey was undertaken with UK-based organisations using a CATI approach^{[footnote 10]}. Fieldwork was carried out from 30th June to 18th August 2023.

Table 4.1: Survey response count by sector

Industry Group	Allocation
Agriculture, forestry, fishing, mining	41
Energy and water supply	40
Manufacturing	40
Construction	30
Wholesale and retail trade	30
Transport and storage	30
Accommodation and food services	20
Information and communication	30
Financial and insurance activities	30
Real estate activities	20
Professional, scientific and technical activities	40
Administrative and support services	30
Public administration, defence, social security	30
Education	21
Human health and social work activities	41
Other services (please specify)	31
Total	504

Source: PwC analysis.

4.1. Survey overview

Through the survey, we collected information from a representative sample of UK businesses on a defined set of their characteristics, including their financial performance, employment practices, and investment strategies including current and planned investment in emerging technologies. Moreover, we gathered information from respondents on their organisations use of technology and factors that enable or slow down technology adoption.

Note that while our projection timeline extends to 2035, we limit our survey to inquiring about planned investments only up to the next 5 years. The primary rationale behind this approach is that respondents typically possess more accurate and detailed information regarding their planned investments over the next 5 years compared to the subsequent 10-15 years.

Why a survey is relevant to anticipate adoption of technology - and potential limitations

The survey-based method does rely on at least some prospective technology users being able to predict their future behaviour. This does not require that firms have perfect information however – even imperfect information, or constrained estimates, could produce meaningful estimates of likely adoption. This is especially the case given the 5-year horizon – over which most firms have developed forms of investment plans. To ensure that survey respondents are knowledgeable about these investment decisions, we targeted general managers and above. Table 4.2 below provides a detailed breakdown of the respondents by their job grades.

Table 4.2: Survey response count by role within company

Role	Count	Proportion
Chief Executive Officer	10	1.98%
Chief Financial Officer	16	3.17%
Chief Technology Officer	29	5.75%
Chief Information Officer	18	3.57%
Chief Strategy Officer	19	3.77%
Chief Operating Officer	40	7.94%
Finance Director	71	14.09%
IT Director	82	16.27%
General manager	132	26.19%
Other	87	17.26%
Total	504	100%

Source: PwC analysis.

A weighted sampling approach is used to target key industries in terms of adopting technologies and their economic centrality. The survey is supplemented with desk research and expert opinions to minimise potential inaccuracies for certain technologies where they are at the early stages of development. To mitigate the impact of outliers, we implement a 2-step procedure. First, during the data cleaning process, we carefully examined the survey responses to identify potential outliers. Second, we take into account the context of the question and the respondents to accurately and reliably identify outliers and remove them from the sample. We identified 2 outliers from our sample, reducing our total responses to 502.

The survey of businesses was rolled out across diverse sectors and industries to collect data on their investments and uses of a set of 15 emerging technologies and assessments of their business impacts. The questionnaire mixes questions of past and future planned spend in the selected emerging technologies alongside self-reported assessments on the barriers to adoption, expected impacts, the timescale of those impacts, and broad information on emerging technology knowledge.

4.2. Business investments across technologies

From the survey, we were able to gather information on current and planned investments from many large and established organisations. Over 39% of responses came from companies with over £10 million in revenue, with 13% being greater than £100 million. Additionally, 52% of responses came from companies with 10 or more years of operating history.

From the data collected, we can understand the characteristics of firms captured, and scale investments appropriately to represent the UK economy. This enables us to estimate the current levels of investment in emerging technologies within the UK, as well as the planned investments up to 2028 (see Table 4.3).

The levels of investment for these technologies are underpinned and propelled by the recognition of benefits the technologies bring to business operations. Typically across technologies, increasing efficiency or satisfying customer needs is seen as the main business benefit of adopting new technologies, followed by responding to competition. In our survey, 72% of responses indicated that increasing efficiency, responding to competition, or customer needs are key motivators for technology adoption.

Our estimates show that businesses are planning to invest approximately £76 billion in emerging technologies over the next 5 years^{[footnote 11]}. However this projection may vary, with a potential downside estimate of £60 billion and a potential upside estimate of £87 billion, given the uncertainties inherent in the adoption of emerging technologies. Our survey results also show that the largest future investments could be in AI and Machine Learning, Synthetic/ Engineering Biology, Augmented Reality, Therapeutics, and Robotics and autonomous systems.

It is crucial to add a note of caution regarding the investment figures presented in Table 4.3, particularly when addressing technologies with concentrated spending (see Table 4.4). A notable example is Synthetic Biology, where our survey reveals substantial spending concentration within the manufacturing sector with the top 5 adopters contributing more than 95% of the total investment in Synthetic Biology. This concentration poses a potential risk of overestimating the technology’s overall figures. The reason is that our approach assumes other firms will adopt at the same rate as the hypothetical leader firm. However, if the leader firm is unrepresentative or unattainable for most companies, the figures in Table 4.3 may be overstated.

Additionally, large individual historical spenders for certain technologies (Table 4.4), where the investment may be concentrated in a few investors could signal a potential hurdle where current technology maturity only provides returns for large investments or could signal a potential barrier to finance. These issues raise the barrier for entry for new entrants, making it difficult to drive down the cost of this technology without specific policy interventions.

Table 4.3: Investments in adopting emerging technologies, scaled to the UK (£ billion)

Emerging technology	Total annual spend 2023-2028 (£ billion)
Scenario	Upside	Baseline	Downside
AI and Machine Learning	23.55	21.71	17.11
Synthetic/Engineering Biology	16.77	11.74	8.23
AR/VR/ER	8.69	8.25	7.15
Therapeutics	6.99	6.45	5.25
Robotics and Autonomous Systems	6.01	5.49	4.02
Agritech	4.79	4.77	4.05
Future Telecoms	3.77	3.32	2.60
Advanced Sensing	3.39	3.18	2.48
Future Computing	2.94	2.68	2.20
Digital Twins	2.68	2.17	1.49
Semiconductors	2.10	1.75	1.38
Quantum Technology	1.95	1.73	1.41
Autonomous Vehicles	1.57	1.38	1.04
Photonics	1.22	1.18	0.94
Advanced Materials	0.41	0.41	0.31
Total	86.84	76.22	59.67

Note: The table is formed by comparing how much of the economy we have information on through revenues and industry classifications of survey respondents. We rely on ONS UK industry revenue data to scale survey responses to represent the UK. Source: PwC Analysis.

Table 4.4: Proportion of 2022-23 emerging technologies investment from the top 5 spenders (%)

Emerging technology	Proportion of investment from the top 5 biggest spenders
Synthetic/Engineering Biology	>95%
Quantum Technology	81%
Semiconductors	72%
Photonics	64%
Autonomous Vehicles	64%
Future Telecoms	63%
Digital Twins	55%
Advanced Materials	54%
Therapeutics	53%
Future Computing	52%
Advanced Sensing	51%
Agritech	41%
Robotics and Autonomous Systems	36%
AR/VR/ER	30%
AI and Machine Learning	26%

Note: To reduce the risk of disclosure, we have provided lower bounds for Synthetic Biology/ Engineering Biology where few respondents contribute a large share of total investment. Source: PwC Analysis.

4.3. Current business uses of emerging technologies

Amongst the emerging technologies, AI and Machine Learning are expected to see the earliest material impact, with 44% of businesses expecting to use the technology by 2025. Furthermore, increasing efficiency is seen as the main business benefit of adoption. In our survey, 30% of respondents indicated that increasing efficiency would be the main benefits of investment in emerging technologies (BICS, 2023)^{[footnote 12]}. The use of GenAI is helpful to frame the expected benefits especially under the expectation that the parent technology will bring material impact by 2025. The development of the use cases and application of GenAI is rapidly increasing. For example, PwC (Strategy&, 2023) is advising on the opportunities for transformation and value creation that GenAI presents, as well as building and deploying hundreds of GenAI applications. Furthermore, as part of this PwC found that skills and regulatory barriers may be limiting adoption in some sectors, especially highly regulated sectors where the highest value use cases are yet to be identified.

The new product and service development is the area of a business operations which is found to be most impacted by the emerging technologies. 31% of businesses responded that AI and Machine Learning will be game changing to this area of their business (see Table 4.5). This compared with 29% for workforce efficiency and 27% for capital production and operations. This is due to the emerging technologies helping increase pace of development/R&D, monitoring design costs, improving quality and supporting innovation.

From the survey, it is evident that the level of investment in emerging technologies varies, but the underlying reasons for adoption and the barriers are consistent across technologies. We use these benefits to guide our methodology, to better understand how emerging technology adoption is likely to affect production processes, and the likely magnitude of the impacts of technology adoption both for individual firms and for industries more broadly.

Table 4.5: Impact of technologies by 2028 on new product and service development (%).

Emerging technology
Extent of the impact	Game changing	Major	Moderate	Minor	Not sure
AI and Machine Learning	31%	37%	23%	5%	4%
AR/VR/ER	22%	35%	29%	7%	6%
Robotics and Autonomous Systems	20%	42%	23%	8%	6%
Quantum Technology	20%	30%	34%	10%	6%
Advanced Sensing	19%	34%	32%	8%	7%
Autonomous Vehicles	17%	36%	27%	17%	3%
Future Computing	17%	36%	28%	10%	10%
Future Telecoms	14%	32%	32%	14%	7%
Digital Twins	13%	35%	37%	9%	6%
Synthetic/Engineering Biology	13%	40%	38%	6%	2%
Advanced Materials	13%	33%	39%	10%	5%
Semiconductors	12%	32%	36%	12%	8%
Photonics	11%	40%	25%	15%	9%
Agritech	11%	28%	38%	13%	11%
Therapeutics	7%	31%	33%	20%	9%

Note: New product and service development impacts are defined as processes which embed IT in products; increase pace of development/R&D; monitor design costs; improve quality; support innovation. Source: PwC Analysis.

4.4. Enablers and barriers to technology adoption

There are a series of limitations to the adoption of technologies. The primary 2 barriers are found to be the financial cost and workforce skill gaps of adopting new technologies. Table 4.6 below shows the extent to which the limitations listed are expected to be a barrier to the introduction of emerging technologies as part of the businesses technological investment. It’s important to note that barriers do differ by technology. For example, for quantum technology, lack of supplier, and for advanced sensing, regulatory climate is amongst the most relevant barriers for further uptake.

It is also important to understand how significant these barriers are across all emerging technologies. Table 4.7 below shows that in line with Table 4.6, financial costs and skills gaps remain as significant barriers, which are relevant across all technologies. Businesses will have to overcome these barriers to enable the core benefits of uptake.

Table 4.6: Reported barriers restricting greater technology uptake (%)

Emerging technology
Barriers to adoption	The financial cost	Workforce skills gap	The regulatory climate	Inadequate technology infrastructure	Lack of supplier relationship	The access/availability of finance	Uncertainty in potential benefit	Uncertainty in current or projected company performance
Agritech	84%	70%	73%	70%	66%	77%	64%	64%
Advanced Sensing	82%	81%	81%	68%	76%	65%	73%	64%
Quantum Technology	81%	88%	78%	75%	81%	78%	75%	66%
Future Telecoms	80%	77%	70%	73%	67%	63%	65%	64%
Photonics	79%	79%	76%	68%	79%	62%	65%	59%
Synthetic/Engineering Biology	78%	78%	74%	63%	67%	74%	48%	65%
Therapeutics	78%	70%	76%	75%	67%	65%	66%	67%
AR/VR/ER	78%	77%	72%	72%	65%	70%	70%	69%
Digital Twins	78%	75%	78%	75%	64%	69%	61%	64%
AI and Machine Learning	78%	75%	72%	72%	69%	68%	69%	66%
Autonomous Vehicles	77%	77%	74%	71%	65%	68%	61%	52%
Robotics and Autonomous Systems	77%	82%	76%	70%	69%	68%	68%	64%
Advanced Materials	75%	75%	70%	65%	75%	80%	65%	70%
Future Computing	75%	84%	72%	65%	73%	67%	69%	72%
Semiconductors	74%	75%	76%	72%	71%	65%	69%	68%

Note: The values represent a response where there is an indication of a barrier, either a significant or slight barrier. Since each barrier was presented to all respondents, therefore, percentages in the table exceed 100%. Source: PwC Analysis.

Table 4.7: Reported barriers restricting greater technology uptake (%)

Barriers to adoption
Extent of the impact	A significant barrier	A slight barrier	No barrier	Don’t know
The financial cost	33%	42%	22%	4%
Workforce skills gap	25%	46%	24%	5%
The access/availability of finance	24%	37%	35%	4%
Inadequate technology infrastructure	21%	41%	33%	6%
The regulatory climate	17%	47%	32%	4%
Uncertainty in potential benefits	15%	48%	31%	6%
Lack of supplier relationships	14%	46%	36%	4%
Uncertainty in current or projected company performance	13%	45%	37%	5%

Note: Survey question: To what extent are each of the following is a barrier to introducing these emerging technologies to your organisation’s technological investment? Source: PwC Analysis.

5. Wider Economic Impacts of Technology Use

From the survey we have estimates of the level of investment in the next 5 years. We also understood businesses’ intentions towards the uptake of these technologies. Although specific use cases will vary across technologies and industries, the underlying main rationale of investment is apparent: businesses want to boost efficiency and meet customer needs. Whether that be AI enabling greater personalisation and productivity; synthetic biology reducing the cost of producing malaria medication; or augmented reality enabling more effective collaboration. These are just some of the potential uses of technology implicit in the investment figures we have collected from our survey.

To determine the link, we rely on the academic literature^{[footnote 13]} which provides empirical evidence on how capital investment in technologies translate to a productivity impact. We can then apply these potential productivity impacts to a Computable General Equilibrium (CGE) Model. Our CGE model captures the interactions between different sectors of the UK economy, which informs us how the shock can propagate through the UK industries. For example, increases in productivity in manufacturing likely leads to lower input costs for other sectors, which would fundamentally change employment and costs in industries impacted.

Predicting the economic impacts of emerging technologies involves a significant degree of uncertainty. As previously discussed, the extent and speed of their adoption into various sectors depend on a number of factors, including technical feasibility, the cost required for the development and implementation of technologies, labour market dynamics such as the availability and skills of the workforce, as well as the regulatory and societal acceptance landscape. Given the complexity of these considerations, we have constructed 3 distinct scenarios^{[footnote 14]}.

Our central baseline scenario projects that within an industry, competitive pressure ultimately implies that in the long run, adoption patterns across firms within an industry resemble those of the early adopters. We also consider optimistic (upside) and pessimistic (downside) scenarios to assess the uncertainties involved in the adoption of emerging technologies.

Our upside scenario captures the potential gains from the UK becoming a technological leader in each emerging technology. To construct our upside scenario, we identify comparable countries and industries with higher historical adoption rates than the UK. These are identified from data gathered by the OECD, through a survey looking at certain indicators of technological adoption. This comparison gives us a measure of the gap between UK technology adoption and the international frontier. Our upside scenario assumes that this technology adoption gap is closed by 2028 for the emerging technologies in our study. Ultimately, this increases technology adoption and use both in the short and long term. See Section A2.3 for additional information regarding the formulation of upside scenarios.

The industries that see the biggest increase in technology adoption in our upside scenario is Manufacturing. The technologies with the biggest uplift in adoption in our upside scenario include Synthetic Biology, Digital Twins, Therapeutics, Robotics, and Semiconductors. This reflects the potential GDP gains resulting from increasing technology adoption in Manufacturing industries specifically. Note that these are also technologies where we typically see larger real wage and salary increases relative to their GDP impacts.

Our downside scenario captures the risk of limited diffusion of technology within industries, and the potential for planned spending to be slow to materialise. In our baseline scenarios, we assume that in the long run, industry adoption catches up with the plans of technological leaders: either late adopters eventually adopt the emerging technologies, or they become unable to compete and drop out of the market. In our downside scenario, this competitive pressure to adopt emerging technologies is muted, with technological laggards able to continue to compete in the long run. Furthermore, technological leaders’ 5-year spending plans are only met in the long run.

5.1. Overview of results

From our adoption spend curve, we constructed a measure of emerging technology capital investment. For some technologies, including robotics and semiconductors, the technology is embedded in capital equipment. For other technologies, for example AI/ML, the technology is embedded in intangible capital assets including software, training algorithms, and trained models.

We then combine these estimates of the path of emerging technology adoption with empirical literature on the productivity impacts of technology adoption to calibrate dynamic efficiency shocks in a multi-sector CGE model of the UK economy. Conventional economic theory recognises emerging technologies as main drivers of economic growth through augmenting productivity. Existing literature on innovative technology adoption shows that firms who adopt new and innovative technologies on average experience an increase in capital efficiency through several channels. For instance, their capital becomes more productive, either through being able to complete more tasks at lower cost than existing capital equipment, or through the automation of tasks that were previously completed by employees (Acemoglu and Restrepo, 2019).

Existing literature estimates the elasticity of capital efficiency to the stock of assets embedded with new or innovative technologies. We apply Jungmittag and Pesole’s (2019) estimate of the elasticity of capital efficiency to the stock of assets embedded with emerging technologies, in order to estimate the capital efficiency shocks, for each industry and technology, resulting from the adoption of the emerging technologies (see Appendix 2.5). We subsequently incorporate these capital efficiency shocks as inputs into the CGE model to assess the wider economic impacts of adopting emerging technologies on the UK economy.

In our baseline projections, the cumulative impact of adopting all the emerging technologies within our study is projected to result in an 8.39% growth in real GDP by 2035, equivalent to £223.4 billion when measured in 2023 GBP (see Table 5.1). The economic impact will unfold gradually and in a non-linear way (Figure 5.1 and Figure 5.2). This is to be expected, as adoption curves and input shocks which are inputs to the CGE model follow an S-shaped pattern.

We project that the adoption of emerging technologies has the potential to increase UK GDP growth by 11.89% in our upside scenario, amounting to £317.0 billion in 2023 prices. The differences between the upside and baseline scenarios provides an indication of the potential prize for the UK if it were able to become a leader in technology adoption in the OECD. In particular, the gap is largest for technologies where firms’ future technology adoption plans are significantly higher than their current levels of adoption, and where late adopters have much lower adoption rates than technology leaders. In our downside scenario, we anticipate a 2.65% increase in UK GDP growth, equivalent to £70.5 billion.

Across various technologies, there are significant variations in their impact on GDP. The economic impact stemming from the adoption of emerging technologies is intricately tied to 2 fundamental factors: the scale of the investment dedicated to the technology’s integration and the sectors adopting this innovation. The most important driver of the economic impacts of technology adoption in our model is the reported annual current and future spend of survey respondents on adoption and uptake of the technology relative to their revenue. A higher investment level facilitates a smoother transition to the new technology and in turn, implies a larger overall economic outcome.

Moreover, the scale of the economic impact depends on how effectively the adopting sector disseminates the benefits and disruptions associated with the technology. Technologies with higher adoption in production sectors, such as Manufacturing and Construction, have larger economic impacts relative to their projected investment spend. These industries are embedded in many production networks in the UK, purchasing inputs from and selling inputs to other producers in their own and other industries. An increase in their productivity can reduce costs in other industries, create new market opportunities, and create a ripple effect that cascades through the production network. Technologies that are particularly strongly affected by this network effect are Digital Twins and Models, Synthetic Biology, and Semiconductors, and each of these technologies generate relatively large projected economic impacts relative to their projected investment.

In addition to planned spend and the share of adoption concentrated among production industries, 1 further factor that has a significant impact on ultimate GDP impacts within our projection window is the depreciation rate of investments in the specific emerging technology. In our projections, adoption is growing significantly for all of the emerging technologies over our projection window. We use standard depreciation measures to transform these investments into stocks of emerging technology capital, from which we derive efficiency shocks (See Appendix A2.5). For longer lived technologies, specifically Robotics and Autonomous Systems, much of the economic payoff from adoption investments during the next 10 years is expected to occur after the projection window, while other technologies with higher depreciation rates are projected to have higher short-term impacts relative to their projected adoption. For example, given a firm with similar adoption spending plans in Robotics and Autonomous Systems and Advanced Materials, the investment in Advanced Materials is likely to have a bigger impact on GDP within our projection window, while the investment in Robotics and Autonomous Systems is likely to generate a more persistent impact on economic activity.

Our estimates also show that a significant portion of this cumulative growth across all technologies is expected to come from 5 essential technologies. To be precise, 64.34% of this cumulative growth from all technologies is expected to come from Artificial Intelligence (AI) / Machine Learning (ML), Future Telecoms, Quantum technology, Semiconductors, and Synthetic Biology/Engineering Biology (See Table 5.2 and Figure 5.3). The largest economic impacts result from the adoption of Artificial Intelligence (AI) / Machine Learning (ML) which is expected to generate an uplift to UK real GDP of 2.98% by 2035. This grows further when taking into account the adoption of related supporting and supported technologies, for example Semiconductors. Synthetic Biology / Engineering Biology is expected to generate an uplift of 1.55% to UK real GDP by 2035.

Table 5.1: Real GDP growth by 2035 (%)

Year	2035	2035	2035
Scenario	Upside	Baseline	Downside
AI and Machine Learning	3.88%	2.98%	0.75%
Synthetic/Engineering Biology	2.32%	1.55%	0.63%
AR/VR/ER	0.80%	0.60%	0.22%
Future Telecoms	0.73%	0.55%	0.15%
Digital Twins	0.74%	0.46%	0.14%
Advanced Sensing	0.33%	0.24%	0.09%
Semiconductors	0.46%	0.24%	0.07%
Future Computing	0.28%	0.23%	0.10%
Robotics and Autonomous Systems	0.40%	0.23%	0.06%
Autonomous Vehicles	0.18%	0.11%	0.03%
Advanced Materials	0.09%	0.09%	0.04%
Quantum Technology	0.11%	0.08%	0.03%
Agritech	0.07%	0.04%	0.01%
Photonics	0.05%	0.04%	0,01%
Total	11.89%	8.39%	2.65%

Note: Percentages represent the GDP growth relative to 2023 GDP. Source: PwC analysis

Figure 5.1: Real GDP growth from adopting all emerging technologies by 2035 (£ billion)

The above image represents the projected real GDP growth in the UK by 2035, in £ billion, from adopting all 15 emerging technologies across 3 different scenarios, relative to 2023. See data for Figure 5.1.

Notes: Numbers represent the growth relative to 2023 GDP. Source: PwC analysis.

Figure 5.2: Real GDP growth from adopting all emerging technologies by 2035 (%)

The above image represents the projected real GDP growth in the UK by 2035, expressed as a percentage, from adopting all 15 emerging technologies across 3 different scenarios, relative to 2023. See data for Figure 5.2.

Notes: Numbers represent the growth relative to 2023 GDP. Source: PwC analysis.

Table 5.2: Real GDP impacts of the 5 essential technologies 2030 and 2035

	Empirical result (real increase relative to 2023 values)	2030	2035
Scenario		Baseline	Baseline
AI and Machine Learning	Total GDP change (%) from 2023	1.17%	2.98%
	GDP impact from 2023 (£b)	31.1	79.4
Future Telecoms	Total GDP change (%) from 2023	0.25%	0.55%
	GDP impact from 2023 (£b)	6.5	14.6
Quantum Technology	Total GDP change (%) from 2023	0.05%	0.08%
	GDP impact from 2023 (£b)	1.4	2.1
Semiconductors	Total GDP change (%) from 2023	0.10%	0.24%
	GDP impact from 2023 (£b)	2.6	6.4
Synthetic/Engineering Biology	Total GDP change (%) from 2023	1.18%	1.55%
	GDP impact from 2023 (£b)	31.5	41.2

Note: Real GDP values are deflated by the GDP deflator. Source: PwC analysis

Figure 5.3: Combined contribution of the UK’s 5 essential technologies to GDP by 2035 (%)

The above image represents the projected real GDP growth in the UK by 2035, expressed as a percentage, from adopting 5 essential emerging technologies across 3 different scenarios, relative to 2023. See data for Figure 5.3.

Note: Percentages represent the growth relative to 2023 GDP and prices. Source: PwC analysis.

5.2. Interpretation of our results

Our findings should be interpreted as projections of future economic impacts resulting from productivity improvements. Other research has explored additional channels through which technology (particularly AI) can impact the economy. Consequently, our results may appear smaller in comparison to some of the estimates found in the existing literature.

Projections are based on current and near-term commercial use cases. Our firm survey asks respondents to describe their expected adoption patterns for the emerging technologies over the next 5 years. As a result, our results should be interpreted as projections of future growth derived from technology use cases that are commercialisable in the near future. Some technologies, for example Quantum technologies, may indeed have significant potential to increase GDP, through use cases that are not currently commercially viable. For instance, quantum computers can process massive datasets much faster than classical computers, enabling more sophisticated and accurate ML models. This speed-up can be particularly beneficial in fields like drug discovery, climate modelling, and financial modelling, where analysing vast amounts of data is essential. The synergy between quantum computing and AI could lead to more rapid advancements in AI technology, resulting in broader adoption across various sectors due to enhanced performance, reduced computational costs, and the ability to tackle previously unsolvable problems. This potential for future commercialisable use cases is not captured by our results.

Some technologies may be difficult for our survey respondents to define and interpret. Some of the emerging technologies are unfamiliar to many survey respondents, and may closely overlap with related, more well-established technologies. For example, we observed a high rate of current and planned adoption of Digital Twins among Construction industry firms. It is possible that some of this planned adoption may be for related technologies, including Digital Shadows and Digital Model technologies which may be typically classified under the umbrella of Augmented Reality / Virtual Reality.

Some technologies are likely bundled in a particular use case. A given industrial application may require investments across multiple emerging technologies. For example, an investment in a Robotics system for an industrial application may also require an Artificial Intelligence investment. Similarly, an investment in Semiconductors may be required for some Artificial Intelligence applications. In order to gain a comprehensive understanding of the potential for an individual emerging technology for the UK economy, it may be appropriate to sum over the contributions of other emerging technologies that are complementary in typical business use cases. This may also explain some of the higher estimates in other published estimates, which tend to relate to a particular technology and yet may include some of the related technologies.

Some technologies rely on few adopters. While we have removed clear outliers from our survey responses, there remain some firms with significant investment plans in some emerging technologies that differ markedly from peer firms within their industry. This pattern of concentrated adoption of new technologies is not atypical historically, and in some industries, a large innovative firm can contribute a large share of economic activity. Concentrated adoption patterns may be more vulnerable to uncertainty than more diffuse patterns of technology adoption. Our projections for Synthetic Biology in particular are sensitive to the adoption plans of few large firms with international operations. Adoption of these technologies within the UK may be particularly sensitive to individual firm risk as well as country risk and global economic risks.

Technologies change industries. Our model calibration relies largely on historical production processes and patterns of specialisation across industries. With the introduction of disruptive technologies, these historical patterns may not be good predictors of the future. The traditional separation between production and service firms is more blurred, and future patterns of substitutability between capital and labour may differ from the past. We have taken literature on automation technologies as well as survey respondents’ descriptions of their intended technology use in determining how the emerging technologies change production patterns going forward, but we expect that there remains uncertainty about how the emerging technologies will affect patterns of production going forward.

References

Acemoglu, D., Autor, D., Hazell, J. and Restrepo, P., (2022). Artificial intelligence and jobs: evidence from online vacancies. Journal of Labor Economics, 40(S1), pp.S293-S340.

Acemoglu, D., and Restrepo, P. (2019). “Automation and New Tasks: How Technology Displaces and Reinstates Labor.” Journal of Economic Perspectives, 33 (2): 3-30.

Acemoglu, D., and Restrepo, P. (2017) “Robots and jobs: Evidence from US Labor Markets”, Journal of Political Economy 128 (6), University of Chicago Press.

Agrawal, A., Gans, J. and Goldfarb, A., (2019). “The Economics of Artificial Intelligence: An Agenda,” University of Chicago Press.

Agrawal, A., Gans, J. and Goldfarb, A., (2019). “Economic policy for artificial intelligence.” Innovation policy and the economy, 19(1), pp.139-159.

Aguiar, A., Chepeliev, M., Corong, E., & van der Mensbrugghe, D. (2023). The Global Trade Analysis Project (GTAP) Data Base: Version 11. Journal of Global Economic Analysis, 7(2).

Antras, P. and Voth, H.J., (2003). “Factor prices and productivity growth during the British industrial revolution.” Explorations in Economic history, 40(1), pp.52-77.

Autor, D., and Salomons A., (2018). “Is Automation Labor-Displacing? Productivity Growth, Employment, and the Labor Share.” National Bureau of Economic Research.

Bessen, J. (2017). “AI and Jobs: The Role of Demand”, Boston Univ. School of Law, Law and Economics Research Paper No. 17-46.

Bessen, J., Goos, M., Salomons, A., and van den Berge, W., (2020). “FirmLevel Automation: Evidence from the Netherlands.” In AEA Papers and Proceedings, 110:389–93.

Bloom, Nicholas, John Van Reenen, and Heidi Williams. (2019). “A Toolkit of Policies to Promote Innovation.” Journal of Economic Perspectives 33 (3): 163–84.

Brynjolfsson, E. and McAfee, A. (2011). “Race against the machine how the Digital Revolution is accelerating innovation, driving productivity, and irreversibly transforming employment and the economy.” Erik Brynjolfsson and Andrew McAfee © 2012.

Brynjolfsson, E., Rock, D. and Syverson, C., (2018). “Artificial intelligence and the modern productivity paradox: A clash of expectations and statistics.” In The economics of artificial intelligence: An agenda (pp. 23-57). University of Chicago Press.

Chui, M., Hazan, E., Roberts, R., Singla, A., Smaje, K., Sukharevsky, A., … Zemmel, R. (2023). Economic potential of generative AI. McKinsey [online]. Available at: https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-

Crafts, N.F. and Harley, C.K., (1992). “Output growth and the British industrial revolution: a restatement of the Crafts-Harley view.” Economic History Review, pp.703-730.

Crafts, Nicholas. (2004a). “Output growth and the British industrial revolution: a restatement of the Crafts-Harley view.” The Journal of Economic History 64 (2): 521–35.

Crafts, Nicholas. (2004b). “Steam as a General Purpose Technology: A Growth Accounting Perspective.” The Economic Journal 114 (495): 338–51.

Czarnitzki, D., Fernández, G.P. and Rammer, C., (2023). Artificial intelligence and firm-level productivity. Journal of Economic Behavior & Organization, 211, pp.188-205.

Damioli, G., Van Roy, V. and Vertesy, D. (2021) “The impact of artificial intelligence on Labor Productivity”, Eurasian Business Review, 11(1), pp. 1–25.

Fiszbein, M., Lafortune, J., Lewis, E.G. and Tessada, J., (2020). “Powering Up Productivity: The Effects of Electrification on US Manufacturing.” NBER Working Paper, (w28076).

GOV.UK (2022). Government announces plans for largest ever R&D budget. [online] GOV.UK. Available at: https://www.gov.uk/government/news/government-announces-plans-for-largest-ever-rd-budget.

Graetz, G., & Michaels, G. (2018). Robots at work. The Review of Economics and Statistics, 100(5), 753–768.

Johnson, S. and Acemoglu, D., (2023). “Power and Progress: Our Thousand-Year Struggle Over Technology and Prosperity.” Hachette UK.

Jovanovic, Boyan, and Peter L Rousseau. (2005). “General Purpose Technologies.” In the Handbook of Economic Growth, 1:1181–1224. Elsevier.

Jungmittag, A. and Pesole, A. (n.d.). “The impacts of robots on labour productivity.” JRC Working Papers Series on Labour, Education and Technology 2019/08.

Koch, M., Manuylov, I., & Smolka, M. (2019). Robots and firms. SSRN Electronic Journal.

Kromann, L., Malchow-Møller, N., Skaksen, J. R., & Sørensen, A. (2019). Automation and productivity—a cross-country, cross-industry comparison. Industrial and Corporate Change.

van der Mensbrugghe, D. (2018). “The Standard GTAP Model in GAMS, Version 7.” Journal of Global Economic Analysis, 3(1), 1–83.

Oliner, S. D. and Sichel, D. E. (2000), “The resurgence of growth in the late 1990s: is information technology the story ?”, Journal of Economic Perspectives, vol. 14(4), pp. 3-22.

Oliner, S. D. and Sichel, D. E. (2002), “Information technology and productivity: where are we now and where are we going ?”, mimeo, Federal Reserve Board.

O’Neill, A. (2023). United Kingdom - Gross Domestic Product (GDP) Growth Rate 2024. Statista [online]. Available at: https://www.statista.com/statistics/263613/gross-domestic-product-gdp-growth-rate-in-the-united-kingdom/

Rogers, Everett M. (1962). “Diffusion of Innovations.” The Free Press of Glencoe, NY 32: 891–937.

Solow, R., (1987). We’d better watch out. New York Times Book Review, 36.

Technical Appendix

A.1. Survey sampling strategy and industry aggregation

The questionnaire was designed by PwC in conjunction with GO-Science. GO-Science provided the list of technologies to be studied and the associated definitions. These definitions are available in Section 2 of the main report. In total, 504 interviews were completed. We targeted executives at the director level and above in organisations that have a staff of 15 or more.

A.1.1. Sampling strategy

The central question we face when designing the methodology is how to target respondents by industry. Without such targeting we risk losing valuable information from firms in industries that are critical in the transmission of particular technologies. For example, we expect Therapeutics to be primarily adopted in the Human Health Activities industry (SIC code: Q86), however such firms represent a very small share of UK firms. Sampling at random risks a data set with few or no responses in this sector and no information on current and planned adoption of Therapeutics.

To mitigate these risks, we use information from our desk research on use cases of technologies, and the UK Supply-Use Input Output Tables to inform a more targeted survey sampling scheme. In particular, we take into account the economic centrality of each industry group in UK production networks, as well as the potential for adoption of emerging technologies within each industry group.

We characterised Industry Groups by their importance for their transmission of the emerging technologies into wider economic outcomes. Our characterisation learns from network theory, and considers the pathways from technology products to wider economic benefits. We use the following 3 measures to guide this analysis:

1. Bridge industries. A bridge industry is an industry, where, if removed, 1 or more technologies could be disconnected from the wider economy. An example bridge industry is Agriculture, which is critical to the demand for and economic effects of Agritech products.
2. High use case industries. A high use case industry is an industry with high in-degree centrality. These are industries with potential use cases for many technologies. An example high use case industry is Transportation and Storage, for which our literature review has identified potential use cases for at least 8 of the 15 emerging technologies.
3. Economically central industries. An economically central industry is an industry with a high eigenvalue centrality in the Supply and Use Input-Output tables. These are industries that are embedded in many value chains and production networks. Some goods and services will pass through these industries multiple times in production. Economically central industries have large economic multipliers, as technology improvement in these industries has a big impact on economy-wide productivity.

An illustration of the 3 measures of importance is provided in Figure A.1.

The above image represents a diagram illustrating 3 industry importance measures.

Figure A.1: Illustration of industry importance measures

Notes: Technologies are indicated by squares, industries are circles, and economic outcomes are diamonds. Industries are categorised according to their importance for the transmission of a single technology (bridge industry), their importance for the transmission of many technologies (high use case industry), and the likely economic impact of innovations in that industry (economic centrality). Source: PwC Analysis

The results of this analysis are presented in Table A.1. The Manufacturing and Human Health and Social Work Activities industry groups are bridge industries, making them critical for the analysis of at least 1 of our emerging technologies. Agriculture, Forestry, Fishing, and Mining, Energy and Water Supply, Manufacturing, Transportation and Storage, and Human Health and Social Work Activities are identified as industries with use cases across many technologies, and are therefore also particularly important for our analysis. Manufacturing, Wholesale and Retail Trade, Finance and Insurance Activities, and Professional, Scientific and Technical Activities are the most economically central industries; technology adoption in these industries is likely to have the biggest wider economic impacts.

We targeted our sampling toward the identified bridge, high use case, and economically central industries. More precisely, we targeted 40 respondent firms for the industries identified as being particularly critical for any of our measures, and targeted 20 respondent firms for the industries identified as being less critical across all 3 measures. By targeting additional respondents in critical industries, we aim to receive sufficient responses from firms in industries that are particularly important for the transmission of the emerging technologies into economic impacts.

Table A.1: Weighting measures and proposed allocations for business survey

	Normalised importance measures (0-100 scale)
Industry Group	Economic Centrality	Use Case (in-degree centrality)	Use Case Critical (bridge industry)	Allocation
Agriculture, forestry, fishing, mining	14	57	100	40
Energy and water supply	25	79	100	40
Manufacturing	100	100	100	40
Construction	42	23	0	30
Wholesale and retail trade	57	34	0	30
Transport and storage	47	58	0	30
Accommodation and food services	5	9	0	20
Information and communication	29	30	0	30
Financial and insurance activities	56	12	0	30
Real estate activities	10	8	0	20
Professional, scientific and technical activities	51	50	100	40
Administrative and support services	30	12	0	30
Public administration, defence, social security	4	50	0	30
Education	4	8	0	20
Human health and social work activities	1	63	100	40
Other services (please specify)	3	26	100	30
Total				500

Source: PwC analysis, and various PwC studies (e.g. AI, drones and automation…).

Using this method, our survey panel of firm respondents is not directly representative of the UK economy, as we over-sample in industries where differences in adoption patterns will have a bigger impact on our final results. We use ONS input-output data to construct survey weights to scale our industry level results to be consistent with UK business demographics.

A.1.2. Industry aggregation

A difficulty that arises when exclusively relying on our survey industry classifications is that we encounter small counts within certain industries. This leads to a significant uncertainty when constructing our S-Curves which will eventually have a knock-on impact on the outcomes of our broader economic impact assessment. In order to mitigate this uncertainty, we further aggregate survey industries into more high-level industry categories.

To map survey industries into high level industries we rely on and perform the following:

• ONS Standard Industrial Classification: We rely on the industry classifications provided by the Office for National Statistics (ONS) as a foundational reference point in mapping survey industries into high-level counterparts.
• Alignment with GTAP Sectors: There exists a misalignment between the industry classifications employed in our survey data and those in the Global Trade Analysis Project (GTAP) dataset. We aggregate industries so that they are aligned with the GTAP sectors at the high-level classification level.

Table A.2 below shows these high-level sectors.

Table A.2: High Level Industries

High-level industries	Agriculture, forestry, fishing, mining	Energy and water supply	Manufacturing	Construction	Trade, transport, accommodation and entertainment	ICT, finance, professional and administrative	Real estate activities	Public services
Survey industries	Agriculture, forestry, fishing, mining	Energy and water supply	Manufacturing	Construction	Wholesale, retail and repair of motor vehicles	Information and communication	Real estate activities	(Admin, defence, education, and health
					Transport and storage	Financial and insurance activities		Public administration, defence, social security
					Accommodation and food services	Professional, scientific and technical activities		Education
						Administrative and support services	-	Human health and social work activities

Source: PwC analysis.

This approach acknowledges the inherent limitations of precision when dealing with certain industries, ensuring that our estimates remain both robust and interpretable. The implication, however, is that our analysis and results will be reported at this high-level of industry aggregation. For instance, instead of reporting GVA for specific industries such as education, we will present GVA for broader high-level industries like public services which includes public administration, defence, social security, education, and human health and social work activities.

A.2. Future technology adoption

From the survey, we obtain information on the current and planned investment on emerging technologies. This provides insight into current investment levels and a projection of where we will be in 5 years, assuming firms follow through with their investment plans. However, this information doesn’t fully specify the adoption path over time. The speed of this process is mainly driven by the diffusion of technology, which refers to the process by which a new technology is adopted and spreads throughout an economy.

A.2.1. Building technology adoption curves: baseline scenario

1 of the earliest and most influential models of technology diffusion is the “S-shaped Curve” proposed by Rogers (1962). He argued that the adoption of new technology follows an S-shaped Curve, with a small group of innovators initially adopting the technology, followed by early adopters, the early majority, the late majority, and finally, the laggards. The adoption of emerging technologies and their impact on labour productivity is influenced by a variety of factors, including the type of technology, the level of investment in technology, and the skills and training of workers. The literature points out that technology adoption and diffusion can be facilitated by factors such as strong intellectual property rights that encourage firms to invest in R&D, the presence of skilled workers who can absorb and implement new technologies, and the availability of funding and infrastructure to support technology development and diffusion (Bloom, Van Reenen, and Williams, 2019)^{[footnote 15]}. In addition, emerging technologies that are complementary to existing jobs, such as collaborative robots (or cobots), are more likely to be adopted as they can lead to significant increases in productivity when integrated into existing work processes. The level of investment in technology is also a critical factor in determining the impact of emerging technologies on labour productivity. Finally, the skills and training of workers are also crucial in determining the impact of emerging technologies on labour productivity^{[footnote 16]}. In our survey, over 70% of organisations cite financial costs and workforce skills gaps as a barrier to adoption.

Following this literature, we assume that the adoption follows an S-shaped Curve. More precisely, we specify a Logistic cumulative distribution function (CDF) with the functional form:

The above diagram represents a mathematical equation for adoption following an S-Curve.

where A is the maximum eventual adoption of the technology; k is the logistic growth rate which controls the shape of the curve, for larger values adoption starts later but is more rapid when it occurs; t_m is the year in which adoption reaches half its eventual maximum A; and t is the year. Our adoption curve (A.1) has 3 unknown parameters, A, k, and t_m. Thus, we need 3 equations/observations in order to identify the 3 parameters. From our survey results, we observe firms’ spend on adoption in 2023, and firms’ planned spend on adoption over the next 5 years. Therefore, we require 1 further observation in order to identify all 3 parameters. To do so, we propose to estimate the maximum eventual adoption of the technology A, using the following approach which we refer to as the leaders-laggards approach.

A.2.2. Identifying leaders and laggards

The approach seeks to use differences in survey responses to identify A, the maximum eventual adoption of the technology. Our approach is closely connected to the diffusion of innovations theory developed by Rogers (1962). The theory proposes a distinction between early adopters of new technologies (i.e., leaders), and late adopters (i.e., laggards). Ultimately, technology diffuses to all firms in the industry, this can be via the eventual adoption by the laggards, increased market share of the leaders, or by the entry of new firms.

In order to identify leaders and laggards in terms of technology adoption, we use data on a firm’s total technology spend (as a share of revenue), and their total current and planned spend on the emerging technologies. We construct a probabilistic model that places a higher probability of leadership for firms who exhibit a high total technology spend relative to their industry peers, a high total emerging technology spend relative to their industry peers, and a high emerging technology spend on a specific technology relative to all other firms in the survey. This approach ensures that we can identify potential technology leaders for whom the emerging technologies in the study might not be appropriate for their business, but also ensures that where any firm is particularly important for the adoption of a given emerging technology, they are identified as a leader. The approach also allows us to identify leaders within industries that might have low average technology spend.

Using this leaders-laggards approach, we identify the leader firms in each industry. We then set the maximum value of the adoption curve to be such that all firms in the industry adopt the technology to the level of the planned adoption by the leading firms:

The above image represents a mathematical equation for maximum value of adoption.

Equations (A.1) and (A.2) and 2 observations on firms’ spend (i.e., adoption in 2023, and planned spend on adoption over the next 5 years) are sufficient to identify the unknown parameters.

Illustrative Example:

We present here an illustrative example demonstrating how we construct S-curves of adoption spending and accumulated emerging technology from our survey responses. For this example, we assume that we start with 1 leader firm and 1 laggard firm in an industry.

Step 1: Maximum eventual adoption (A)

To calculate the maximum adoption rate, and given a classification of firms into leaders and laggards, we need to the following inputs from our survey:

• Current spend in 2023 and 2028 on the adoption of a specific technology.

From this we identify the leader firms and their adoption rate (spend per unit revenue). Next we can calculate with the formula below,

The above image represents a mathematical equation demonstrating how to derive the maximum value of adoption in a working example.

Step 2: Growth rate (k) and midpoint (t_m)

To calculate t_m, and k we need to the following additional inputs from our survey:

• Current spend in 2023 on the technology
• Planned spend in 2028 on the technology

Then we can simply substitute these into the equation to solve for t_m and k. The steps to derive the formulas below involve equating the logistic S-Curve function in 2023 and 2028 through k, and solving for t_m and then k. The solutions are as follows:

The above image represents a mathematical equation demonstrating how to derive the midpoint and growth rate of adoption in a working example.

Illustrative example:

Item	Source	Data	Leaders	Laggards
A	Survey	Spend in 2023 (£m)	100	50
B	Survey	Spend in 2028 (£m)	250	200
C	Survey	Revenue (£m)	5000	10,000
			Industry
D	[Total A]/[Total C]	a(2023) (%)	1%
E	[Total B]/[Total C]	a(2028) (%)	3%
F	[B Leaders]/[C Leaders]	A (%)	5%

Note: Our baseline scenarios in our full model also include a small market share adjustment, where leaders are assumed to increase market share between 2023 and 2028. This step is not included in our illustrative example here.

Using the formulas above for t_m, and k, we need to calculate items D, and E from the table above. Given the figures in the table, we compute t_m = 2026.9, and k = 0.36.

A.2.3. Formulating upside scenarios

We will construct upside scenarios in 3 steps:

1. Determine comparable countries and indicators.
2. Calculate the adoption potential (%) for each industry.
3. Construct potential maximum scenarios.

Step 1: Determine comparable countries and indicators

The primary goal of this task is to understand where the UK sits relative to comparable OECD leaders in technology adoption, providing a feasible way to determine our potential maximum point of technology adoption. This potential maximum is used in our modelling to consider where the UK adopted similar policies to comparable leaders in similar sectors across OECD countries. To do so, we rely on indicators of technological advancement from the OECD to determine for each industry where the UK is on the technological frontier, and how far the UK is to the leading OECD country. More specifically, we rely on OECD statistics on ICT access and use by businesses across sectors, which contains 51 indicators, based on the 2nd revision of the survey on ICT access and usage by businesses

The next step is to decide the countries/industries which will serve as the benchmark. For our analysis to be grounded on real-world observations, it is always important to ensure that the benchmarking countries must be similar to the UK. Some countries have more favourable conditions for adoption of a specific technology, compared to the UK. An example of this is that Iceland is geographically very different to the UK, and benefits from different emerging technologies, therefore, Icelandic policies related to tech adoption will be inherently different to the UK, and thus not comparable.

Therefore, we have applied the following filtering to determine the most pertinent and comparable indicators of technological advancements, given modelling needs and data availability, to exclude the below:

• Countries with a total population of fewer than 10 million people^{[footnote 17]}. As smaller countries have distinct economic structures from the UK, with their economy mostly centred in a single city and limited number of industries, we exclude them from our benchmarking exercise.
• Indicators that have a value of less than 25% adoption for the UK. We consider this reasonable, as low adoption rates imply that the UK is earlier on the S-Curve, therefore inclusion of these indicators could create scenarios which don’t reflect potential adoption, rather another point on the adoption curve itself.
• Indicators which do not cover at least 9 sectors^{[footnote 18]}. To calculate a comparable set of adoption potentials, we should rely on a consistent set of indicators across industries. Therefore, any indicators which cannot adequately cover the majority of sectors can be dropped for comparability and relevance (e.g. businesses with formal policy to manage ICT privacy risks due to lack of data, and businesses using social media due to lack of coverage across sectors).

Given the above criteria, we have identified a set of indicators to proxy a country’s’ technological adoption for OECD sectors^{[footnote 19]}. Example technology adoption and use indicators include:

• Persons employed regularly using a computer in their work (%)
• Businesses with a broadband connection -includes both fixed and mobile (%)
• Businesses using ERP (Enterprise Resource Planning) software (%)
• Businesses using CRM (Customer Relationship Management) software (%)
• Businesses which employ ICT specialists, within the last 12 months (%)

Step 2: Calculate the adoption potential (%) for each industry

To compile an adoption potential (%) for each industry, we first calculate a simple percentage difference between the UK and the leading adoption rate, for every indicator. We then assign each indicator to labour or capital & others, depending on whether the indicator reflects labour related technology adoption, or capital & others. Next, we calculate a simple average of labour and capital & other adoption potentials respectively. This is then used to compute a final adoption potential by weighting each simple average by the labour to capital ratio from ONS IO tables.

We calculate the labour and capital ratio by looking at the ratio of the compensation of employees and the gross operating surplus and mixed incomes, for the different industry groups. This is important as different industries, use labour and/ or capital at different intensities, and since these technology adoption indicators cover both aspects^{[footnote 20]}, we must control for this ratio. The alternative would be applying a simple average across all indicators for each industry and assume that each indicator is equally important to that industry and all other industries. This is clearly a strong assumption, which can fail at the extremes - for example, in manufacturing we would expect higher reliance on capital for production, therefore, indicators of labour adoption of technologies (e.g. percentage of businesses employing a ICT specialist within the last 12 months) will not be a good indication of the general adoption in that industry.

This final adoption potential (%), reflects for each industry, in the UK, the potential maximum adoption of technologies, based on a set of comparable countries which are leading in the indicators associated with the industry.

Step 3: Apply the adoption potential to our adoption S-Curve

The adoption potential (%), is applied to the 2028 investment / spend plan indicated by business users in the survey. This is reflected in the S-Curve of adoption through a larger total investment/ spend in 2028 for all respondents (see Figure A.2). The policy interpretation would be as follows:

• Changing policies in the UK today, to deliver the same adoption rates as leading OECD countries which materialise in 2028.
• Given assumptions and filtering to ensure that the OECD countries are comparable to the UK, means that policies used in leading countries could be feasibly implemented in the UK.
• Therefore, it would be plausible to consider adoption potential as a feasible maximum.

Furthermore, the adoption S-Curve is an input to the efficiency shocks, which are inputs to the CGE model to measure the wider economic impacts. Therefore, since efficiency shocks are derived from our adoption of S-Curves, the plausible maximum adoption will also inform the plausible maximum efficiency shock.

Figure A.2: Adoption S-Curve illustration of the upside scenario compared to baseline

The above image represents a line chart depicting the illustrative scenario only for the upside of an S-Curve tracking the rate of uptake of a technology.

Source: PwC Analysis.

A.2.4. Formulating downside scenario

To construct downside scenarios, we follow the following steps:

1. We assume the same rate of adoption and growth as the baseline in 2023;
2. Assume firms do not meet their planned spend between 2023 - 2028, but it will be achieved beyond 2028 as the market’s eventual adoption rate.

Step 1: We assume the same rate of adoption and growth as the baseline in 2023.

In this step, we assume that the downside scenario in 2023 mirrors the baseline, in terms of the adoption rate (%) of firms with a similar rate of uptake (%) in the near term (2023 - 2028) (See Figure A.3). To achieve this, we initially calculate the adoption rate (%) under the baseline scenario, using it as the starting point for the downside scenario in 2023. This calculation helps establish a growth rate in the near term that aligns with the baseline, taking into account the growth necessary to reach the eventual downside adoption rate (%), which is consistently lower than the baseline (as explained in step 2 below).

The step above shapes the narrative for the downside scenario, representing a deviation from the baseline, signifying a delay in the 2023 - 2028 spending from our survey, due to various reasons, including: technologies not maturing as expected (resulting in use cases not materialising), and/ or firms being overly optimistic about their financial prospects in the next 5 years.

Step 2: Assume firms do not meet their planned spend between 2023 - 2028, but it will be achieved beyond 2028 as the market’s eventual adoption rate.

We assume that firms won’t meet their planned spending between 2023 and 2028; instead, they will achieve these levels of spending beyond 2028, which informs the market’s eventual adoption rate. This delay is modelled by deferring firms’ investment decisions related to emerging technology and their impact until after 2028. In the modelling, this delay is reflected in the assumed maximum eventual adoption rate (%).

To calculate the eventual adoption rate for the downside scenario, we consider the impact of technology spending in relation to market share beyond 2028. We assume that the rate of adoption beyond 2028 corresponds to the 2028 level of adoption indicated in our survey. Differences in adoption rates between market leaders and laggards lead to shifts in market shares (measured by revenues) between the 2 groups. These shifts are driven by relative productivity and cost advantages for the group with greater adoption, calculated through our estimated betas.

The interaction described above informs the market shares of leaders and laggards beyond 2028. Combining this information with the adoption rates from both groups (using the 2028 data point) allows us to calculate the revenue-weighted average adoption rate for the market beyond 2028. This represents the downside scenario’s eventual maximum adoption rate.

Notably, since the market rate of adoption is the revenue-weighted adoption rate for leaders and laggards, the downside scenario will always have a maximum eventual adoption rate lower than the baseline approach. This is because, in the baseline, the eventual rate of adoption of all firms corresponds to the rate of leaders, effectively representing a market of leaders only.

Crucially, the downside scenario makes the following assumptions:

• Delay in spending: Firms will not invest in 2028, as reported in the survey. Instead, they will invest at this level beyond 2028.
• Unit elasticity^{[footnote 21]}: To establish predictable market change dynamics, we assume that customers’ relative demand for products produced by leaders and laggards is unit elastic. This means changes in cost savings translate 1-to-1 with changes in market share (e.g. a 3% increase in cost savings for leaders results in a 3% increase in revenues).
• There is imperfect competition: The market is not perfectly competitive, allowing laggards to remain competitive with leaders beyond 2028, even without similar levels of emerging technology adoption.

Figure A.3: Adoption S-Curve illustration of the downside scenario compared to baseline

The above image represents a line chart depicting the illustrative scenario only for the downside of an S-Curve tracking the rate of uptake of a technology.

Source: PwC Analysis.

A.2.5. Constructing efficiency curves

Jungmittag and Pesole (2019) estimate the effect of the use of robot technologies on productivity for manufacturing businesses. They compute a robot-intensity index, which equates to the share of robot assets within a firm’s broader capital assets. They estimate an elasticity of multifactor productivity with respect to their robot index of 0.6; a 10% increase in the capital share of robots generates a 6 percent increase in multifactor productivity. We use this estimate as the starting point for our mapping from technology adoption to efficiency shocks in our model.

The starting point elasticity of 0.6 is not appropriate for all technologies / industries. Jungmittag and Pesole’s (2019) estimates are based on the stock of robot capital, rather than the flow of robot adoption spend. In order to construct comparable stock variables for our study, we build emerging technology capital stock S-curves from our adoption spend curves described above, using technology-specific depreciation rates. For example, we apply a 10% per year depreciation rate for Robotics and Autonomous Systems.

By using a common starting point elasticity, rather than identifying separate elasticities for different industry/technology pairs, we can limit the influence of different estimation methodologies contributing to different estimates by technology or industry. The key assumption underlying our approach to constructing technology elasticities is that firms investing in 2 different technologies are expecting a similar return on those investments, after taking depreciation into account.

We then impose 2 adjustments based on the technology and the industry to construct elasticities mapping technology adoption to capital efficiency shocks.

Industry elasticity adjustment. We apply shocks to capital efficiency, rather than multifactor productivity. We adjust our elasticities to maintain a consistent impact on multi-factor elasticity. Ultimately, this means that for 2 firms with similar adoption as a share of revenue, we apply a larger capital efficiency shock to the firm with smaller capital share in their production. This ensures the 2 firms have similar overall productivity impacts from similar sized investments, and reduces the potential impact of misallocating a firm into an incorrect industry.

Technology elasticity adjustment. We impose an adjustment to the efficiency elasticity of technology adoption that depends on typical depreciation rates of investments in each technology. The Jungmittag and Pesole (2019) estimates relate specifically to robots, which have the lowest typical depreciation rate of all of the emerging technologies in our study. Firms investing in robot technologies expect a return on that investment over a longer time period than investments in other technologies, and our technology capital accumulation curves build in this longer useful life. We adjust the technology elasticities upwards for technologies with higher depreciation rates to ensure that an individual firm investing in multiple technologies would expect those investments to earn a similar return.

A.3. Overview of CGE model

A CGE model captures the interactions between different economic agents in the economy as a whole. This means it is able to track the potential impact of emerging technologies because it represents the interlinkages between different sectors of the UK economy. It has multiple agents (e.g. firms, households and government) which interact in markets for capital, labour, and goods and services. The model finds an equilibrium in the economy where supply and demand are balanced for all goods and services. This is achieved by solving a system of equations that represent the behaviour of households, firms, and the government.

CGE models are powerful tools to assess the likely effects of policy changes and shocks, and are widely used by international institutions such as the World Bank, IMF, OECD, and European Commission and the UK Government. The benefits of these models stems from their ability to estimate the impact of ‘shocks’ which may have significant general equilibrium effects, even when the direct impact is confined to 1 sector, affecting consumption choices and firms’ output and employment decisions over time. CGE models are designed to capture the feedback effects of changes in 1 sector of the economy on other sectors and the economy as a whole.

To facilitate this policy analysis, a CGE model captures the economic behaviour of all agents in the economy through a system of equations. Figure A.4 illustrates these economic interactions between agents. To determine the general equilibrium in the economy, a system of simultaneous equations should be solved to obtain a set of prices and an allocation of commodities and production factors that support the equilibrium. When the economy is in equilibrium, the economy must satisfy the following conditions:

• All markets in the economy must clear i.e. the demand and supply of all commodities and production factors are balanced; and
• Income in the economy must balance i.e. all economic agents must exhaust their budgets.

Figure A.4: Primary interrelationships within the CGE model

The above image represents a diagram showing the primary interrelationships within the computable general equilibrium model.

Source: PwC Analysis.

To simulate the impact of policy or technology changes, particular elements of the model are impacted or ‘shocked’, which leads to a new general equilibrium in the economy. For instance, if the efficiency of a sector in the economy is adjusted, then economic agents adjust to these price changes by reallocating consumption and production decisions until equilibrium in the economy is restored. The economic impact of the policy or economic shock is then estimated by comparing the baseline output variables to the output variables under the new equilibrium.

A.3.1. Model features

In this Section, we explain the key features of our CGE model and the underlying assumptions behind the model.

Firms in our CGE model are producers of goods and services. They use factors of production such as labour, capital, and land to produce goods and services, which they sell to households and the government. Firms are assumed to operate in different industries and make investment decisions that consider both current output and their expectations about the future.

Households are represented as economic agents who own factors of production such as labour and capital. They supply these productive factors to firms in exchange for income, which they use to purchase goods and services. Households are assumed to have preferences for various goods and services, and they allocate their income among different types of consumption and saving. The model assumes a representative household in the economy which maximises an intra-period utility function across all time periods, subject to a budget constraint.

The government collects all the taxes in the economy such as income taxes, taxes on production such as foreign worker levies and property taxes, and taxes on products such as customs and excise duties and the Value Added Tax (VAT). The government’s expenditure includes consumption of goods and services and transfers to households. All welfare payments accrue to households.

The model is dynamic and tracks the evolution of the economy over time as it responds to shocks. This dynamic approach has the distinct advantage that it captures the intertemporal aspect of agents’ decision-making. Table A.3 summarises the primary model features.

Table A.3: Summary of model features

Model features	Description
Household	A representative household is assumed to be rational and maximises its utility in each period
Firms	The model captures the activities of firms in a variable number of aggregated industry groups
Dynamic	The model tracks the evolution of economic variables over time
Labour market	The workforce is disaggregated into 2 types of workers, characterised by occupation characteristics. Workers of both types can change jobs, shifting from lower pay industries to higher pay industries.
Imperfect competition	Within each industry, firms offer differentiated products, and by limiting their supply they can charge a higher price than their marginal production costs
Capital adjustment costs	Firms are assumed to incur costs when planning and installing new capital in response to changes in the economy
Government closure rules	The government raises revenue largely through taxation, and funds public services.
Rest of the world	The UK economy trades goods and services with the rest of the world, subject to physical trade and transportation costs in addition to tariff and non-tariff trade barriers.

Source: PwC Analysis

A.3.2. Model interactions

Figure A.5 shows the economic interactions between economic agents in more detail, displaying the income flows in the economy. All income in the economy must balance so that all economic agents exhaust their budgets and all markets in the economy clear i.e. the demand and supply of all commodities and production factors are balanced. These income flows can be considered in terms of the consumption block and consumption block.

The production block shows the allocation of output between factors of production and intermediate inputs. The creation of a product required intermediate inputs, of which typically some are sourced domestically and some are imported. The remainder of output consists of returns to capital and labour, which produce factor incomes which then feed into consumption. Households, firms and governments spend their income on a mixture of domestic and imported goods with the remainder used as savings. Savings are then used by firms to replenish depleted capital or invest in new capital for the production of goods and services to meet domestic and foreign demand. Residual income from firms are retained as profits which accrue to the providers of capital.

Figure A.5: Circular flow of income in the global economy

The above image represents a diagram showing the circular flow of income in the global economy between economic activities.

Source: PwC Analysis.

Linkages between economic activities, such as firms’ decisions on the relative quantities to sell to domestic and foreign markets, are represented by elasticities which are parameters that attempt to capture the behavioural responses of economic agents. We discuss these parameters later in the report. To illustrate how the CGE model has been set up, we now describe in greater detail key economic activities and linkages. Our discussion is segmented into 2 main components – the production block and consumption block.

A.3.3. Production block

The level of production for each sector in the economy is determined by a nested production structure as shown in Figure A.6. In the first stage, firms choose the amounts to supply to the domestic market and overseas market. The model assumes there are costs to switching between these markets, in light of particular costs involved in export activities, such as the costs of establishing and maintaining a client base overseas or producing goods that satisfy foreign tastes. The CGE model takes these costs into account through a Constant Elasticity of Transformation (CET) function. Output is initially exported or consumed in some baseline proportion in the CGE model. These proportions then adjust according to changes in relative export and domestic prices and the elasticity parameter in the CET function governs the rates at which these proportions change by. For example, if elasticity equals 3, a rise in the relative price of exports by 1% causes firms to increase the relative quantity of exports by 3%.

Figure A.6: Nested structure of the production block

The above image represents a diagram showing the nested structure of the production block in the computable general equilibrium model.

Source: PwC Analysis.

In this CGE model, the firm produces multiple outputs, an export good and a domestic good. This equation captures transformation between these outputs. The functional form of the CET function is as follows:

The above image represents the functional form of the Constant Elasticity of Transformation function.

where:

Variable	Description
Y	Total output
E	Efficiency
α	Share of output consumed domestically
D	Output consumed domestically
X	Output exported
η	Elasticity of substitution between different inputs

In the second stage, firms combine intermediate input goods M with their own capital and labour factors of production, which contribute to the value added component of production Y_va:

The above image represents the functional form describing how firms produce the final goods by combining intermediate goods with their own capital and labour.

where:

Variable	Description
Yp	Total output
A	Output efficiency
Yva	Value added component of production
M	Intermediate inputs
αva	Share parameter
γ	Elasticity of substitution between different inputs

Output efficiency increases the total amount of production, holding productive factors and intermediate inputs constant. The value added component of production combines labour and capital employed by firms in the industry. Note that in GTAP11, the elasticity of substitution between value added and intermediate inputs is set to 1, γ = 1, for all industries:

The above image represents the functional form describing how firms produce the final goods by combining capital and labour.

where capital and labour inputs are scaled by factor input efficiency parameters A_k, A_L respectively. An increase in A_L means that the industry can reduce their labour employed L and maintain the same level of output. The capital-labour substitution elasticity is denoted by ϵ. When the elasticity is low, an increase in labour efficiency A_L will likely decrease the labour demanded by the industry. This would be an example of firms using technology to replace workers. When the elasticity is high, an increase in labour efficiency A_L will likely increase the labour demanded by the industry. This would be an example where as labour becomes more productive, firms employ more workers. The elasticities vary by industry, and are calibrated to the GTAP11 database.

The aggregator M bundles together intermediate production input goods, purchased from other industries as well as from other firms within the same industry:

The above image represents the functional form aggregating intermediate production input goods, purchased from other industries as well as from other firms within the same industry.

where A_mi denotes the intermediate input efficiency for an intermediate input purchased from industry .

A.3.4. Productivity channels

Adoption of a new technology increases the efficiency shifters A, A_K. The specific efficiency parameter depends on the application area of the technology, as reported by firms in our survey. The application areas relate to specific uses of technologies, for example to reduce wastage, or to improve products, or to train staff, or to better utilise capital. Specifically, we model the effects of emerging technology adoption primarily through the capital efficiency shifter A_K, aside from in the Construction Sector, where we model the effects of technology adoption through the multifactor efficiency shifter A.

The emerging technologies studied in this report primarily take the form of new capital equipment, and intangible automation-focused technologies. Automation technologies, including AI, can be used to shift tasks from labour to capital (Acemoglu and Restrepo, 2019). For both types of technologies, it is generally appropriate to model the effects of their adoption through the resulting increase in capital efficiency. We make an exception for the Construction sector, where the adoption of the emerging technologies is modelled as a shock to multifactor efficiency A. This modelling decision takes into account the use cases of the emerging technologies in Construction, as well as differences between the emerging technologies and existing widely used capital equipment in Construction.

A.3.5. Consumption block

Demand from households and the government drives consumption in the economy, with the aggregate consumption determined by a nested consumption structure, as shown in Figure A.7. The model assumes a representative household in the economy which maximises her utility subject to capital and labour income and welfare transfers from the government. In the model, the representative household and government consume both domestic and imported goods.

A CES function with different Armington elasticities governs the substitution possibilities across the range of commodities. Such preferences mirror real-world trading patterns where countries simultaneously import and export goods in the same product category. In particular, household spending on domestic goods and services is governed by a Stone-Geary (SG) utility function. The functional form of the SG utility function requires households to have a minimum level of consumption. Over the medium term horizon of the model, this minimum level of consumption reflects the difficulty of adjusting consumption spending on essential goods and services, which can include housing, food expenditure and energy expenditure.

Figure A.7: Nested consumption structure of the domestic economy

The above image represents a diagram showing the nested structure of the consumption block in the computable general equilibrium model.

Source: PwC Analysis.

The model assumes a representative household in the economy which maximise an intra-period utility function across all time periods utility

The above image represents the functional form describing household in the economy maximise an intra-period utility function across all time periods.

subject to a budget constraint. Households’ utility is a function of leisure and aggregate consumption given in a CES function:

The above image represents the functional form describing Households’ utility is a Constant Elasticity of Substitution function of leisure and aggregate consumption.

where:

Variable	Description
Ct	Utility deriving from consumption in year t
Lt	Leisure in year t
U(.)	Utility function of the household
ρ	Discount rate parameter
σ	Substitution parameter
1 / 1 − σ	Elasticity of substitution between leisure (L) and aggregate consumption utility (C)
α	Share of leisure in household utility

The flow of utility deriving from consumption is then represented as a Stone-Geary function given as:

The above image represents the functional form of the Stone-Geary function.

where:

Variable	Description
C	Utility deriving from consumption for a household in a certain period
qi	Level of consumption for good/service i
γi	Subsistence level of each good/service i
βi	Share of discretionary expenditure for good/service i

Finally, domestic consumption is a CES aggregate of imports and domestically produced goods given as:

Figure A.7: Nested consumption structure of the domestic economy

The above image represents the functional form describing domestic consumption is represented as a Constant Elasticity of Substitution aggregate of imports and domestically produced goods.

where σ is the Armington elasticity of substitution between imports (qm_j) and domestic supply (qa_j).

A.3.6. Government

In our model, the government collects all the taxes in the economy such as income taxes, taxes on production such as foreign worker levies and property taxes, and taxes on products such as customs and excise duties and the Value Added Tax (VAT). The government’s expenditure includes consumption of goods and services and transfers to households. The government can also regulate economic activity through policies such as trade barriers and subsidies. The levels of these taxes and trade barriers are calibrated with current UK policy and specific parameter values are provided by the GTAP database, as described in the following paragraph.

A.3.7. Model calibration

Our CGE model is written in General Algebraic Mathematical Software (GAMS), a widely-used software package that facilitates the development of CGE models and allows the user to simulate interactions within an economy. Our model is closely related to the Standard GTAP Model version 7.0, as described in van der Mensbrugghe (2018).

We use the Global Trade Analysis Project (GTAP) database version 11 to calibrate model parameters including supply chain interactions between industries and the consumption behaviour of households and governments (for an overview of GTAP version 11 see Aguiar et al., 2023). The GTAP database is compiled and updated by an international network of economists coordinated by Purdue University.

Our model is also calibrated to be consistent with the GTAP supply and use tables. The supply table shows the value of products supplied by each of these sectors, and the use table shows the value of products used by each of these sectors. The GTAP tables are internationally consistent, ensuring for instance that UK exports sum to the total international imports of UK products.

A.4. Further results

Table A.3: Real GDP impacts of the 5 essential technologies 2030 and 2035

Priority UK emerging technology	Empirical result (real increase relative to 2023 values)	2030	2035
		Upside	Upside
AI and Machine Learning	Total GDP change (%) from 2023	1.49%	3.88%
	GDP impact from 2023 (£b)	39.7	103.4
Future Telecoms	Total GDP change (%) from 2023	0.32%	0.73%
	GDP impact from 2023 (£b)	8.6	19.4
Quantum Technology	Total GDP change (%) from 2023	0.07%	0.11%
	GDP impact from 2023 (£b)	1.9	3.0
Semiconductors	Total GDP change (%) from 2023	0.17%	0.46%
	GDP impact from 2023 (£b)	4.5	12.2
Synthetic/Engineering Biology	Total GDP change (%) from 2023	1.82%	2.32%
	GDP impact from 2023 (£b)	48.4	61.9

Note: Real GDP values are deflated by the GDP deflator. Source: PwC analysis

Table A.4: Real GDP impacts of the 5 essential technologies 2030 and 2035

Priority UK emerging technology	Empirical result (real increase relative to 2023 values)	2030	2035
		Downside	Downside
AI and Machine Learning	Total GDP change (%) from 2023	0.60%	0.80%
	GDP impact from 2023 (£b)	16.1	20.0
Future Telecoms	Total GDP change (%) from 2023	0.12%	0.15%
	GDP impact from 2023 (£b)	3.2	4.1
Quantum Technology	Total GDP change (%) from 2023	0.02%	0.03%
	GDP impact from 2023 (£b)	0.6	0.7
Semiconductors	Total GDP change (%) from 2023	0.05%	0.07%
	GDP impact from 2023 (£b)	1.4	1.8
Synthetic/Engineering Biology	Total GDP change (%) from 2023	0.56%	0.63%
	GDP impact from 2023 (£b)	14.92	16.88

Note: Real GDP values are deflated by the GDP deflator. Source: PwC analysis

Annex

Data for Figure 5.1: Real GDP growth from adopting all emerging technologies by 2035 (£ billion)

	2023	2024	2025	2026	2027	2028	2029	2030	2031	2032	2033	2034	2035
Upside	0	6.80258019	15.6095	27.3951075	44.8209046	73.8347956	113.835952	154.333359	191.707249	226.455942	259.506137	290.033543	316.997119
Baseline	0	5.94968023	13.3865339	22.9184926	36.1426351	56.4750995	83.6824605	111.387355	136.902333	160.537972	183.113924	204.282516	223.386193
Downside	0	5.9964442	12.7177301	20.2314384	29.1561561	40.0336151	50.5032212	58.2985987	63.2271667	66.2475869	68.2215398	69.559542	70.4778524

Data for Figure 5.2: Real GDP growth from adopting all emerging technologies by 2035 (%)

	2023	2024	2025	2026	2027	2028	2029	2030	2031	2032	2033	2034	2035
Upside	0.0%	0.3%	0.6%	1.0%	1.7%	2.8%	4.3%	5.8%	7.2%	8.5%	9.7%	10.9%	11.9%
Baseline	0.0%	0.2%	0.5%	0.9%	1.4%	2.1%	3.1%	4.2%	5.1%	6.0%	6.9%	7.7%	8.4%
Downside	0.0%	0.2%	0.5%	0.8%	1.1%	1.5%	1.9%	2.2%	2.4%	2.5%	2.6%	2.6%	2.6%

Data for Figure 5.3: Combined contribution of the UK’s 5 essential technologies to GDP by 2035 (%)

	2023	2024	2025	2026	2027	2028	2029	2030	2031	2032	2033	2034	2035
Upside	0.0%	0.1%	0.3%	0.5%	0.9%	1.6%	2.8%	3.8%	4.8%	5.6%	6.3%	6.9%	7.4%
Baseline	0.0%	0.1%	0.2%	0.4%	0.7%	1.2%	2.0%	2.7%	3.4%	4.0%	4.5%	5.0%	5.4%
Downside	0.0%	0.1%	0.2%	0.4%	0.6%	0.9%	1.2%	1.4%	1.5%	1.5%	1.6%	1.6%	1.6%

This report has been prepared for the Government Office for Science and solely for the purpose and on the terms agreed with the Government Office for Science and cannot be relied on by anyone else. It does not constitute professional advice, and anyone other than our client should not act upon the information contained in this report without obtaining specific professional advice. No representation or warranty (express or implied) is given as to the accuracy or completeness of the information contained in this document, and, to the extent permitted by law, PricewaterhouseCoopers LLP, its members, employees and agents do not accept or assume any liability, responsibility or duty of care for any consequences of anyone acting, or refraining to act, in reliance on the information contained in this report or for any decision based on it.

© 2024 PricewaterhouseCoopers LLP. All rights reserved. ‘PwC’ refers to the UK member firm, and may sometimes refer to the PwC network. Each member firm is a separate legal entity. Please see www.pwc.com/structure for further details.

1. Emerging technologies are those that are in the early stages of development or have only recently been introduced. These technologies are often innovative, presenting new approaches and solutions that were previously not possible. They are characterised by rapid change, high levels of uncertainty, and potential for significant societal and economic impacts. Examples include artificial intelligence (AI), blockchain, quantum computing, and advanced biotechnologies. See (PwC, 2024). ↩

2. In 2021 the UK Government released the UK Innovation Strategy, setting out the national goal to become a ‘global hub for innovation by 2035’ (GOV.UK, 2021). ↩

3. Based on a recent OBR study (Economic and Fiscal Outlook, November 2023), the UK’s average annual growth rate forecast for the period 2024-2028 is 1.6%, aligning with this assumption. ↩

4. For instance, a recent study by Acemoglu and Restrepo (2019) finds negative effects of robots on employment and wages. ↩

5. Solow (1987) found that during the 1990s, a significant surge in ICT investment did not result in a corresponding increase in productivity growth, a phenomenon known as the “Solow paradox”. ↩

6. A general purpose technology or GPT is a term coined to describe a new method of producing and inventing that is important enough to have a protracted aggregate impact. See Jovanovic and Rousseau (2005). ↩

7. See also Crafts (2004a) and Antras and Voth (2003). ↩

8. See Czarnitzki, D., Fernández, G., and Rammer, C., (2023). ↩

9. See Appendix A1 for a detailed description of the sampling strategy. ↩

10. A Computer-Assisted Telephone Interviewing (CATI) approach is a survey sampling method commonly employed in market research, wherein telephone interviews are conducted to facilitate and enhance the survey. ↩

11. We rely on ONS UK industry revenue data to scale survey responses to represent the UK. To calculate the projections, we aggregated the planned investments reported for the next 5 years across various sectors. We then weighted these investments based on each sector’s contribution to the total output, obtained from input-output tables. ↩

12. Business and Insights Conditions Survey (BICS) finds that of those businesses currently using or planning to use 1 of the specified AI applications, the most common reason for doing so was creating efficiencies (35%). See (BICS, 2023). ↩

13. Section 3 provides a comprehensive literature review on how emerging technologies affect labour productivity. ↩

14. For a technical perspective on how these scenarios were constructed see Section 3. ↩

15. See, for instance, Bloom, Van Reenen, and Williams (2019). A toolkit of policies to promote innovation. Journal of economic perspectives, 33(3), pp.163-84. ↩

16. See Autor and Salomons (2018). ↩

17. Based on OECD averages for total population between 2018 - 2021. ↩

18. OECD report the indicators for 11 sectors: Manufacturing, Construction, Wholesale trade, except of motor vehicles and motorcycles, Retail trade, except of motor vehicles and motorcycles, Transportation and storage, Accommodation and Food and beverage service activities, Information and communication, Financial and insurance activities, Real estate activities, Professional, scientific and technical activities And, Administrative and support service activities. ↩

19. OECD does not collect data on financial and insurance activities; we have proxied this industry group with information and communication. ↩

20. e.g. persons employed regularly using a computer in their work (%) is related to labour, and businesses using ERP (Enterprise Resource Planning) software (%) is related to capital & others. ↩

21. This is a conservative assumption that enables laggards to retain a significant market share even as their production costs remain high relative to industry leaders. ↩