《Linux基金會：2025開源人工智能的經濟與勞動力影響研究報告（英文版）（39頁）.pdf》由會員分享，可在線閱讀，更多相關《Linux基金會：2025開源人工智能的經濟與勞動力影響研究報告（英文版）（39頁）.pdf（39頁珍藏版）》請在三個皮匠報告上搜索。

1、The Economic and Workforce Impacts of Open Source AIInsights from Industry,Academia,and Open Source Research PublicationsAnna Hermansen,The Linux FoundationCailean Osborne,PhD,The Linux FoundationMay 2025Commissioned by According to industry research,nearly all developers have experimented with open

2、 models and almost two thirds(63%)of companies are using an open model.89%of organizations who have adopted AI use open source AI(OSAI)in some form in their infrastructure.AIs 50%+reduction in business unit costs,coupled with open source softwares cost savings,suggests that OSAI holds significant po

3、tential for revenue gains.Two thirds of organizations report that OSAI is cheaper to deploy than proprietary AI,and nearly half choose OSAI because of their cost savings.Open source positively impacts AI innovation:increased inter-organizational collaboration leads to faster development of high-qual

4、ity models.Smaller businesses are adopting OSAI at a higher rate than larger businesses.OSAI will be foundational for small models powering privacy-preserving edge applications and reasoning models with higher inference-time compute.In healthcare,open models have proven to be on par with proprietary

5、 models,demonstrating that institutions can adopt OSAI without sacrificing performance.OSAI is poised to have a high impact in manufacturing,where open models provide the flexibility needed to integrate AI directly into operational processes.The Economic and Workforce Impacts of Open Source AICopyri

6、ght 2025 The Linux Foundation|May 2025.This report is licensed under the Creative Commons Attribution-NoDerivatives 4.0 International Public License.Please cite this research report when using these infographics.See citation guidance on the reports back page.THE ECONOMIC AND WORKFORCE IMPACTS OF OPE

7、N SOURCE AI|3ContentsExecutive summary 4Introduction 5Adoption rates of OSAI 7Comparing adoption by small and large businesses 9Economic benefits of OSAI 11Economic impacts of open source 11Economic impacts of AI 14Economic impacts of OSAI 17Impact on the workforce 20Market impacts of AI by industry

8、 22Healthcare 23Agriculture 24Construction 25Manufacturing 26Energy 27Conclusion 29References 31About the authors 38Acknowledgments 38THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|4Executive summaryThis study assesses and predicts the market impacts of open source artificial intelligence(OSAI

9、)through a comprehensive analysis of prior academic and industry literature as well as empirical data from Linux Foundation(LF)Research.The research is global in scope,with U.S.-and Europe-specific insights where possible.The study first reviews the adoption rates of OSAI,finding that most businesse

10、s are already adopting AI:an LF study shows that 94%of surveyed organizations have already adopted AI tools and models,and 89%of those AI adopters use some form of open source in their infrastructure.In a review of the relationship between organization size and adoption,the literature shows that the

11、re is an inverse relationship between company size and adoption rates of OSAI in particular,with smaller companies prioritizing tools that are open source.The review then analyzes existing evidence on the economic benefits of OSAI,including productivity growth,cost savings,and revenue gains.Findings

12、 from LF Research show that two thirds of surveyed organizations believe that OSAI is cheaper to deploy than proprietary AI,and nearly half cite choosing open source because of the cost savings.Research from the Harvard Business School on open source software(OSS)shows that the adoption of OSS leads

13、 to companies spending 3.5 times less than what they would if OSS didnt exist,along with increases in productivity and innovation.When aligned with evidence on AIs positive impacts on cost savings and productivity,the study provides some findings and predictions on OSAIs additional impacts on organi

14、zational costs and innovation.AIs impact on the workforce and the potential for job creation over job replacement shows that in the short term,AI is not poised to displace workers.LF Research found that 95%of surveyed hiring managers over the past two years do not plan to reduce headcount due to AI.

15、Even in the long term,AI will not necessarily lead to replacement:while AI may replace some jobs in supply chain/inventory management and customer service,most jobs will experience only partial exposure to AI automation,meaning it will become a job complement.In fact,holding skills in AI may increas

16、e workers wages by over 20%.Finally,examining the adoption and potential impacts of AI in healthcare,energy,agriculture,construction,and manufacturing,prior research predicts AIs value in the billions for each of these sectors.However,some sectors experience distinct impacts.For example,the energy s

17、ector is experiencing increased demands from data centers to meet the needs of AI,but AI can also bring significant advancements and efficiencies to the sector through monitoring energy consumption,predicting energy demand and supply,the design and deployment of new electric plants,autonomous operat

18、ion and maintenance,emissions prediction,and identification of new materials.AIs potential in healthcare is also incrediblethe productivity and resource savings could lead to$150-$260 billion additional value for the sector globallywhile the privacy and financial constraints this sector faces make o

19、pen source an attractive option.The report concludes with recommendations for future empirical research to address evidence gaps and better understand the market impacts on OSAI.THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|5IntroductionOpen models are increasingly competing withand in some c

20、ases,already surpassingproprietary models across various capabilities.Perhaps no development captured this shift better than when DeepSeeks R1 model hit the headlines in January for demonstrating capabilities matching or exceeding industry-leading proprietary models on certain benchmarks for cents o

21、n the dollar.1,2 But this is not the only example.The open source flywheel is in full motion,with other open models topping benchmark leaderboards.3 As Nathan Lambert from Ai2 articulated upon the release of OLMo 2 32B:“For a long time,people have asked for a truly open source version of ChatGPT,and

22、 we finally have it.”4Quantifiable trends reflect this momentum.According to Stanfords 2024 AI Index Report,66%of the 149 foundation models released in 2023 were open,a substantial increase from 44%in 2022 and 33%in 2021.5 The 2025 AI Index Report shows that in 2024 alone,there was a 40%increase in

23、the number of AI repositories on GitHub.6 At the same time,the number of repositories on Hugging Face Hubthe central platform for collaboration among the OSAI community7has skyrocketed,surpassing 1.5 million models and 350,000 datasets.Download data from platforms like Hugging Face Hub shows Metas L

24、lama models,Mistrals Mixtral models,and Alibabas Qwen models have been downloaded millions,if not billions,of times.8Defining OSAIThere is an ongoing debate about the definition of OSAI.Broadly speaking,it concerns the release of AI systems and their constituent components,such as software,data,mode

25、l parameters(i.e.,pre-trained weights and biases),tools,and documentation,under free and open source licenses that permit their use,study,modification,and redistribution.9 In this report,when we use the term OSAI,we focus specifically on open models in the domain of generative AI,which we define as

26、follows:Open models are defined in the Generative AI Commons Model Openness Framework as machine learning models whose architecture,parameters(i.e.,pre-trained weights and biases),and documentation are released under permissive licenses that permit their use,study,modification,and redistribution.10

27、Generative AI refers to AI systems and models that create novel outputs,such as text,images,audio,video,and/or code,by learning patterns and distributions from training data rather than following explicit programming.Generative AI includes but is not limited to:language models,which enable tasks suc

28、h as text generation and summarization;vision models,which enable tasks such as image generation and modification;and multimodal models,which are trained on data of multiple modalities,such as text,images,and audio,and accordingly enable the generation of outputs across different modalities,such as

29、text-to-image creation or image-to-text reasoning.Among these,foundation models,which are characterized by their large scale,training on diverse datasets,and adaptability to various downstream tasks,play a crucial role in the development and application of generative AI systems.10THE ECONOMIC AND WO

30、RKFORCE IMPACTS OF OPEN SOURCE AI|6OSAI has well-documented benefits for research and innovation,including enhancing reproducible research and promoting security through widespread scrutiny.11,12,13,14 Now,the market impacts of OSAI are beginning to bubble to the surface,from challenging the dominan

31、ce of industry leaders to enabling small and large businesses alike to build custom AI applications at accessible costs.15 This trend is already evident across industries,as 60%of business decision-makers report significant cost savings associated with open source,and 81%of developers place growing

32、professional value on experience with open source tools.16The market impacts of OSAI,in particular open models,are less well understood,but may bear similarities to the impacts of open source software(OSS).Prior research shows the enormous economic impact of OSS,from business productivity and entrep

33、reneurship to GDP contributions.17,18 For example,the global supply-side value of widely used OSS is an estimated$4.15 billion,with demand-side value reaching$8.8 trillion,suggesting businesses would need to spend 3.5 times more on software if OSS didnt exist.19 In the U.S.,investment in OSS in 2019

34、 was an estimated$37.8 billion,with a current-cost net stock of$74.3 billion.20 Similarly,in the EU,companies invested around 1 billion in OSS in 2018,generating between 65 billion and 95 billion for EU GDP and a cost-benefit ratio above 1:4 for companies.21 It has been well documented that choosing

35、 OSS over proprietary software contributes to cost savings,avoids vendor lock-in,and accelerates innovation through knowledge sharing,as will be discussed below.21,22The rise of OSAI raises similar questions:How will value be created and captured?How will competitive dynamics in the AI industry shif

36、t?What new business models will emerge?This report examines the market impacts of OSAI by building on this established body of research while acknowledging the unique characteristics that distinguish AI from traditional software.After reviewing adoption rates of OSAI,this literature review turns to

37、an analysis of market impacts of OSAI,how the technology is already impacting different sectors,and whether and to what extent it is impacting the workforce.The report concludes with an overview of key findings and a discussion on areas for further exploration.THE ECONOMIC AND WORKFORCE IMPACTS OF O

38、PEN SOURCE AI|7Adoption rates of OSAIIn software development,AI adoption is already ubiquitous.A 2024 study by GitHub that surveyed 2,000 software developers in the U.S.,Germany,India,and Brazil revealed that 97%of respondents had used AI tools at some point in and outside of work.23 Despite these h

39、igh numbers,the survey found some hesitation in formalized organizational adoption,as companies have to reconsider certain processes,governance,and compliance activities,as well as build a culture that trusts the technology and produces measurable results.Individualized use of AI increased beyond so

40、ftware developers.For example,a 2025 McKinsey report found that 53%of C-level executives use it regularly at work.24 This report and others find its use concentrated in certain sectors,large enterprises,and startups.25A 2024 Linux Foundation report by Lawson et al.found that,at the time of the surve

41、y,94%of respondent organizations have some level of AI adoption.26 More than half of these organizations were in the U.S.or Canada(55%),18%in Asia-Pacific,and 17%in Europe.The report then investigated the extent to which those organizations that have adopted AI are using open source.Findings show th

42、at,on average,41%of their code infrastructure is open source and 89%of adopters have at least some open source in their AI infrastructure(see FIGURE 1).FIGURE 1.AI ADOPTERS SHARE OF AI INFRASTRUCTURE THAT IS OPEN SOURCE11%of all AI adopters responded that none of their AI infrastructure is open sour

43、ce;25%responded that one-quarter to one-half of their infrastructure is open source;17%responded one-half to three-quarters;and 20%responded more than three-quarters.Source:Lawson,A.,Hendrick,S.,Rausch,N.,et al(2024 November).Shaping the Future of Generative AI:The Impact of Open Source Innovation.T

44、he Linux Foundation.https:/www.linuxfoundation.org/hubfs/LF%20Research/lfr_genai24_111924.pdfTHE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|8McKinsey,the Mozilla Foundation,and the Patrick J.McGovern Foundation recently published findings from their global survey fielded across 41 countries to

45、 over 700 technology leaders.16 Similar to the findings above,they find that nearly two thirds(63%)of respondent organizations are already using open models,and 76%expect their organization to increase its adoption of OSAI over the next few years.They also find that organizations that view AI as imp

46、ortant to their competitive advantage are more likely to use open models and tools.As Lawson et al.(2024)explain,open source is a differentiator for organizations with advanced AI capabilities that require the adaptability and control of open source,as well as the foundation for innovation and susta

47、inability it provides.26“Open source communities continuously push the boundaries of model architectures,tools,and libraries,”the report states,by offering access to the latest developments in models,frameworks,and techniques.GitHubs 2024 survey demonstrates how widespread OSAI really is,with nearly

48、 all survey respondents stating they have experimented with open models.27 Hugging Faces Open Source AI Year in Review best demonstrates this ubiquity in their collection of data at the end of 2024 to capture a snapshot of the trends in OSAI.8 Overall,its data shows the exponential growth in open mo

49、dels,increasing from a few thousand in 2022 to over a million in 2024(see FIGURE 2).FIGURE 2.GROWTH IN HUGGING FACE MODELS,FROM 2022-2024Source:Hugging Face(n.d.).Open-source AI:year in review 2024.Retrieved April 22,2025,from https:/huggingface-open-source-ai-year-in-review-2024.static.hf.space/ind

50、ex.htmlTHE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|9As noted above,research into the adoption of AI has found that the size,industry,and geographic location of the organization are significant.An analysis from 2018 Annual Business Survey(ABS)data shows that larger businesses are higher adop

51、ters of AI,and adoption among small and medium-sized businesses is low.25 However,when specifically examining startups,the researchers found that companies with high growth early in their inception tend to adopt and use AI,which they argue is the area that matters most for economic growth.As startup

52、s gain a foothold in the economy,this could fuel greater diffusion of AI and have a large influence over AIs market impact.Five years later,a 2023 survey from the Small Business&Entrepreneurship Council found that three-quarters of small businesses(75%)are using AI.28 The main drivers of AI use incl

53、uded research,time and cost savings,competitive pressures,and influence from peers.The survey found that the samples median investment in AI tools per year is$1,800,with the vast majority planning to increase their investment in the following 12 months.When looking specifically at OSAI,the Linux Fou

54、ndations 2024 survey on generative AI found an inverse relationship between company size and adoption rates(see FIGURE 3).29 When segmenting OSAI adoption by company size,the findings show that smaller companies are adopting open source at a higher level,while larger companies are prioritizing it le

55、ss.The findings also show that OSAI is more of a priority for small and medium-sized companies(see FIGURE 4).Larger enterprises may be adopting AI at a greater rate,but they appear less concerned with the openness of the model or tool.However,the small and medium-sized businesses that are adopting A

56、I are predominantly choosing open source for their environments.As the McElheran et al.(2024)study points out,the innovation and high-growth potential of startups make them important for economic growth and dynamism.25 Their prioritization of OSAI suggests a greater diffusion and impact of OSAI in p

57、articular.Comparing adoption by small and large businessesThe report by McKinsey,the Mozilla Foundation,and the Patrick J.McGovern Foundation includes a country-specific view of OSAI adoption.16 They find that India,the United Kingdom,and the United States are the countries with highest open model u

58、se,arguing that this is most likely due to the relative maturity of their respective technology sectors.THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|10FIGURE 3.RELATIONSHIP BETWEEN COMPANY SIZE AND ADOPTION OF OSAIThe larger companies(250-10,000+employees)have the highest representation in t

59、he 1%-50%range,while the smaller companies(1-249 employees)have comparatively higher representation in the 50%-75%+range.Source:The Linux Foundation(2024).2024 Generative AI Survey.Data.world.https:/data.world/thelinuxfoundation/2024-generative-ai-surveyFIGURE 4.RELATIONSHIP BETWEEN COMPANY SIZE AND

60、 THE CHOICE TO ADOPT OSAIRespondents from small companies(1-249 employees)indicated strongly positive opinions around open source,as compared to respondents from mid-sized companies(250-9,999 employees),who felt positively about open source without making it a priority.The largest companies(10,000+e

61、mployees)indicated greatest neutrality around the choice to use OSAI.Source:The Linux Foundation(2024).2024 Generative AI Survey.Data.world.https:/data.world/thelinuxfoundation/2024-generative-ai-surveyTHE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|11Economic benefits of OSAIWe first present r

62、esearch on the economic impacts of open source and then the impacts of AI to extrapolate the benefits of OSAI.It is well documented that open source software(OSS)is widely used at the company level because of the cost savings,increased productivity,and innovation it affords.Linux Foundation Research

63、s annual World of Open Source survey found that,in 2024,the community considered these three effects as some of the top benefits of OSS(see FIGURES 5&6).32Economic impacts of open sourceExpert opinions from the open source communityWhen asked for his perspective on small business adoption of OSAI,Fr

64、ank Nagle,Assistant Professor at Harvard Business School,explained that“prior research has shown that open source can substantially benefit small businesses and startups,particularly due to its lower costs than related proprietary software.OSAI will allow these resource-constrained companies to more

65、 fully benefit from cutting-edge technologies and will help them keep pace in competitive markets.”30Matt White also sees evidence of this trend in his role as Executive Director of the PyTorch Foundation and General Manager of AI at the Linux Foundation.“The data is clear that smaller businesses ar

66、e embracing OSAI at higher rates than larger enterprises,and for good reason.For startups and small businesses operating with limited resources,open models provide sophisticated AI capabilities without the prohibitive costs of building from scratch or licensing proprietary solutions.This levels the

67、playing field,allowing innovative smaller companies to compete based on their unique applications rather than being blocked by access barriers.”31THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|12FIGURE 5.INNOVATION,PRODUCTIVITY AND REDUCED COSTS ARE TOP BENEFITS OF OPEN SOURCEFIGURE 6.THE EXTE

68、NT TO WHICH OPEN SOURCE LOWERS COSTS,IMPROVES PRODUCTIVITY,AND FACILITATES INNOVATIONSource:The Linux Foundation(2024).2024 World of Open Source:Global Spotlight.data.world.https:/data.world/thelinuxfoundation/2024-world-of-open-source-global-spotlightSource:The Linux Foundation(2024).2024 World of

69、Open Source:Global Spotlight.data.world.https:/data.world/thelinuxfoundation/2024-world-of-open-source-global-spotlightLower cost of software ownershipImproved productivityLess vendor lock-inImproved software qualityMake the organization a better place to workFacilitates innovationLower cost of IT o

70、perationsImproved securityLess development time to marketTHE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|13When it comes to OSS cost savings,a 2024 study by Harvard Business School valued OSS at$4.15 billion on the supply side(the cost to recreate widely used OSS once)and$8.8 trillion on the de

71、mand side(the cost for companies to replace each piece of OSS they use if OSS did not exist at all).19 The researchers also found that,if OSS did not exist,companies would have to spend 3.5x more on software than they currently do.Using an alternate approach,Chesbrough(2023)surveyed business leaders

72、 about the benefits and costs of open source and asked respondents to calculate the cost of alternatives.33 46%of respondents said that it would have cost them at least two times the cost of OSS to write the code themselves.Korkmaz et al.(2024)applied a cost estimation model to GitHub development ac

73、tivity data across 7.6 million repositories to estimate OSS investment at$37.8 billion in 2019.20Looking specifically at Europe,EU companies invested an estimated 1 billion in OSS in 2018,which had an impact of 65-95 billion on the European economy.21 Furthermore,an estimated 10%increase in OSS cont

74、ributions would annually generate an additional 0.4%to 0.6%of GDP.Compared to the Korkmaz et al.(2024)findings,where the U.S.investment was$37.8 billion in 2019,the European investment was approximately 3%of the global investment(at the end of 2018,1 billion=$1.14 billion).According to BlackDucks 20

75、24 Open Source Security and Risk Analysis report,96%of codebases have some open source components.34 As open source becomes more prevalent in businesses,Nagle(2018)examined its impacts on productivity and found a statistically significant positive return for businesses with complementary capabilitie

76、s,such as strong IT technical expertise,IT-intensive operations,or being in IT-producing industries.17 Open source can also enable innovation.In a 2023 study,Wright et al.measured the relationship between OSS and entrepreneurship by studying new venture founding and participation on GitHub.18 The an

77、alysis found that an increase in participation on GitHub generates an increase in a countrys new technology ventures in the following year.It also found that contributing to OSS leads to higher-quality and more mission-oriented ventures engaging in socially impactful activities.The evidence from thi

78、s study indicates that OSS has a positive relationship with entrepreneurial activity.THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|14Using the same economic measurescost savings,productivity gains,and increased innovationthe literature suggests that AI also has a meaningful impact on the mark

79、et.Academic and industry analyses have thoroughly measured AIs impact on productivity.24,35-41 A 2024 PwC report found that global GDP will rise 14%by 2030 as a result of AI adoption,equivalent to$15.7 trillion.42 It argues that this economic growth will be due in part to productivity gains from aut

80、omating and complementing jobs.The Hatzius et al.(2023)report from Goldman Sachs finds a slightly smaller but similarly sizable impact,as it estimates AIs labor boost will cause an almost$7 trillion increase in global GDP over ten years.43 A third report from McKinsey(2023)also predicts a multiple-t

81、rillion-dollar boost to the global economy due to productivity gains from AI.37 The report reviews the impact annually across more than 60 use cases,finding a$2.6-$4.4 trillion addition to GDP per year.The researchers identify four main areas where AI will deliver value:software engineering,marketin

82、g and sales,customer operations,and research and development.Lawson et al.(2024)also found similar use cases for AI,including customer service,code generation,and research(see FIGURE 7).26Economic impacts of AIFIGURE 7.TOP USE CASES FOR AISource:Lawson,A.,Hendrick,S.,Rausch,N.,et al(2024 November).S

83、haping the Future of Generative AI:The Impact of Open Source Innovation.The Linux Foundation.https:/www.linuxfoundation.org/research/gen-ai-2024THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|15Focusing specifically on the impacts of AI on software development,various studies on GitHub Copilot

84、demonstrate the productivity gains for software developers.A 2022 GitHub study found that,when using the tool,over 90%of developers complete tasks faster55%faster than developers not using the tool,according to their study.36 Copilot impacts more than developer speed88%of developers feel more produc

85、tive,and over half the respondents indicated that the tool allows them to feel more fulfilled,experience less frustration,and stay in a flow state.36 An independent analysis by Faros(2024)found similar productivity gains,citing that code merges 50%faster with Copilot and lead time to production decr

86、eased by 55%for those using the tool.38 The analysis also found that code quality and security either increased or remained steady,concluding that“a 55%improvement in lead time with no collateral damage to code quality is a phenomenal ROI.”The 2024 GitHub survey findings from the previous section th

87、at found a 97%adoption rate also referenced multiple productivity benefits of using AI coding tools,including development efficiency,code-quality improvements,streamlined workflows,and faster upskilling and onboarding.23 Increased productivity allows developers to re-invest the saved time into colla

88、borative and system design tasks.Looking beyond software developers,Noy and Zhang(2023)analyzed productivity gains from ChatGPT,sampling from mid-level,college-educated professionals.39 In their experiment,they drew from a range of occupationsmarketers,grant writers,consultants,data analysts,human r

89、esource professionals,and managersand asked participants to complete 20-to 30-minute tasks relevant to their professions,which included writing press releases,short reports,analysis plans,and emails.Half the participants used ChatGPT.They found that,in comparison to the control group,the use of Chat

90、GPT decreased the time it took to complete a writing task by 40%and the quality of the output rose by 18%.The adoption of AI has also led to cost savings and revenue gains for businesses.The study by Lawson et al.(2024)asked respondents to estimate how much their investment in AI has converted to re

91、venue gain.26 Among those with higher adoption rates,35%have seen substantial or significant revenue gain,and another 20%have experienced moderate revenue gain(see FIGURE 8).THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|16The 2025 McKinsey report found that organizational adoption of AI reduc

92、ed costs across various business units.24 Specifically,in the second half of 2024,respondents reported over 50%cost decreases across six of their business units:61%in supply chain and inventory management,58%in service operations,56%in strategy and corporate finance,56%in HR,52%in software engineeri

93、ng,and 51%in risk,legal,and compliance.A 2022 study by Lee et al.investigated the relationship between AI adoption intensity and revenue growth.44 They found that a businesss revenue only increases after a certain level of investment in AI.Investments in cloud computing and database systems,as well

94、as the pursuit of R&D strategies that are specific to the venture,positively impact the relationship between AI investment and revenue.These findings show that the level of commitment to AI plays a role in AIs impact on the business.Additionally,and similar to the Nagle(2018)study,17 complementary t

95、echnologies play an important role in revenue growth.Due to its ability to increase the speed and quality of production,AI is also seen as a key technology for innovation.A 2024 Deloitte survey found that 46%of respondents cited“new ideas”and“innovation and growth”as key benefits of AI.45 More speci

96、fically,some of AIs applications,such as deep learning,could lead to“changes in the very nature of the innovation process within those domains,”according to Cockburn,Henderson,and Stern(2018).46 From an R&D perspective,the authors discuss how AI automates the process of discovery while also expandin

97、g the feasibility of research questions and the scope to address them.With these capabilities,they argue,FIGURE 8.RELATIONSHIP BETWEEN DIFFERENT LEVELS OF AI ADOPTION AND THE REVENUE GAIN FROM AISource:Lawson,A.,Hendrick,S.,Rausch,N.,et al(2024 November).Shaping the Future of Generative AI:The Impac

98、t of Open Source Innovation.The Linux Foundation.https:/www.linuxfoundation.org/research/gen-ai-2024THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|17Source:Lawson,A.,Hendrick,S.,Rausch,N.,et al(2024 November).Shaping the Future of Generative AI:The Impact of Open Source Innovation.The Linux Fo

99、undation.https:/www.linuxfoundation.org/research/gen-ai-2024There is some literature discussing the economic impacts of OSAI.A handful of studies have analyzed the impact of open source on an organizations costs when adopting AI.Lawson et al.(2024)asked respondents about the cost savings of OSAI in

100、two different ways.26 First,66%of survey respondents agreed that OSAI has a lower overall cost than proprietary AI(see FIGURE 9).Second,the survey asked whether the open source nature of a tool or model influences the adoption decision and why;46%stated cost efficiency as the justification for adopt

101、ing open source(see FIGURE 10).Economic impacts of OSAIFIGURE 9.AGREEMENT AROUND THE BENEFITS OF OSAIAI tools will entirely change research methods,allowing a transition from labor-intensive and routinized practices to the use of prediction algorithms on large datasets.Increasing and broadening rese

102、arch products in this way will lead to greater and faster innovation.THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|18FIGURE 10.HOW OPEN SOURCE INFLUENCES ADOPTION DECISIONSSource:Lawson,A.,Hendrick,S.,Rausch,N.,et al(2024 November).Shaping the Future of Generative AI:The Impact of Open Source

103、 Innovation.The Linux Foundation.https:/www.linuxfoundation.org/research/gen-ai-2024McKinsey,the Mozilla Foundation,and the Patrick J.McGovern Foundation(2025)found similar benefits related to OSAI implementation.16 In their global survey,respondents rated cost savings as the greatest benefit of usi

104、ng OSAI,and over half of the decision-makers in the survey found they had lower implementation and maintenance costs than with proprietary AI.This same group also said that OSAI performed better and was easier to use than proprietary tools.GitHubs 2024 survey also found cost savings to be an importa

105、nt factor in survey respondents choice to use open models.27Other studies have investigated OSAIs impacts on innovation.The report from McKinsey,the Mozilla Foundation,and the Patrick J.McGovern Foundation(2025)described how collaboration around open models provides an environment for accelerated in

106、novation that diminishes redundant development and allows for collective momentum.16 Open sourcing a model may impact the speed of innovation,as greater community collaboration accelerates the development and quality of a product.47 As found in the Linux Foundations 2025 report on open standards,eng

107、agement in open,collaborative activities is a better indicator of innovation than patents.48 Through a case study on Metas contribution of the deep learning framework PyTorch to the Linux Foundation,Yue and Nagle(2024)examine the effects of governance changes in OSAI software projects on innovation

108、and collaboration when they transition from unilateral governance under one company to open governance under a non-profit foundation.49 They found three trends:there was a significant decrease in contributions from Meta;there was a large increase from external companies,especially from the developer

109、s of complementary technology,such as chip manufacturers;and there was no change in participation from PyTorch users,such as app developers.This evidence shows that open governance promotes broader participation and increased contributions and decreases the dominance of any single company in the dev

110、elopment of industry-leading OSAI software.Intelligent Power Plant OperationsCost EfficiencySecurity32%46%52%THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|19Expert opinions from the open source communityWhen asked about the economic impacts of OSAI,Frank Nagle pointed to the broad and large e

111、conomic impact of OSS in his existing research.He argued that“as AI becomes more prevalent and more impactful than standard software,the economic impact of OSAI is likely to be substantially larger than traditional OSS.”30 In his response,Matt White also compared known OSS impacts to OSAI.“The econo

112、mic impact of open source software gives us a compelling preview of OSAIs potential.Research shows companies would spend 3.5 times more on software if open source didnt exist,and a similar multiplier effect is emerging with open models,where organizations are already reporting cost efficiencies of o

113、ver 60%compared to proprietary alternatives.OSAI is democratizing access to technology that would otherwise remain financially out of reach for many organizations,especially startups and research labs.”31As Hoffmann,Nagle,et al.(2024)found,companies would have to spend 3.5 times more on software if

114、open source software(OSS)didnt exist.19 Applying this to the AI market,OSAI could contribute to similar savings in business costs.While AI is already reducing the costs of some business units by over 50%,coupling this with the savings from OSS could mean that adopting OSAI leads to even higher savin

115、gs than Hoffmann,Nagle,et al.s estimate for OSS.Looked at another way,AI increases a business revenue gains.It is possible that revenue gains from productivity increases are even higher with OSAI due to its lower costs than proprietary AI.Extrapolating general open source software benefits to the AI

116、 landscape:PredictionsAs for innovation,open source will undoubtedly play a significant role in the diffusion and adoption of AI,as we have seen in other technology domains.The report from McKinsey,the Mozilla Foundation,and the Patrick J.McGovern Foundation(2025)predicts OSAI will have an impact in

117、 two key areas:small language models powering privacy-focused edge applications and reasoning models with higher inference-time compute.16 Open sources role in advancing these technology domains may lead to greater AI adoption overall,as both are key to widespread use.THE ECONOMIC AND WORKFORCE IMPA

118、CTS OF OPEN SOURCE AI|20Impact on the workforceThere has been significant speculation that AI will cause widespread job displacement through automation.However,research has demonstrated that this is not necessarily the caseor at least,that the picture is more nuanced than that.50 Looking at the near

119、er term,the Linux Foundations 2024 study on technical talent surveyed hiring managers around the world and found that AI reduced organization headcounts for only 5%of respondents in 2023 and 2024.51 In fact,these respondents predicted that AI would have little impact on headcount in the next year or

120、 may actually increase headcount as hiring managers prioritize AI as an area for staffing(see FIGURES 11&12).The latest data from the 2025 survey shows a continuation of this trend,where more organizations are hiring than downsizing due to AI.52FIGURE 11.AIS IMPACT ON ORGANIZATIONAL HEADCOUNTSource:

121、Lawson,A.(2024,April).2024 State of Tech Talent Report:Survey-Based Insights into the Current State of Technical Talent Acquisition,Retention,and Management Globally.The Linux Foundation.https:/www.linuxfoundation.org/research/open-source-jobs-report-2024THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SO

122、URCE AI|21FIGURE 12.AI IS A TOP STAFFING PRIORITYSource:Lawson,A.(2024,April).2024 State of Tech Talent Report:Survey-Based Insights into the Current State of Technical Talent Acquisition,Retention,and Management Globally.The Linux Foundation.https:/www.linuxfoundation.org/research/open-source-jobs-

123、report-2024As adoption rates increase and AI moves more into the core infrastructure of organizations,various studies indicate that“AI has the potential to change the anatomy of work.”37 The automation of activities will impact some jobs and skill sets more than others.The 2025 McKinsey report found

124、 that over the next three years,AI will decrease headcount in specific areas,including service operations and supply chain.24 Similarly,Hatzius et al.(2023)estimate that AI could automate one-fourth of current work duties,with the greatest impact in administrative and legal jobs and a lower impact i

125、n construction and other trades occupations.43 Eloundou et al.(2023)show that 19%of workers in the U.S.may see 50%or more of their tasks affected by AI,and employees could complete about 15%of all tasks in the U.S.significantly faster while maintaining the same level of quality when using an LLM.53H

126、owever,those areas with greater AI exposure will not necessarily lead to workforce displacement and instead will augment work.54 PwCs 2024 AI Jobs Barometer found that in the sectors with greater AI penetration,productivity growth is increasing steadily,and jobs that require AI skills have up to a 2

127、5%wage premium.55 Those workers in these sectors need to learn how to use AI to stay relevant and build these new skills on the job.56 As Stephany and Teutloff(2024)find,AI skills are valuable because they have high skills complementarityin that workers can combine them with a high and diverse numbe

128、r of other valuable skillsand they find that having these skills will increase worker wages by an average of 21%.57 Hatzius et al.(2023)found that most jobs will only be partially exposed to AI,meaning that AI will help complete the job instead of replacing the worker.43A recent study by the OECD,Bo

129、ston Consulting Group,and INSEAD(2025)surveyed 1,007 enterprises around the world and identified a scarcity of skills and specialized talent needed for greater AI adoption.58 In the survey,76%of respondents indicated that information on accreditation schemes would be helpful in the adoption of AI,as

130、 enterprises face challenges in understanding the specific skills they seek in their employees for successful and sustainable AI adoption.According to theTHE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|22report,existing academic certifications are not sufficient in providing this information,an

131、d new qualification frameworks are needed to describe candidates competencies.Moving forward,public-private collaborations are key to address these gaps by creating relevant training and certification.One existing example is the practitioner certification program from scikit-learn,a widely used OSS

132、for machine learning.59Looking at the long term,Hatzius et al.(2023)predict through modeling that AIs labor cost savings,new job creation,and productivity boost will lead to a labor productivity boom similar to that of historic technologies such as the personal computer.43 They estimate global produ

133、ctivity will increase by 1.4%over a ten-year period.MicKinseys 2023 report calculates an annual labor productivity growth of 0.1%to 0.6%through 2040,depending on adoption rates and redeployment of workers.37 Looking specifically at software developers,Hoffmann,Boysel,et al.(2024)studied the individu

134、al-level effects of GitHub Copilot on task allocation.60 They found that the use of this tool allowed developers to shift away from project management,with these activities decreasing by 10%.Instead,they could focus on core coding tasks,increasing this activity by 5.4%.They also found that Copilot i

135、ncreased developers exposure to new programming languages,which opens up opportunities to command a higher salary.As discussed in the previous section,the productivity gains for other workers using tools such as ChatGPT could lead to a shift in task allocation,with more time available for higher-imp

136、act work.39An industry-specific analysis of AI is useful to acknowledge its widespread but asymmetric impacts on the global economy.For example,in their analysis of the 2018 ABS,McElheran et al.(2024)found that American manufacturing,information,and healthcare sectors have the highest rates of AI us

137、e,as each sector has 11%-12%of businesses reporting some use of AI.25 On the lower end of the scale,construction businesses were at 4%,and agriculture,mining,and utilities were grouped together at under 6%.The 2025 McKinsey report found the highest AI workplace adoption in 2024 in technology(88%),pr

138、ofessional services(80%),advanced industries(79%),and media and telecom(79%).24 On the lower end were healthcare(63%)and energy and materials(59%).Comparing the 2018 ABS data with the 2024 data provides insight into how AI adoption has evolved and shifted from 2018 to today.Importantly,AI adoption h

139、as significantly increased across all sectors.In the following section,we review the market impacts of AI across five sectors:healthcare,agriculture,construction,manufacturing,and energy.This study chose these sectors for in-depth analysis because of their role in providing critical everyday goods a

140、nd services and their significant share of the global economy.Combined,these sectors represent 44%of global GDP and employ approximately 1.5 billion people around the world.Market impacts of AI by industryTHE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|23It is well established that AI promises

141、significant productivity gains for the healthcare sector.61 However,it appears that its use remains more of a promise than a fact in the clinical domain.A 2024 Tebra survey found that only 10%of healthcare professionals use AI in their work,with 50%planning to adopt it in the future.62 The use of AI

142、 in healthcare will free up clinical resources,save costs,and potentially increase efficiency through its ability to automate tasks and decision support,aid in diagnoses and detection of other symptoms,and predict clinical outcomes such as hospital wait times and ICU transfers.62,63,64 The 2023 McKi

143、nsey report found that the global healthcare sector stands to gain$150-$260 billion in value from AI by applying it across business functions,with supply chain and operations,marketing and sales,and customer operations representing the top three areas.37When it comes to the value of OSAI in healthca

144、re,some research exists.Lawson(2024a)found that,in healthcare in particular,AI and ML stand to benefit the most from open source,ahead of other technology realms such as cybersecurity and the cloud(see FIGURE 13).65HealthcareFIGURE 13.KEY OPEN SOURCE TECHNOLOGIES IN HEALTHCARE1 Healthcare expenditur

145、e represented 10%of global GDP in 2022 and the sector employed 65 million workers worldwide in 2020;Agriculture represented 4%of global GDP in 2023 and employed 892 million people worldwide in 2022;Current data suggests that construction represents 13%of global GDP and employs more than 100 million

146、people worldwide;Manufacturing represented 15%of global GDP in 2023 and employs 400 million people worldwide as of 2025;Energy rents(oil,natural gas,and coal)represented 2.1%of the worlds GDP in 2021 and employed 41 million people across the globe in 2019Source:Lawson,A.(2024,December).2024 Global S

147、potlight Insights Report:The Role of Open Source in Uniting Innovation,Collaboration,and Resilience Across Regions and Industries.The Linux Foundation.https:/www.linuxfoundation.org/research/world-of-open-source-global-20242024 World of Open Source Survey,Q14,Q29,Q36,Sample Sizes=59,58,54 Respective

148、ly THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|24OSAI is particularly attractive in settings with limited resources,where free and flexible tools are important.66 In August 2024,Buckley et al.(2024)ran an analysis of the Llama 3.1 model against OpenAIs GPT-4 and found the results of Llama 3

149、.1 on par with GPT-4.67 In their discussion,they defined Llama as an open model and argued that,given their findings,institutions may be able to start adopting open source solutions when building their own custom models to run locally,without sacrificing privacy or performance.An editorial from Spri

150、nger Nature discusses the benefits of OSAI in healthcare settings,using DeepSeeks DeepThink(R1)as an example.68 The authors developed 12 key aspects of open models,including cost efficiency,scalability,and fine-tuning abilities.Combined with AIs clinical productivity gains in automation,detection,an

151、d prediction,the ability to rely on open models to produce customized,cost-effective,and privacy-preserving solutions trained on local data increases revenue gains and model accuracy.On top of these organization-level gains,the open source nature of a model such as DeepThink allows for continuous le

152、arning by integrating with publicly available datasets.These models can stay up to date on the latest scientific research and advancements in healthcare,enhancing the models performance.Demonstrating these capabilities,the author indicates that this leads to faster,lower-cost discoveries.Therefore,O

153、SAI provides cost-saving opportunities within the healthcare process while also accelerating the science of medicine.AgricultureThe 2023 McKinsey report analyzed AIs impacts on the agriculture sector.37 The report found that AI could add$40-$70 billion in value to the sector when companies primarily

154、 apply it in marketing and sales,software engineering,and supply chain and operations.The PwC report ran a modeling exercise in 2020 that scored the impact of key AI use cases across a few different sectors,including agriculture.69 The report identified key applications,such as robotics,precision mo

155、nitoring of environmental conditions,land use management,and crop monitoring.It found that these applications in combination have the potential to increase global GDP by 0.2%to 0.3%.DeClerq et al.(2024)studied the use of AI in the agriculture sector and found that AI could have an impact on global f

156、ood production challenges by providing on-demand advice and training to farmers,controlling machinery,advancing research through data wrangling,bridging linguistic barriers,and monitoring agricultural shocks.70 The World Economic Forum(WEF)considers the adoption of AI and analytical modeling an impo

157、rtant component of“regenerative agriculture,”a new approach to farming that is estimated to produce profit increases as high as 120%.71 As mentioned above,AI allows farmers to build advanced monitoring systems and predictive analytics.In particular,applying AI this way could boost the agricultural G

158、DP of low-and middle-income countries by$450 billion or more,according to the WEF.71THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|25For lower-income countries,access to cost-saving open source technologies is critical to also reap the benefits of digital agriculture.Semios is an example of an

159、 open source agricultural tool,providing precision crop monitoring and environmental conditions tracking.72 The tool uses the open source TensorFlow in its AI insect monitoring tool.The AI assistant Farmer.Chat is another example of a tool,providing localized and tailored advice to farmers built usi

160、ng the Llama model.73 Farmers all over the world can adopt these tools while keeping their advisory costs minimal.ConstructionThe 2023 McKinsey report on AIs economic impacts included an analysis of the construction sector.37 It found that AI will boost the industry by$90-$150 billion(0.7%-1.2%of it

161、s revenue)when companies apply it primarily to marketing and sales,product R&D,and supply chain and operations functions.A 2021 Adroit Market Research study identified AIs key use case as the analysis of past projects to predict delays,identify risks,build schedules,forecast bottlenecks,and provide

162、proactive advice for managers.74 In the Asia-Pacific region,a 2025 Deloitte and Autodesk survey of the construction sector found that 37%of respondents were already using AI,with 33%planning to use it in the future.75 AI adoption in construction is low,but the sector is well positioned to adopt AI b

163、ecause of its reliance on data-driven decision-making.76 The fastest-growing market appears to be Asia-Pacific,while the largest market is North America.77 AI use in pre-construction phases,such as planning,designing,and building information modeling,indicates the value of developing an open and acc

164、essible foundation model so that new builds take a standardized approach across the sector.Expert opinions from the open source communityBriq,a software solution that provides digital labor to sectors such as construction,relies on an open model to fine-tune its solution based on industry-specific t

165、opics and issues.78 AI provides important cost savings to its clients through task automation,and using an open model means it can optimize solutions to meet its clients specific contexts and needs.Bassem Hamdy,CEO and Co-founder of Briq,stated that“Briqs digital workers use open source LLMs to unde

166、rstand natural language,machine learning to optimize decisions,and automation engines to take action across dozens of platforms.They can read submittals,validate risk documentation,process payroll,and even forecast revenue.They replicate the faculties of a human worker(seeing,reading,thinking,decidi

167、ng,and acting)and serve as project engineers,compliance officers,financial analysts,and risk managers.They dont just reduce coststhey change how companies operate.”79THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|26ManufacturingThe McKinsey analysis(2023)found that AI will boost the advanced m

168、anufacturing industry by$170-$290 billion when companies apply it primarily to product R&D,marketing and sales,and software engineering functions.37 Lawson(2024a)found that AI is the second area to most benefit from open source,behind operating systems(see FIGURE 14).65FIGURE 14.KEY OPEN SOURCE TECH

169、NOLOGIES IN MANUFACTURINGSource:Lawson,A.(2024,December).2024 Global Spotlight Insights Report:The Role of Open Source in Uniting Innovation,Collaboration,and Resilience Across Regions and Industries.The Linux Foundation.https:/www.linuxfoundation.org/research/world-of-open-source-global-2024This se

170、ctor produces enormous amounts of data1,1812 petabytes per year,according to a 2020 Deloitte reportand processing and using this data greatly enhances decision-making processes along the manufacturing line.80 For example,the implementation of AI for smart production allows for the automation of fact

171、ory tasks,order management,and scheduling.Heimburger et al.(2024)examine the factors that determine AI adoption in production and manufacturing environments and discuss the technologys potential in the areas of maintenance,quality control,and production planning.81 AIs potential in manufacturing is

172、massivethe global AI market size for the manufacturing industry has been worth more than$70 billion since 2023,and the vast majority of companies believe it to be a pivotal technology for growth and innovation in the sector.80 However,this potential is still nascent.Only 15%of companies that Deloitt

173、e surveyed are in the implementation stage,while the rest are in the proposal and pilot stages.802024 World of Open Source Survey,Q14,Q29,Q36,Sample Sizes=101,99,94 Respectively THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|27EnergyThe energy sector is embracing AI,with nearly three-quarters(

174、74%)of energy and utility companies around the world indicating they have implemented or explored AI in 2023.82The impacts of AI on energy demand:AI puts significant demand on data center power.83,84,85 This pressure on energy demand will necessitate innovation toward energy-efficient solutions,such

175、 as model software optimizations,new techniques at the data center level,or models designed for more efficient chips.86 Developing AI infrastructure with energy efficiency in mind can cause a re-drawing of AIs energy demand,meaning that the industry impacts can decrease over time.84,87The impacts of

176、 AI on energy Operations:AI will also have an operational impact on the energy sector.The 2023 McKinsey report found that incorporating AI into business functions such as customer operations and marketing and sales could boost the revenue of the global energy sector by$150-240 billiona 1%-1.6%increa

177、se in the industrys revenue.37The 2020 PwC report also scored AIs impacts in the energy sector.69 The report identified the main applications of AI as monitoring energy consumption,predicting energy demand and supply,coordinating decentralized networks,and increasing the efficiency of assets.It pred

178、icted a 1.6%-2.2%increase in global GDP by 2030.The report argued that smart monitoring lowers energy costs,which boosts economic activity.It also argued that coordinating decentralized networks increases distribution which leads to greater productivity for the energy sector.Expert opinions from the

179、 open source communityWhen asked about sector-specific impacts,Frank Nagle commented:“the sectors where OSAI will have the greatest impact are,first and foremost,in software development,where the users are already tech-savvy and we already see substantial impacts;and second,in manufacturing,where op

180、en models provide the flexibility to be integrated directly into operational processes to great effect.”30THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|28There are a number of different opportunities specific to the U.S.energy market for AI to solve problems in the sector.In nuclear energy,po

181、wer grid operations,carbon management,energy storage,and energy materials management,AI can accelerate and facilitate various activities such as licensing processes,design and deployment of a plant,autonomous operation and maintenance,emissions prediction,and identification of new materials.88 As th

182、e Argonne National Laboratory estimated in 2024,AI has the potential to reduce commercial power plant design and licensing schedules by around 20%across new clean energy,with a savings potential in the hundreds of billions of dollars by 2050.88A 2024 LF Energy whitepaper discusses similar use cases

183、for AI,including forecasting energy demand and supply,optimizing energy systems,managing asset reliability and performance,and performing long-term planning(see FIGURE 15).89 The whitepaper argues that using AI in these capacities saves costs and increases productivity by replacing human resources w

184、ith AI for these activities,improving the breadth and depth of information we have on energy systems and increasing the speed of these activities.However,the whitepaper notes the relatively low industry adoption of AI.It points to concerns around privacy and regulatory compliance,lack of AI readines

185、s,and the difficulty in setting up industry-academia collaborations.To address these obstacles,the whitepaper points to the value of open source tools to increase collaboration and standardization to build an industry-wide pre-competitive layer on which players can build their own solutions.As the I

186、EA argues,“The technology is uniquely placed to support the simultaneous growth of smart grids and the massive quantities of data they generate.”90 Using open models and tools can ease the pressure on energy demand just as fast as the optimization of the grid.FIGURE 15.AI USE CASES IN THE ENERGY SEC

187、TORSource:LF Energy(2025,January).Unlocking AIs Potential for the Energy Transition through Open Source.The Linux Foundation.https:/lfenergy.org/unlocking-ais-potential-for-the-energy-transition-through-open-source/AI AssistantsAsset Management and ReliabilityEnvironmental Impacts and ResiliencyInte

188、lligent Power Plant OperationsDemand Scheduling and FlexibilityGrid Interactive Smart CommunitiesLong Term PlanningOptimized DesignsForecasting Supply and DemandAccelerated Optimization and SimulationData Access,Digital Twins and Realistic Open Benchmarks for AI InnovationsTHE ECONOMIC AND WORKFORCE

189、 IMPACTS OF OPEN SOURCE AI|29It is possible that this sector will face resistance to OSAI adoption.As Groopman and Lindstrom(2023)point out in their analysis of the microgrid sector,the novel application of AI is a significant differentiator for those managing the distribution and control of energy.

190、As a result,these entities may wish to remain proprietary as a form of competitive advantage,at least in the near term.92Expert opinions from the open source communityAlex Thornton,Executive Director of LF Energy,identifies two key features of open source that will drive its adoption in this sector:

191、collaboration and transparency.“OSAI presents an approach to uniquely address many of the challenges facing the intersection of AI with energy,”he comments.“Open source collaboration has already yielded step-change improvements in AI compute and energy efficiency,and I expect this trend only to cont

192、inue as OSAI commoditizes proprietary approaches.In applying AI to energy systems,trust is essential.Open source is the only way to have this necessary trust through extreme transparency and confidence in the digital supply chain.”91ConclusionThis report provides a review of prior research and empir

193、ical data on the adoption rates,market impacts,and workforce effects of OSAI.The evidence base shows that the use and adoption of AI tools are already becoming ubiquitous,with open source making up a significant portion of this adoption.A majority of organizations using AI have adopted open models,a

194、nd an average of 41%of adopters code infrastructure is open source.Adoption is asymmetrical in certain sectors,geographies,and occupations,with open source being a particular priority of small and medium-sized businesses.The economic impacts of OSAI can be predicted on the basis of the proven econom

195、ic impacts of open source software(OSS).Research has shown that the top benefits of OSS are cost savings,productivity gains,and faster time to innovation.If OSS did not exist,companies would have to spend 3.5x more on software than they currently do.The literature also showed that increased OSS is r

196、elated to returns on productivity and entrepreneurship.When examining AIs economic impacts,its potential boost to GDP is in the trillions(as much as$15 trillion42)due to increased productivity and innovation.Although its adoption is not uniformly high across all sectors,it shows significant potentia

197、l within healthcare,energy,agriculture,construction,and manufacturing.When looking at its THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|30impact on the workforce,it appears that AI will act more as a task complement than a job replacement mechanism,and the skills necessary to use AI will comm

198、and a wage premium.This review reveals significant evidence gaps in our understanding of OSAIs economic impacts.To advance our understanding of the economic impacts of AI,we recommend the following directions for future research that draw on econometric methodologies previously used to quantify the

199、economic value of OSS.Specifically,we recommend future studies to:Empirical evidence on these questions,among others,would contribute to a more comprehensive,evidence-based understanding of the adoption and economic impacts of OSAI,in particular open models,ultimately helping to guide future OSAI in

200、vestment,policy,and adoption decisions.1.Investigate the effects of OSAI adoption,particularly open models,on total AI market growth,including but not limited to complementary innovation,services,and applications;2.Measure the economic returns on investment in OSAI infrastructure,providing insights

201、for both policymakers and organizational decision-makers considering resource allocation toward open models,datasets,and related components.3.Examine the relationship between OSAI adoption and innovation,such as new venture creation,patent applications,and R&D efficiency;4.Measure the cost different

202、ial between implementing open versus proprietary AI solutions across organizational sizes,sectors,and geographies;5.Quantify the impacts on worker productivity and satisfaction across different tasks and sectors that are attributable to the adoption of open models.THE ECONOMIC AND WORKFORCE IMPACTS

203、OF OPEN SOURCE AI|31References1.DeepSeek-AI(2025).DeepSeek-R1:Incentivizing Reasoning Capability in LLMs via Reinforcement Learning.Hugging Face.https:/huggingface.co/deepseek-ai/DeepSeek-R12.Heikkil,M.(2025,January 28).DeepSeeks aha moment creates new way to build powerful AI with less money.Financ

204、ial Times.https:/ 13).Gemma 3,OLMo 2 32B,and the growing potential of open-source AI.Interconnects.https:/www.interconnects.ai/p/gemma-3-olmo-2-32b-and-the-growing 4.Lambert,N.(2025,March).A very exciting day for open-source AI!We at Ai2 are releasing our biggest open source model yet Post.LinkedIn.

205、https:/ al(2024,April).The AI Index 2024 Annual Report.AI Index Steering Committee,Institute for Human-Centered AI,Stanford University,Stanford,CA.https:/hai.stanford.edu/ai-index/2024-ai-index-report 6.Maslej,N.,Fattorini,L.,Perrault,R.,et al(2025,April).The AI Index 2025 Annual Report.AI Index Ste

206、ering Committee,Institute for Human-Centered AI,Stanford University,Stanford,CA.https:/hai.stanford.edu/ai-index/2025-ai-index-report7.Osborne,C.,Ding,J.&Kirk,H.R.(2024).The AI community building the future?A quantitative analysis of development activity on Hugging Face Hub.Journal of Computational

207、Social Science,7,20672105.https:/doi.org/10.1007/s42001-024-00300-8 8.Hugging Face(n.d.).Open-source AI:year in review 2024.Retrieved April 22,2025,from https:/huggingface-open-source-ai-year-in-review-2024.static.hf.space/index.html9.The Open Source AI Definition 1.0.(2024).Open Source Initiative.h

208、ttps:/opensource.org/ai/open-source-ai-definition 10.White,M.,Haddad,I.,Osborne,C.,et al.(2024,October 18).The Model Openness Framework:Promoting Completeness and Openness for Reproducibility,Transparency,and Usability in Artificial Intelligence.ArXiv.https:/doi.org/10.48550/arXiv.2403.13784 11.Tiwa

209、ri,U.(2024,March 27).Policy Readout.The Columbia Convening on Openness and AI.https:/foundation.mozilla.org/en/research/library/policy-readout-columbia-convening-on-openness-and-ai/12.Eiras,F.,Petrov,A.,Vidgen,B.et al(2024,May 29).Risks and Opportunities of Open-Source Generative AI.ArXiv.https:/arx

210、iv.org/abs/2405.08597THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|3213.Seger,E.,Dreksler,N.,Moulange,R.et al(2023,September 29).Open-Sourcing Highly Capable Foundation Models:An evaluation of risks,benefits,and alternative methods for pursuing open-source objectives.ArXiv.https:/arxiv.org/ab

211、s/2311.0922714.Basdevant,A.,Franois,C.,Storchan,V.et al(20214,May 17).Towards a Framework for Openness in Foundation Models:Proceedings from the Columbia Convening on Openness in Artificial Intelligence.ArXiv.https:/arxiv.org/abs/2405.15802 15.Statement from Economists on the Importance of Open Sour

212、ce AI.(2024,December 10).Mozilla.https:/open.mozilla.org/economists/16.Open source technology in the age of AI.(2025,April).McKinsey&Company,the Patrick J.McGovern Foundation,and the Mozilla Foundation.https:/ 4).Open Source Software and Firm Productivity.Management Science,65(3),1191-1215.https:/do

213、i.org/10.1287/mnsc.2017.2977 18.Wright,N.L.,Nagle,F.,Greenstein S.(2023).Open source software and global entrepreneurship.Research Policy,52(9).https:/doi.org/10.1016/j.respol.2023.104846 19.Hoffmann,M.,Nagle,F.,&Zhou,Y.(2024 January).The Value of Open Source Software.Harvard Business School Working

214、 Paper,No.24-038.https:/www.hbs.edu/ris/Publication%20Files/24-038_51f8444f-502c-4139-8bf2-56eb4b65c58a.pdf20.Korkmaz,G.,Calderon,J.B.S.,Kramer B.L.,et al(2024).From GitHub to GDP:A framework for measuring open source software innovation.Research Policy,53(3).https:/doi.org/10.1016/j.respol.2024.104

215、954 21.Blind,K.,Bhm,M.,Grzegorzewska,P.,et al(2021).The impact of Open Source Software and Hardware on technological independence,competitiveness and innovation in the EU economy,Final Study Report.European Commission,Brussels.https:/www.ospi.es/export/sites/ospi/documents/documentos/CNECT_OpenSourc

216、eStudy_EN_28_6_2021_LMBhSihnCeC7JEDsHXkK1JlZ0_79021_compressed.pdf22.The Linux Foundation(2024).2024 World of Open Source:Global Spotlight.data.world.http:/data.world/thelinuxfoundation23.Daigle,K.&GitHub Staff(2024,August 20).Survey:The AI wave continues to grow on software development teams.GitHub

217、 Blog.https:/github.blog/news-insights/research/survey-ai-wave-grows/24.Singla,A.,Sukharevsky,A.,Yee,L.,et al(2025,March).The state of AI:How organizations are rewiring to capture value.QuantumBlack AI by McKinsey.https:/ al(2024,January 24).AI adoption in America:Who,what,and where.Journal of Econo

218、mics&Management Strategy,33(2),375-415.https:/doi.org/10.1111/jems.12576THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|3326.Lawson,A.,Hendrick,S.,Rausch,N.,et al(2024 November).Shaping the Future of Generative AI:The Impact of Open Source Innovation.The Linux Foundation.https:/www.linuxfoundat

219、ion.org/research/gen-ai-2024 27.Finley,K.(2025,January 28).Open source AI is already finding its way into production.GitHub Blog.https:/github.blog/ai-and-ml/generative-ai/open-source-ai-is-already-finding-its-way-into-production/28.Small Business AI Adoption Survey(2023 October).Small Business&Entr

220、epreneurship Council.https:/sbecouncil.org/wp-content/uploads/2023/10/SBE-Small-Business-AI-Survey-Oct-2023-FINAL.pdf 29.The Linux Foundation(2024).2024 Generative AI Survey.data.world.https:/data.world/thelinuxfoundation/2024-generative-ai-survey 30.F.Nagle,personal communication,April 23,2025.31.M

221、.White,personal communication,April 25,2025.32.The Linux Foundation(2024).2024 World of Open Source:Global Spotlight.data.world.http:/data.world/thelinuxfoundation33.Chesbrough,C.(2023 March).Measuring the Economic Value of Open Source:A Survey and a Preliminary Analysis.The Linux Foundation.https:/

222、www.linuxfoundation.org/research/measuring-economic-value-of-os 34.2024 Open Source Security and Risk Analysis Report:Your guide to securing your open source supply chain(2024).BlackDuck.https:/ Simple Macroeconomics of AI.NBER Working Paper,No.32487.DOI:10.3386/w32487 36.Kalliamvakou,E.(2022,Septem

223、ber 7).Research:quantifying GitHub Copilots impact on developer productivity and happiness.GitHub Blog.https:/github.blog/news-insights/research/research-quantifying-github-copilots-impact-on-developer-productivity-and-happiness/37.Chui,M.,Hazan,E.,Roberts,R.,et al(2023,June).The economic potential

224、of generative AI:The next productivity frontier.McKinsey&Company.https:/ 17).Is GitHub Copilot Worth It?Real-World Data Reveals the Answer.Faros.https:/www.faros.ai/blog/is-github-copilot-worth-it-real-world-data-reveals-the-answer 39.Noy,S.,&Zhang,W.(2023,July 13).Experimental evidence on the produ

225、ctivity effects of generative artificial intelligence.Science,381(6654),187-192.https:/doi.org/10.1126/science.adh2586THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|3440.Czarnitzki,D.,Fernandez,G.P.,Rammer,C.(2023,July).Artificial intelligence and firm-level productivity.Journal of Economic Be

226、havior&Organization,211,188-205.https:/doi.org/10.1016/j.jebo.2023.05.008 41.Is generative AI a game changer?(2024,February 14).J.P.Morgan.https:/ the prize:Whats the real value of AI for your business and how can you capitalise?(2017)PwC.https:/ al(2023,March 26).The Potentially Large Effects of Ar

227、tificial Intelligence on Economic Growth.Goldman Sachs.https:/ 44.Lee,Y.S.,et al(2022,December).When does AI pay off?AI-adoption intensity,complementary investments,and R&D strategy.Technovation,118.https:/doi.org/10.1016/j.technovation.2022.10259045.Rowan,J.,Ammanath,B.,Perricos,C.,et al(2025,Janua

228、ry).Now decides next:Generating a new future.Deloitte.https:/ 46.Cockburn,I.M.,Henderson,R.,Stern,S.(2018,March).The Impact of Artificial Intelligence on Innovation.NBER Working Paper,No.24449.DOI 10.3386/w24449 47.Habibi,M.(2025,January 22).Open Sourcing GPTs:Economics of Open Sourcing Advanced AI

229、Models.ArXiv.https:/arxiv.org/abs/2501.11581 48.Burson,J.,&Gerosa,M.(2025,March).The State of Open Standards:Standardization and Patents in Organizations.The Linux Foundation.https:/www.linuxfoundation.org/research/state-of-open-standards-202449.Yue,D.,&Nagle,F.(2024).Igniting Innovation:Evidence fr

230、om PyTorch on Technology Control in Open Collaboration.Harvard Business School Working Paper No.25-013.https:/www.hbs.edu/ris/Publication%20Files/25-013_e2daa894-2ccc-4ea5-b338-85e5bf073fdc.pdf 50.Hampole,M.,et al(2025,February).Artificial Intelligence and the Labor Market.NBER Working Paper,No.3350

231、9.DOI:10.3386/w3350951.Lawson,A.(2024,April).2024 State of Tech Talent Report:Survey-Based Insights into the Current State of Technical Talent Acquisition,Retention,and Management Globally.The Linux Foundation.https:/www.linuxfoundation.org/research/open-source-jobs-report-2024 52.The Linux Foundati

232、on(2025).The 2025 State of Tech Talent.data.world.http:/data.world/thelinuxfoundationTHE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|3553.Eloundou,T.,et al.(2023,August 22).GPTs are GPTs:An Early Look at the Labor Market Impact Potential of Large Language Models.ArXiv.https:/arxiv.org/pdf/2303.

233、1013054.Gmyrek,P.,et al(2023).Generative AI and jobs:A global analysis of potential effects on job quantity and quality.ILO Working Paper 96.Geneva:International Labour Office.https:/www.ilo.org/publications/generative-ai-and-jobs-global-analysis-potential-effects-job-quantity-and55.PwCs 2024 AI Job

234、s Barometer:How will AI affect jobs,skills,wages,and productivity?(2024).PwC.https:/ 56.Goel,S.(2023,December 11).Competence over Credentials:The Rise of Skills-Based Hiring.Boston Consulting Group.https:/ 57.Stephany,F.,&Teutloff,O.(2024,January).What is the price of a skill?The value of complement

235、arity.Research Policy,53(1).https:/doi.org/10.1016/j.respol.2023.104898 58.OECD,BCG,INSEAD(2025).The Adoption of Artificial Intelligence in Firms:New Evidence for Policymaking.OECD Publishing,Paris.https:/doi.org/10.1787/f9ef33c3-en 59.Gittos,Penelope(2024,October 31).Official Scikit-learn Certifica

236、tion Launch.Probabl.https:/papers.probabl.ai/official-scikit-learn-certification-launch 60.Hoffmann,M.,Boysel,S.,et al(2024).Generative AI and the Nature of Work.Harvard Business School Working Paper,No.25-021.https:/ 2).Will Artificial Intelligence Deliver on Its Promise in Healthcare?JAMA,331(1),6

237、5-69.https:/doi.org/10.1001/jama.2023.25054 62.Noyes,J.(2024,March 14).Perceptions of AI in healthcare:What professionals and the public think.Tebra.https:/ 63.Hermansen,A.(2024,October).An Open Architecture for Health Data Interoperability:How Open Source Can Help the Healthcare Sector Overcome the

238、 Information Dark Ages.The Linux Foundation.https:/www.linuxfoundation.org/research/health-data-interoperability 64.From code to cure,how Generative AI can reshape the health frontier:Unlocking new levels of efficiency,effectiveness,and innovation(2023).Deloitte.https:/ Global Spotlight Insights Rep

239、ort:The Role of Open Source in Uniting Innovation,Collaboration,and Resilience Across Regions and Industries.The Linux Foundation.https:/www.linuxfoundation.org/research/world-of-open-source-global-2024 THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|3666.Mondillo,G.,Colosimo,S.,Perrotta,A.,et

240、al(2025,January 28).Comparative Evaluation of Advanced AI Reasoning Models in Pediatric Clinical Decision Support:ChatGPT OI vs.DeepSeek-R1.medRxiv.https:/doi.org/10.1101/2025.01.27.25321169 67.Buckley,T.A.,Crowe,B.,Abdulnour,R-E.E.et al(2025,March 14).Comparison of Frontier Open-Source and Propriet

241、ary Large Language Models for Complex Diagnoses.JAMA Health Forum,6(3).doi:10.1001/jamahealthforum.2025.0040 68.Temsah,A.,Alhasan,K.,Altamimi,I.,et al(2025,February 18).DeepSeek in Healthcare:Revealing Opportunities and Steering Challenges of a New Open-Source Artificial Intelligence Frontier.Cureus

242、,17(2).DOI 10.7759/cureus.7922169.Herweijer,C.et al(2020).How AI can enable a sustainable future.PwC.https:/www.pwc.co.uk/sustainability-climate-change/assets/pdf/how-ai-can-enable-a-sustainable-future.pdf 70.DeClerq,D.et al(2024,October 24).Large language models can help boost food production,but b

243、e mindful of their risks.Frontiers in Artificial Intelligence,7.https:/doi.org/10.3389/frai.2024.1326153 71.Rowe,J.(2025,January 6).Delivering regenerative agriculture through digitalization and AI.World Economic Forum.https:/www.weforum.org/stories/2025/01/delivering-regenerative-agriculture-throug

244、h-digitalization-and-ai/72.Semios:Helping growers produce more sustainable,profitable crops(n.d.).Google Cloud.https:/ 73.Farmer.Chat.(n.d.)Digital Green.https:/farmerchat.digitalgreen.org/74.Generative AI in Construction Market(2021).Adroit Market Research.https:/ of Digital Adoption in the Constru

245、ction Industry 2025(2025,February).Deloitte and Autodesk.https:/ 76.Barbosa,F.,et al.(2023,September 19).AI:The next frontier of performance in industrial processing plants.McKinsey&Company.https:/ In Construction Market Size&Share Analysis-Growth Trends&Forecasts(2025-2030)(n.d.).Mordor Intelligenc

246、e.https:/ Briq Technology Suite(n.d.).Briq.https:/ 79.B.Hamdy,personal communication,April 28,2025.THE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|3780.AI Enablement on the Way to Smart Manufacturing:Deloitte Survey on AI Adoption in Manufacturing(2020).Deloitte.https:/ 81.Heimberger,H.et al(20

247、24,August 23).Exploring the factors driving AI adoption in production:a systematic literature review and future research agenda.Information Technology and Management.https:/doi.org/10.1007/s10799-024-00436-z 82.New IBM Study Data Reveals 74%of Energy&Utility Companies Surveyed Embracing AI(2024,Febr

248、uary 26).IBM.https:/ 83.AI to drive 165%increase in data center power demand by 2030(2025,February 4).Goldman Sachs.https:/ 84.Kemene,E.et al(2024,July 22).AI and energy:Will AI help reduce emissions or increase power demand?Heres what to know.World Economic Forum.https:/www.weforum.org/stories/2024

249、/07/generative-ai-energy-emissions/85.AI is set to drive surging electricity demand from data centres while offering the potential to transform how the energy sector works (2025,April 10).IEA.https:/www.iea.org/news/ai-is-set-to-drive-surging-electricity-demand-from-data-centres-while-offering-the-p

250、otential-to-transform-how-the-energy-sector-works86.Kann,S.(2025,March 6).A skeptics take on AI electricity load growth.Latitude Media.https:/ 29).Does DeepSeek call the data center boom into question?Latitude Media.https:/ al(2024,April).Advanced Research Directions on AI for Energy.Argonne Nationa

251、l Laboratory.https:/www.anl.gov/sites/www/files/2024-04/AI-for-Energy-Report_APRIL%202024.pdf 89.LF Energy(2025,January).Unlocking AIs Potential for the Energy Transition through Open Source.The Linux Foundation.https:/lfenergy.org/unlocking-ais-potential-for-the-energy-transition-through-open-sourc

252、e/90.Rozite,V.,et al(2023,November 2).Why AI and energy are the new power couple.IEA.https:/www.iea.org/commentaries/why-ai-and-energy-are-the-new-power-couple 91.A.Thornton,personal communication,April 25,2025.92.Groopman,J.,&Lindstrom,J.(2023,June).The Open Source Opportunity for Microgrids:Five W

253、ays to Drive Innovation and Overcome Market Barriers for Energy Resilience.The Linux Foundation.https:/www.linuxfoundation.org/research/open-source-opportunity-for-microgridsTHE ECONOMIC AND WORKFORCE IMPACTS OF OPEN SOURCE AI|38AcknowledgmentsThe authors would like to thank the subject matter exper

254、ts who advised on this project and provided data and feedback:Frank Nagle,Alex Thornton,Rick Justis,Matt White,Hilary Carter,and Adrienn Lawson.The authors would also like to thank the Linux Foundation Creative Services team for the production of the PDF.About the AuthorsAnna Hermansen is a Research

255、er and the Ecosystem Manager for Linux Foundation Research where she supports end-to-end management of the Linux Foundations research projects.She has conducted qualitative and systematic review research in health data infrastructure and the integration of new technologies to better support data sha

256、ring in healthcare,and has presented on this research work at conferences and working groups.Her interests lie at the intersection of health informatics,precision medicine,and data sharing.She is a generalist with experience in client services,program delivery,project management,and writing for acad

257、emic,corporate,and web user audiences.Prior to the Linux Foundation,she worked for two different research programs,the Blockchain Research Institute and BC Cancers Research Institute.She received her Master of Science in Public Health and a Bachelor of Arts in International Relations,both from the U

258、niversity of British Columbia.Cailean Osborne,PhD is a Senior Researcher at the Linux Foundation,where he co-develops and advocates tools for promoting openness in AI.He has a PhD in Social Data Science from the University of Oxford,where he researched inter-company collaboration dynamics in the ope

259、n source AI ecosystem.Previously,he worked in AI policy at the UK Government,where he co-authored the UKs National AI Strategy and served as a UK Government Delegate at the Global Partnership on AI,among others.He is based in Berlin,Germany.Copyright 2025 The Linux FoundationThis report is licensed

260、under the Creative Commons Attribution-NoDerivatives 4.0 International Public License.To reference this work,please cite as follows:Anna Hermansen and Cailean Osborne,“The Economic and Workforce Impacts of Open Source AI:Insights from Industry,Academia,and Open Source Research Publications,”The Linu

261、x Foundation,May 2025.Founded in 2021,Linux Foundation Research explores the growing scale of open source collaboration,providing insight into emerging technology trends,best practices,and the global impact of open source projects.Through leveraging project databases and networks,and a commitment to best practices in quantitative and qualitative methodologies,Linux Foundation Research is creating the go-to library for open source insights for the benefit of organizations the world over.Commissioned byAll dollar figures in this report are in USD,unless noted