人工智能开放的好处：经合组织七国集团讨论文件

Benefits of AI openness OECD discussion paper for the G7 This work is published under the responsibility of the Secretary-General of the OECD. The opinionsexpressed and arguments employed This document was prepared by the Organisation for Economic Co-operation and Development (OECD) at the request of the 2026G7Presidency ofFrance.It was drafted by the OECDDirectorate for Science, Technology and Innovation (STI).The paper aims to inform This document, as well as any data and map included herein, are without prejudice tothe status of or sovereignty over any territory, to the The statistical data for Israel are supplied by and under the responsibility of the relevant Israeli authorities. The use ofsuch data by the OECD Cover image: ©Gorodenkoff/Shutterstock ©OECD2026 Attribution4.0 International (CCBY4.0) This work is made available under the Creative Commons Attribution4.0International licence. By using this work, you accept to be bound by the terms of this licence(https://creativecommons.org/licenses/by/4.0/). Attribution–you must cite the work. Adaptations–you must cite the original work and add the following text:This is an adaptation of an original work by the OECD. The opinions expressed and arguments employed Third-party material–the licence does not apply to third-party material in the work. If using such material, you are responsible for obtaining permission from the third party and Acknowledgements Thispaperwas written by Jeff Mollins andLuis Arandafrom theOECDDirectorate for Science, Technologyand Innovation (STI).Strategic direction was provided by Jerry Sheehan,Audrey Plonk, Sara Rendtorff-Smith and Galia Daor from theOECD Directorate for Science, Technology and Innovation.The authors The authorsare also gratefultothose who have contributed to the report throughout its development.ThisincludesVincent Bataille,Manuel Betin,Celine Caira,Julia Carro, Andreia Furtado,Peter Gal, DanielHaerle andBenedicte Rispalfrom the OECD Secretariat,as well asJesse Dunietz (USNIST)and FrankNagle (MIT).The authorsalsowish to thank members oftheOECD.AIExpert Group on AICompute and Table of contents Acknowledgements Executive summary6 1 Introduction 2 Analytical framework: Openness across the AI stack 3 Economic benefits of AI openness 12 Micro-level effects: Firm-level gains in cost efficiency and performance from AI opennessQuality-to-price ratio of open-vs closed-weight modelsOpen-weight models are increasingly competitive at the AI economic frontierBreak-even analysis: When self‑hosting open‑weight models becomes cost-effectiveMacro-level effects: Indicative impacts of open‑source AI code on economic growthMain finding 4 AI openness as a driver of stronger and more resilient national AI ecosystems20 Local value creation and downstream innovationTechnological sovereignty and strategic autonomySector-specific adaptation 202222 5 Conclusion 24 Annex A. Break-even analysis assumptions 25 Annex B. Methodology for macro-level analysis27 References33 Tables Table2.1. Illustrative extended mapping of the AI stack and associated openness considerationsTable3.1. Token usage scenariosTable3.2. Estimated fixed CapEx costs for private hosting (USD)Table3.3. Breakeven months for cloud API vs private GPU installation scenariosTable3.4. Impact of AI open-source activity on GDPTableB.1. Sources and descriptionsTableB.2. Impact of domestic and foreign AI-related open-source activity on GDP Figures Figure2.1. Illustrative high-level view of the AI stack9Figure3.1. Cloud-provided open‑weight text models maintain stronger value‑for‑money asthey age13Figure3.2. Cloud-provided open‑weight models represented approximately 60% of models at the economicfrontier from January 2024 to April 202514Figure3.3. Open-source AI codecontributions have expanded rapidly17Figure4.1. Open datasets drive faster creation of models with open-source licenses21Figure4.2. Models with open-source license diffuse faster: they make up about 65% of all models with licence Executivesummary Recent advances in artificial intelligence (AI), particularly the rapid development of foundation models,have renewed policy attention to the role ofopenness across the AI ecosystem. AI openness affectsmultiple layers of the AI stack, from infrastructure and data to models and applications, as well ascross-cutting enablers such as governance, skills, and investment. Its implications extend beyond technical This report analyses the economic and strategic benefits of AI openness, drawing on OECD data andthird-party research. It focuses on bothmicro-level impacts(performance, costs, and competitiveness)andmacro-level outcomes(economic growth and knowledge diffusion), as well as the role of openness At themicroeconomiclevel, evidence suggests that open-weight AI models are becoming increasinglycompetitivewith proprietary alternatives.While open models achieve approximately 90%of the Cost dynamics vary by scale. For smaller workloads, cloud-bas