| 594 | 0 | 687 |
| 下载次数 | 被引频次 | 阅读次数 |
在生成式人工智能、大模型训练和产业智能化加速推进的背景下，算力已由一般性技术投入上升为影响企业成长边界、产业组织方式与资本配置结构的关键生产要素，算力换股权也逐步成为连接技术供给、资本投入与产业协同的重要制度安排。本文在不完全契约和融资约束框架下构建理论模型，比较债务融资、普通股权融资与算力换股权的差异。演绎逻辑展示，算力换股权的形成源于要素属性跃升、融资结构调适、契约关系重塑和制度环境催化，能够在一定条件下缓解现金流约束、保留需求方研发激励，并激励供给方提高交付、适配、运维和扩容努力。然而，随着算力资本化、生态协同和金融嵌套不断加深，定价锚点不稳、平台锁定强化、估值泡沫累积以及跨境监管与合规责任不清等矛盾开始集中显现，成为算力换股权功能发挥过程中亟待解决的关键问题。基于此，本文提出四个方面的解题思路：完善会计确认、估值审计和权益工具法律规则，优化开放协同与竞争治理结构，健全信息披露和金融监管安排，并加强国内制度规则与跨境监管框架的衔接。循此，算力换股权才能真正转化为促进科技创新、产业升级与经济高质量发展的重要支点。
Abstract：Against the backdrop of accelerated development in generative artificial intelligence,large-scale model training,and industrial intelligence,computing power has evolved from a general technological input into a key production factor that shapes firm growth boundaries,industrial organization,and capital allocation structures.In this context,computing power for equity has gradually become an important institutional arrangement linking technology supply,capital investment,and industrial coordination.This paper develops a theoretical model based on incomplete contracts and financing constraints to compare debt financing,ordinary equity financing,and compute-for-equity.The model shows that,under certain conditions,compute-for-equity can ease cash flow constraints,preserve the R&D incentives of compute demanders,and encourage compute providers to exert greater effort in delivery,adaptation,operation and maintenance,and capacity expansion.Its emergence is driven by the upgrading of compute as a production factor,maturity mismatch in financing,the reshaping of contractual relations,and an enabling institutional environment.However,as the capitalization of computing power,ecosystem coordination,and financial nesting continue to deepen,tensions such as unstable pricing anchors,strengthened platform lock-in,accumulated valuation bubbles,and unclear cross-border regulatory and compliance responsibilities have become increasingly prominent.Accordingly,four pathways for improvement are proposed:improving accounting recognition,valuation auditing,and legal rules for equity instruments;optimizing open collaboration and competition governance;strengthening information disclosure and financial regulation;and enhancing coordination between domestic institutional rules and cross-border regulatory frameworks.Following these pathways,computing power for equity can be transformed into an important fulcrum for advancing technological innovation,industrial upgrading,and high-quality economic development.
国家发展改革委、国家数据局、工业和信息化部.国家数据基础设施建设指引[EB/OL].2025-01-06.https://www.nda.gov.cn/sjj/zwgk/zcfb/0106/20250106095112713400492_pc.html.
国家发展改革委员会.“东数西算”工程系列解读之一|“东数西算”工程助力我国全面推进算力基础设施化[EB/OL].2022-03-07.https://www.ndrc.gov.cn/xxgk/jd/jd/202203/t20220317_1319467.html.
侯佳忆,王浩宇,何德旭.货币政策对企业技术创新的影响——基于不同融资结构的分析[J].经济与管理,2025(1).
马涛,钟化鑫.超大规模算力建设与“十五五”时期重大生产力布局优化[J].改革,2026(3).
四川数字资产交易中心.成都智算中心[EB/OL].https://www.sdex.com.cn/computinginfo/747408183353409676.html.
谢丹夏,魏文石,李尧.数据与算力经济:一个数字经济的一般均衡理论[J].管理世界,2026(3).
新华网.国家数据局发布《数字中国发展报告(2024年)》[EB/OL].2025-04-30.https://www.news.cn/digital/20250430/2b7f40dbc70d4ab39303a9b3b7705 8ae/c.html.
徐浩宇.人工智能算力互联互通:生成逻辑、制度梗阻及其分层治理[J/OL].西安交通大学学报(社会科学版),1-18[2026-05-10].
徐浩宇.行政规制与市场规则:算力资源配置的双阶法治构造[J].学习与实践,2025(11).
许诺,毛聚,毛新述,王彦超.算力部署、数据跨域流动与企业全要素生产率——来自智算中心的证据[J].中国工业经济,2025(4).
于潇宇.从“东数西算”到“东数西训”——新形势下中国算力布局的演进逻辑与优化路径[J].宏观经济研究,2025(12).
余东华.算力:数字经济时代的新质生产力[J].财贸研究,2024(7).
张红,叶兆驰.论算力作为一种新型财产权利[J].广东社会科学,2025(2).
卓乎知芯.600亿美元,5年长约!AMD与Meta达成超大型AI芯片供应协议[EB/OL].2026-02-25.https://mp.weixin.qq.com/s?__biz=Mzg5MzkzNzM1OA%253D%253D&mid=2247496245&idx=1&sn=e711e7828aaf761b6fe7e65e1433b06e&chksm=c17bb755748809296e16028b84896a149159dbb17c2b58e6c192900698ec73dcfda328fa765 8&scene=0&xtrack=1%23rd.
Acemoglu D.,A.Ozdaglar,and A.Tahbaz-Salehi,2015.Systemic Risk and Stability in Financial Networks.American Economic Review,105(2):564-608.
Advanced Micro Devices,Inc.AMD and Meta Announce Expanded Strategic Partnership to Deploy 6 Gigawatts of AMD GPUs[EB/OL].2026-02-24.https://www.amd.com/en/newsroom/press-releases/2026-2-24-amdand-meta-announce-expanded-strategic-partnersh.html.
Armbrust M.,A.Fox,R.Griffith,A.D.Joseph,R.Katz,A.Konwinski,G.Lee,D.Patterson,A.Rabkin,I.Stoica,and M.Zaharia,2010.A View of Cloud Computing.Communications of the ACM,53(4):50-58.
Bakkar,Y.,Z.Dhifaoui,S.A.M.I.Ben Jabeur,and J.Pillot,2025.Artificial Intelligence Narratives and Stock Market Returns,SSRN Working Paper No.5407132.
Brynjolfsson E.,P.Hofmann,and J.Jordan,2010.Cloud Computing and Electricity:Beyond the Utility Model.Communications of the ACM,53(5):32-34.
Data Center Dynamics.Nvidia data center GPU shipments totaled 3.76m in 2023,equating to a 98%marketshare-report[EB/OL].2024-06-12.https://www.datacenterdynamics.com/en/news/nvidia-gpu-shipments-totaled-376m-in-2023-equating-to-a-98-market-share-report.
Eisfeldt A.L.,G.Schubert,and M.B.Zhang,2023.Generative AI and Firm Values,NBER Working Paper.
Gartner.Gartner Forecasts Worldwide GenAI Spending to Reach644 Billion in 2025[EB/OL].2025-03-31.https://www.gartner.com/en/newsroom/press-releases/2025-03-31-gartner-forecasts-worldwide-genaispending-to-reach-644-billion-in-2025.
Goldfarb A.,and C.Tucker,2019.Digital Economics.Journal of Economic Literature,57(1):3-43.
Grossman S.J.,and O.Hart,1986.The Costs and Benefits of Ownership:A Theory of Vertical and Lateral Integration.Journal of Political Economy,94(4):691-719.
Holmstrom B,and J.Tirole,1997.Financial Intermediation,Loanable Funds,and the Real Sector.The Quarterly Journal of Economics,1997,112(3):663-691.
Holmstrom B.,1982.Moral Hazard in Teams.The Bell Journal of Economics,13(2):324-340.
Holmstr?m B.,1979.Moral Hazard and Observability.The Bell Journal of Economics,10(1):74-91.
IDC.Worldwide Server Market Value Will Reach 366 billion in 2025 Representing a 45%Growth Compared to2024,according to IDC[EB/OL].2025-06-26.https://my.idc.com/getdoc.jsp?containerld=prUS53643125.
Jones E.,2023.Digital Disruption:Artificial Intelligence and International Trade Policy.Oxford Review of Economic Policy,39(1):70-84.
Kalyani A.,N.Bloom,M.Carvalho,T.Hassan,J.Lerner,and A.Tahoun,2025.The Diffusion of New Technologies.The Quarterly Journal of Economics,140(2):1299-1365.
Klotz A.Nvidia's H100 AI GPU shortages ease as lead times drop from up to four months to 8-12 weeks[EB/OL].Tom's Hardware,2024-04-10.https://www.tomshardware.com/pc-components/gpus/nvidias-h100-ai-gpushortages-ease-as-lead-times-drop-from-up-to-four-months-t0-8-12-weeks.
Korinek A.,and J.Vipra,2025.Concentrating Intelligence:Scaling and Market Structure in Artificial Intelligence.Economic Policy,40(121):225-256.
Ljungqvist A.,and M.P.Richardson,2003.The Cash Flow,Return and Risk Characteristics of Private Equity,Working Paper.
McElheran K.,M.J.Yang,Z.Kroff,and E.Brynjolfsson,2025.The Rise of Industrial AI in America:Microfoundations of the Productivity J-Curve.US Census Bureau,Center for Economic Studies Working Paper.
OpenAI,2025a.“星际之门”(Stargate)项目正式启动[EB/OL].2025-01-21.https://openai.com/zhHans-CN/index/announcing-the-stargate-project.
OpenAI,2025b.OpenAI、甲骨文与软银为扩展星际之门(Stargate)计划宣布新增五处AI数据中心[EB/OL].2025-09-23.https://openai.com/zh-Hans-CN/index/five-new-stargate-sites.
OpenAI,2025c.AMD and OpenAI announce strategic partnership to deploy 6 gigawatts of AMD GPUs[EB/OL].2025-10-06.https://openai.com/index/openai-amd-strategic-partnership.
Reuters,2025a.OpenAI expects business to burn115 billion through 2029,The Information reports[EB/OL].2025-09-06.https://www.reuters.com/technology/openai-expects-business-burn-115-billion-through-2029-information-reports-2025-09-06.
Reuters,2025 b.AMD signs AI chip-supply deal with Ope nAI,shares surge over 34%[EB/OL].2025-10-06.https://www.reuters.com/business/amd-signs-ai-chip-supply-deal-with-openai-gives-it-optiontake-10-stake-2025-10-06.
Reuters,2025c.CoreWeave signs14 billion AI infrastructure deal with Meta[EB/OL].2025-09-30.https://www.reuters.com/technology/core weave-signs-14-billion-ai-deal-with-meta-bloomberg-news-reports-2025-09-30.
S&P Global Ratings.Presale:QTS Issuer ABS I LLC And QTS Co-Issuer A[EB/OL].2025-05-28.https://www.spglobal.com/ratings/en/regulatory/article/-/view/sourceld/13496042.
Stiglitz J.E.,and A.Weiss,1981.Credit Rationing in Markets with Imperfect Information.The American Economic Review,71(3):393-410.
Synergy Research Group.GenAI Helps Drive Quarterly Cloud Revenues to119 Billion as Growth Rate Jumped Yet Again in Q4[EB/OL].2026-02-05.https://www.srgresearch.com/articles/genai-helps-drive-quarterlycloud-revenues-to-119-billion-as-growth-rate-jumped-yet-again-in-q4.
(1)注：关系专用性安排是指算力供给方围绕特定客户的大规模、连续性和定制化算力需求所形成的专用化资源配置、技术适配和履约保障安排。具体包括：一是资源配置安排，即为特定客户预留或优先配置高端芯片、算力容量和相关基础设施资源；二是技术适配安排，即围绕客户模型架构、训练框架、推理场景和系统环境进行软件栈优化、系统集成和机架级部署适配；三是履约保障安排，即提供持续运维、性能调优、故障响应、弹性扩容和供应链协调等支持。上述安排依赖特定客户的技术路线、部署环境和商业化节奏，通常难以在不发生额外转换成本、效率损失或机会成本的情况下完全转用于其他客户，因而具有资产专用性和关系专用性特征。
基本信息:
中图分类号:F49;F832
引用信息:
[1]胡海峰,刘斯诺.人工智能时代算力要素金融创新研究——基于算力换股权现象的探讨[J].金融经济学研究().
基金信息:
国家社科基金重大项目(21&ZD111)
2026-07-23
2026-07-23
2026-07-23