Bibliography: AI Capital Expenditure ROI Measurement

Date: November 2025 Scope: Academic papers, industry reports, financial disclosures, and central bank analyses

Academic Papers

Productivity and Economic Impact

Acemoglu, D. (2024). The Simple Macroeconomics of AI. MIT Economics. - URL: https://economics.mit.edu/sites/default/files/2024-05/The%20Simple%20Macroeconomics%20of%20AI.pdf - Key Contribution: Task-based model estimating 0.5-1.6% GDP impact over 10 years; only 4.6% of tasks exposed to AI - Methodology: Quantitative framework: AI Impact = (Fraction of Tasks Exposed) × (Cost Savings per Task) × (GE Multiplier)

Brynjolfsson, E., Li, D., & Raymond, L. R. (2023). Generative AI at Work (NBER Working Paper No. 31161). National Bureau of Economic Research. - DOI: https://www.nber.org/papers/w31161 - Key Finding: 14% productivity gain for customer service agents; 34% improvement for novice workers - Methodology: Randomized controlled trial with ~5,000 workers at Fortune 500 company

Brynjolfsson, E., Rock, D., & Syverson, C. (2017). Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics (NBER Working Paper No. 24001). - DOI: https://www.nber.org/papers/w24001 - Key Contribution: Productivity J-curve framework for general purpose technologies; implementation lags explain paradox - Application: Justifies 5-10 year timeline for AI infrastructure ROI

McElheran, K., Yang, M., Brynjolfsson, E., & Kroff, Z. (2025). The Rise of Industrial AI in America: Microfoundations of the Productivity J-curve(s) (Census Working Paper CES-WP-25-27). U.S. Census Bureau. - URL: https://www2.census.gov/library/working-papers/2025/adrm/ces/CES-WP-25-27.pdf - Key Finding: AI adoption initially reduced productivity by 1.33 pp on average; 60 pp when correcting for selection bias - Methodology: Causal analysis of U.S. manufacturing firms using Census microdata

METR (2025). Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity. - URL: https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/ - Key Finding: Experienced developers using AI took 19% longer than without AI - Controversy: Contradicts most productivity claims; highlights measurement challenges

Infrastructure Investment and Distributed Lags

Bom, P. R., & Ligthart, J. E. (2014). What Have We Learned From Three Decades Of Research On The Productivity Of Public Capital? Journal of Economic Surveys, 28(5), 889-916. - DOI: https://ideas.repec.org/a/bla/jecsur/v28y2014i5p889-916.html - Key Finding: Meta-analysis of 578 estimates; short-run output elasticity 0.083, long-run 0.122 - Application: Establishes 5-10 year lag for infrastructure productivity gains

Kumar, R. (2024). Infrastructure and Its Impact on Economic Growth Using the Autoregressive Distributed Lag (ARDL) Model: India as a Case Study. International Journal of Economics and Finance. - URL: https://www.researchgate.net/publication/390244088 - Key Finding: Infrastructure road investment affects GDP by 0.41% with 3-year lag; land infrastructure by 1.51% with 3-year lag - Methodology: ARDL model with quarterly data 2000-2020

El Alaoui, A. (2021). Transport Infrastructure and Economic Growth: An Econometric Analysis Using the Autoregressive Distributed Lag Model ARDL. Research in Transportation Economics. - URL: https://www.researchgate.net/publication/350567819 - Key Finding: Positive effects on economic growth in both short and long term; distributed lag structure validated

Investment Bank Research Reports

Goldman Sachs

Goldman Sachs Research (2024, June). Gen AI: Too Much Spend, Too Little Benefit? Top of Mind, Issue 129. - URL: https://www.goldmansachs.com/insights/top-of-mind/gen-ai-too-much-spend-too-little-benefit - PDF: https://www.goldmansachs.com/intelligence/pages/gs-research/gen-ai-too-much-spend-too-little-benefit/report.pdf - Key Finding: ~$1T AI infrastructure spending with “little to show for it” beyond efficiency gains - Expert Views: Acemoglu (pessimistic: 0.5% productivity) vs. Briggs/GS (optimistic: 15% productivity) - Infrastructure Concern: Chip supply and power constraints could limit growth

Goldman Sachs Research (2024). Will the $1 Trillion of Generative AI Investment Pay Off? - URL: https://www.goldmansachs.com/insights/articles/will-the-1-trillion-of-generative-ai-investment-pay-off - Analysis: Investment scale justification, technology adoption timeline, infrastructure requirements

Morgan Stanley

Morgan Stanley Research (2024). GenAI Revenue Growth and Profitability. - URL: https://www.morganstanley.com/insights/articles/genai-revenue-growth-and-profitability - Key Projection: GenAI starts generating profits in 2025 (34% margin estimated) - Revenue Forecast: $153B (2025) → $1.1T (2028), 20× increase - Infrastructure Investment: Hardware/networking $98B (2024) → $276B (2028)

Mauboussin, M. J., & Callahan, D. (2014). Measuring the Moat: Assessing the Magnitude and Sustainability of Value Creation. Morgan Stanley Counterpoint Global. - URL: https://www.shareholderforum.com/returns/Library/20140604_Mauboussin-Callahan.pdf - Key Contribution: ROIC decomposition framework; Return on Incremental Invested Capital (ROIIC) methodology - Application: Evaluating return on new investment programs (AI capex) vs. average ROIC

Citigroup

Citigroup Research (2025, September). Big Tech AI Spending Forecast Through 2029. - Coverage: Bloomberg, Reuters, BNN Bloomberg - URL: https://www.bnnbloomberg.ca/business/2025/09/30/citigroup-forecasts-big-techs-ai-spending-to-cross-us28-trillion-by-2029/ - Key Projection: $2.8T total AI infrastructure spending through 2029 (revised up from $2.3T) - Power Requirements: 55 GW new capacity needed by 2030; $2.8T global power infrastructure spend ($1.4T U.S.) - Cost per GW: ~$50B

Sequoia Capital

Cahn, D. (2024, June). AI’s $600B Question. Sequoia Capital Blog. - URL: https://www.sequoiacap.com/article/ais-600b-question/ - Key Argument: Infrastructure investment requires ~$600B revenue to justify; current GenAI revenue ~$100B - Gap: $500B revenue shortfall raises ROI sustainability questions - Counterargument: Natural 5-10 year time lag between infrastructure and application revenue

Central Bank and Financial Stability Reports

International Monetary Fund (IMF)

Georgieva, K. (2024, October). Remarks at Milken Institute Global Conference. International Monetary Fund. - Coverage: CNBC, Al Jazeera - URL: https://www.cnbc.com/2025/10/09/imf-and-bank-of-england-join-growing-chorus-warning-of-an-ai-bubble.html - Key Statement: “Uncertainty is the new normal” — “buckle up” - Warning: Current stock valuations “heading toward levels we saw during the bullishness about the internet 25 years ago” - Risk: If sharp correction occurs, “tighter financial conditions could drag down world growth”

Bank of England

Bank of England (2024, October). Financial Stability Report. - URL: https://www.bankofengland.co.uk/ - Coverage: The Guardian, Financial Times - URL (News): https://www.theguardian.com/business/2025/oct/08/bank-of-england-warns-of-growing-risk-that-ai-bubble-could-burst - Key Finding: “The risk of a sharp market correction has increased” - Valuation Analysis: “Equity market valuations appear stretched, particularly for technology companies focused on AI” — comparable to 2000 dot-com peak - Systemic Risk: High concentration in small cluster of AI-heavy companies; decline could ripple globally

Bank for International Settlements (BIS)

Bank for International Settlements (2024). Annual Economic Report 2024. - URL: https://www.bis.org/

Bank for International Settlements (2024). Intelligent Financial System: How AI is Transforming Finance (BIS Working Paper No. 1194). - URL: https://www.bis.org/publ/work1194.pdf - Key Points: AI will affect financial systems as well as productivity, consumption, investment, and labor markets - Risk Factors: Poor governance, opaque decision-making, overreliance on third-party providers - Measurement Gap: “Lack of globally accepted definition of AI prevents better understanding of AI use cases”

Industry Research and Forecasts

Gartner

Gartner (2024, July). Gartner Forecasts Worldwide IT Spending to Grow 7.5% in 2024 [Press Release]. - URL: https://www.gartner.com/en/newsroom/press-releases/2024-07-16-gartner-forecasts-worldwide-it-spending-to-grow-7-point-5-percent-in-2024

Gartner (2025, March). Gartner Forecasts Worldwide GenAI Spending to Reach $644 Billion in 2025 [Press Release]. - URL: https://www.gartner.com/en/newsroom/press-releases/2025-03-31-gartner-forecasts-worldwide-genai-spending-to-reach-644-billion-in-2025 - Key Projection: $644B GenAI spending in 2025 - Growth Driver: Enterprise adoption scaling from 11% (2024) to higher penetration

IDC (International Data Corporation)

IDC (2024). Worldwide Artificial Intelligence IT Spending Forecast, 2024–2028. - URL: https://my.idc.com/getdoc.jsp?containerId=US52635424

IDC (2024, August). IDC’s Worldwide AI and Generative AI Spending — Industry Outlook. - URL: https://blogs.idc.com/2024/08/21/idcs-worldwide-ai-and-generative-ai-spending-industry-outlook/ - Key Finding: $235B AI spending (2024) → $632B (2028) - ROI Estimate: $3.70 return per $1 invested (average); top 5% achieve $10 per $1

McKinsey & Company

McKinsey & Company (2024). The Economic Potential of Generative AI: The Next Productivity Frontier. - URL: https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier - Key Estimate: GenAI could add $2.6T to $4.4T annually to global economy - Impact: Equivalent to adding entire UK economy annually

McKinsey & Company (2024). The State of AI in Early 2024: Gen AI Adoption Spikes and Starts to Generate Value. - URL: https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai-2024 - Key Finding: 65% of organizations regularly use GenAI (up from 33% in 2023) - Value Realization: 46 out of 876 respondents (high performers) attribute >10% of EBIT to GenAI deployment - Challenge: 97% struggle to demonstrate GenAI business value

Bloom Energy

Bloom Energy (2025). 2025 Data Center Power Report. - URL: https://www.bloomenergy.com/news/data-centers-are-turning-to-onsite-power-sources-to-address-35-gw-energy-gap-by-2030/ - Key Finding: 84% of data center operators rank power availability among top 3 site selection considerations - Bottleneck: Power availability is “the major bottleneck” for data center growth - Projection: 38% of facilities expected to use onsite generation by 2030 (vs. 13% in 2024); 27% fully powered onsite (vs. 1%)

Dell’Oro Group

Dell’Oro Group (2024). Hyperscale Capex Surges 82 Percent in 3Q 2024. - URL: https://www.delloro.com/news/hyperscale-capex-surges-82-percent-in-3q-2024-fueled-by-ai-infrastructure-spending/ - Key Data: Q3 2024 top 4 hyperscalers spent $58.9B capex (63% YoY growth) - Capex-to-Revenue: 22% in 2024 vs. historical 11-16%

Stanford HAI

Stanford HAI (2024). The 2024 AI Index Report. - URL: https://aiindex.stanford.edu/report/ - Scope: Comprehensive annual analysis of AI trends, benchmarks, adoption metrics

Financial Methodology

ROIC and Valuation

Damodaran, A. (n.d.). Return on Capital (ROC), Return on Invested Capital (ROIC) [Working Paper]. NYU Stern School of Business. - URL: https://pages.stern.nyu.edu/~adamodar/pdfiles/papers/returnmeasures.pdf - Key Contribution: Definitive treatment of ROIC calculation; NOPAT definition; invested capital adjustments - Application: Foundation for AI-specific ROIC decomposition

Mauboussin, M. J., & Callahan, D. (2014). Measuring the Moat: Assessing the Magnitude and Sustainability of Value Creation. Morgan Stanley. - Key Formula: ROIIC = Δ NOPAT / Δ Invested Capital - Application: More practical for evaluating new investment programs (AI capex) than average ROIC

Valuation Frameworks

Koller, T., Goedhart, M., & Wessels, D. (2020). Valuation: Measuring and Managing the Value of Companies (7th ed.). Wiley Finance. - ISBN: 978-1-119-61088-5 - Key Contribution: McKinsey DCF methodology; ROIC-driven valuation; terminal value formulas - Application: Foundation for conservation-consistent terminal multiples

Damodaran, A. (2012). Investment Valuation: Tools and Techniques for Determining the Value of Any Asset (3rd ed.). Wiley Finance. - ISBN: 978-1-118-20656-0 - Key Contribution: Comprehensive valuation methods; cost of capital estimation; growth assumptions

Company Financial Disclosures

Microsoft

Microsoft Corporation (2024). 2024 Annual Report. - URL: https://www.microsoft.com/investor/reports/ar24/ - Key Disclosure: FY2024 capex $48.4B; Azure AI business $13B annual run rate (Q2 FY2025, 175% YoY growth) - Segments: Intelligent Cloud revenue $75B+ (34% growth)

Alphabet (Google)

Alphabet Inc. (2024). Form 10-K for fiscal year ended December 31, 2024. U.S. Securities and Exchange Commission. - URL: https://www.sec.gov/edgar/ - Key Disclosure: Google Cloud $50B+ annual run rate; “billions in revenue” from AI (CEO Thomas Kurian) - Capex: $24B Q4 2024 → $62.6B projected 2025

Amazon

Amazon.com, Inc. (2024). Form 10-K for fiscal year ended December 31, 2024. U.S. Securities and Exchange Commission. - URL: https://www.sec.gov/edgar/ - Key Disclosure: AWS Q3 2025 revenue $33B (+20% YoY, fastest pace since 2022); 3.8 GW capacity added in 12 months - Capex Projection: $125B (2025)

Nvidia

Nvidia Corporation (2024). Q2 FY2025 Earnings Call Transcript. - URL: https://www.nvidia.com/en-us/ - Key Claims: “Not unusual to save 90% of computing cost” via accelerated computing; “best ROI computing infrastructure investment” - Case Studies: Commonwealth Bank (640× performance, 80% cost reduction), AT&T (3.3× faster, 60% lower cost)

Technical Resources

GPU Utilization and Performance

Nvidia Corporation (2024). Sustainable Strides: How AI and Accelerated Computing Are Driving Energy Efficiency. - URL: https://blogs.nvidia.com/blog/accelerated-ai-energy-efficiency/ - Key Claims: 5× average energy efficiency increase for apps accelerated with A100 GPUs (NERSC study)

Nvidia Developer Blog. Understanding GPU Performance: Utilization vs. Saturation. - URL: https://developer.nvidia.com/blog/ - Key Insight: nvidia-smi GPU utilization metric is misleading; SM efficiency provides more accurate picture

Data Center Industry

Equinix (2024). Form 10-K. U.S. Securities and Exchange Commission. - Key Disclosure: Risk factors cite “availability of power, increased costs to procure power… challenges of constructing data centers” - Application: Material risk disclosure example

News Coverage and Analysis

Financial Times

Financial Times (2024). Wall Street’s AI ‘Bubble’ Echoes Dotcom Excesses, Ray Dalio Warns. - URL: https://www.ft.com/content/eef8dbc9-bd04-4502-bdc2-1092aa4251b2 - Key Quote: Dalio warns of valuation parallels to 2000 dot-com bubble

Axios

Axios (2025, September). Behind the Curtain: Slow, Hard AI. - URL: https://www.axios.com/2025/09/09/behind-the-curtain-slow-hard-ai - Key Finding: MIT study context (disputed: “95% see zero ROI”); measurement challenges pervasive

Reuters

Reuters (2025, September). Citigroup Forecasts Big Tech’s AI Spending to Cross $2.8 Trillion by 2029. - URL: https://www.reuters.com/world/china/citigroup-forecasts-big-techs-ai-spending-cross-28-trillion-by-2029-2025-09-30/

Investopedia

Investopedia (2024). Meta Says Its Numbers Show that AI Spending Is Paying Off. - URL: https://www.investopedia.com/meta-says-its-numbers-show-that-ai-spending-is-paying-off-11726660 - Key Data: Ad efficiency improvements, engagement rate lifts attributed to AI

CNBC

CNBC (2025, October). IMF and Bank of England Join Growing Chorus Warning of an AI Bubble. - URL: https://www.cnbc.com/2025/10/09/imf-and-bank-of-england-join-growing-chorus-warning-of-an-ai-bubble.html

Measurement Frameworks and Surveys

IBM Research

IBM (2024). AI ROI Research. (Referenced in industry analyses) - Key Finding: Average enterprise AI ROI: 5.9%; high-performing companies: 13% ROI

Forrester Research

Forrester (2024). Q2 2024 GenAI Survey. - Key Finding: 49% of GenAI decision-makers expect ROI in 1-3 years; 44% in 3-5 years

Federal Reserve

Federal Reserve (2024). Workers Using GenAI Survey. - Key Finding: Workers using GenAI reported saving 5.4% of work hours (implying 1.1% workforce productivity increase)

Circular Financing Analysis

Fortune, Seeking Alpha, Harvard Business Review (October 2024). AI Circular Financing Concerns. - Key Examples: Nvidia $100B OpenAI investment; AMD partnership with OpenAI (160M share warrants for 6 GW GPU commitment) - Parallel: Cisco’s vendor financing in 1999 dot-com era - Concern: May give inflated perception of true AI demand

Total Sources: 60+ Academic Papers: 12 Investment Bank Reports: 8 Central Bank/Financial Stability: 4 Industry Research: 10 Company Filings: 5 Technical Resources: 6 News Coverage: 15+

Compilation Date: November 2025 Framework Version: AI ROI Extension v1.0