Market attention on AI investment is excessively focused on hardware, while "non-hardware investments" such as data infrastructure, organizational redesign, and workforce restructuring are rapidly increasing, with their scale and impact now impossible to ignore.
Analysts from Goldman Sachs Global Economics Research, including Joseph Briggs, have released a new report stating that U.S. AI-related hardware capital expenditure has already reached $360 billion (1.1% of GDP). Global hyperscale cloud provider capital expenditure is expected to reach $400 billion in 2025 and exceed $700 billion in 2026. However, this is only half the story.
The analysts calculate that current annual U.S. corporate AI-related labor costs are approximately $153 billion, and the time investment by executives in AI organizational transformation is equivalent to about $40 billion annually. Historical patterns indicate that for every additional dollar of hardware investment, it drives two dollars of intangible capital investment. Based on this, they project that global AI non-hardware investment could surpass $1 trillion in the coming years.
**How Large is Non-Hardware Investment? An Assessment from Four Dimensions**
The analysts estimate the current scale of non-hardware AI investment from four perspectives.
First, IT labor costs. Over the past two to three years, the proportion of AI-related IT positions has risen to about 25%, aligning with survey data showing "20% to 40% of IT budgets allocated to AI projects." Based on this, the analysts estimate that non-tech sector companies' annualized investment in AI tool development through internal IT teams is around $153 billion.
Second, executive time costs. Surveys show that about one-third of performance evaluations and compensation for executives at large firms are linked to AI strategy, and AI has consistently been the top priority for U.S. corporate leaders. Using the Corrado Hulten Sichel (2005) methodology and assuming 20% of executive time is spent on organizational innovation, with 35% of that focused on AI, the analysts estimate the annualized scale of AI-related organizational capital investment exceeds $40 billion (based on a total U.S. executive compensation pool of roughly $600 billion).
Third, workforce restructuring costs. The analysts compiled AI-driven layoff data from companies including Block, Atlassian, HP, Oracle, Accenture, Salesforce, Chegg, and C3.AI, finding an average restructuring cost of about $84,000 per affected employee.
Currently, AI's impact on the job market remains limited. Goldman Sachs estimates AI-related hiring friction reduces monthly employment growth by about 10,000, corresponding to current annual labor restructuring investment of roughly $10 billion. However, extrapolating from the analysts' prior prediction that AI could displace 6% to 7% of the workforce, total U.S. corporate spending on labor restructuring over the entire AI adoption cycle would reach $800 to $900 billion. If spread over a decade, this equates to about $90 billion annually.
Fourth, historical statistical patterns. Using the EU KLEMS database for regression analysis, the analysts found a significant statistical relationship between ICT hardware investment and intangible capital investment: historically, for every additional dollar of hardware investment, it drove two dollars of intangible capital investment—comprising $1.30 in data/software investment, $0.50 in organizational capital, and $0.20 in other intangible assets.
Applying this multiplier to the current scale of AI hardware investment corresponds to roughly $700 billion in supporting intangible capital investment in the U.S. and about $1 trillion globally. The analysts also cite survey data from the Atlanta Fed as corroboration. That survey, covering software, subscription services, hardware, employee training, and IT support, estimates corporate AI-related spending in 2026 will be approximately $280 billion.
**Revenue from Data Management Companies Tells the Story**
The acceleration in non-hardware investment is already evident in market data.
Leading data management and infrastructure companies like Snowflake, Databricks, and Palantir have seen their revenues more than triple since the emergence of ChatGPT in 2022. The combined enterprise value of these three companies has surged from under $100 billion in 2022 to over $650 billion by the end of 2025.
Cloud service revenue also confirms this trend.
Cloud service revenue from Amazon AWS, Microsoft, Google, and Oracle has grown from about $200 billion in 2022 to over $500 billion currently, with the market expecting it to surpass $1 trillion by the end of the decade. Notably, the consensus revenue forecast for 2026 has been revised upward by more than $150 billion since the end of 2022.
**The Nature of Non-Hardware Investment: Intangible Capital**
The analysts categorize these non-hardware investments as "intangible capital investment," encompassing patents, trademarks, brands, software, R&D, employee training, and organizational management capabilities.
According to the EU KLEMS database, intangible capital investment already accounts for over 50% of total investment in the U.S. and the U.K., and about 48% across the G10.
Over the past two decades, the rise in this share has been primarily driven by organizational capital and database/software investments—precisely the areas requiring the most significant outlays for AI and AI agent deployment.
**The Productivity "J-Curve": Present Undervaluation, Future Surprise**
Large-scale intangible capital investment tends to depress GDP and productivity statistics in the short term—a well-documented "J-curve effect" in economics.
The reason is that when companies redirect resources to internal AI tool development, process redesign, and employee retraining, these expenditures are recorded as costs rather than investments in national accounts, and the corresponding asset value is not counted.
The analysts estimate that AI-related organizational investment alone (executive time at $40 billion + labor restructuring at $10 billion, totaling about $50 billion annually) already leads to an underestimation of U.S. GDP by at least 0.2%. If the relevant statistical relationships hold, the underestimation could be as high as 2% of GDP.
In other words, there is room for further improvement in recent U.S. productivity data.
**Who Will Be the Next Generation of "Superstars"?**
This is the report's most direct implication for investors.
Goldman Sachs cites the "4 S" framework proposed by economists Haskel and Westlake in *Capitalism Without Capital* (2017), highlighting the fundamental differences between intangible and traditional tangible capital:
* **Scalability** (marginal cost approaches zero) * **Sunkness** (cannot be resold, higher risk) * **Spillovers** (knowledge and practices easily replicated by competitors) * **Synergies** (different types of intangible assets amplify each other's value)
This cost structure—high fixed costs, low marginal costs—naturally favors first movers. Historical data already validates this: over the past 40 years, the revenue share of "superstar" firms has risen almost in tandem with the scale of intangible capital investment.
The report's regression analysis further shows that for every 1 percentage point increase in the share of intangible capital, the labor cost share of value added decreases by 0.2 to 0.3 percentage points over the next 2 to 4 years, with brand, software/database, and organizational capital being the most significant drivers.
The conclusion is that companies that more effectively invest today in data, labor, and organizational infrastructure to deploy AI agents are likely to become the next generation of superstar firms commanding premium valuations.
Goldman Sachs also notes two important caveats.
First, this analysis does not address profit distribution within the AI technology stack. Chipmakers or foundational model providers could equally become the true AI superstars—this depends on where market power ultimately accumulates, which remains highly uncertain.
Second, the automation and business process standardization brought by AI will enhance productivity for most firms, not just a few winners. The analysts maintain their prior judgment: AI will bring broad, economy-wide productivity gains. They estimate that full AI diffusion in developed economies could ultimately boost labor productivity and GDP levels by approximately 15%.
Comments