The $600 Billion Revenue Gap: Why More AI Spending Is Not Producing More AI Value

By Dr. Felicia Newhouse, Founder, AI-Powered Women

Six hundred billion dollars. That is the approximate annual global spend on AI infrastructure as of 2025, according to analysis from Sequoia Capital. It includes compute, model development, integration, tooling, and the expanding ecosystem of enterprise AI platforms. It is, by any measure, an extraordinary allocation of capital toward a single technology category.

Here is the number that should accompany it in every boardroom conversation: AI-driven revenue remains a fraction of that figure. The gap between what organizations are spending on AI and what they are generating from it has widened, year over year, as infrastructure investment has scaled. Sequoia's analysis makes the trajectory clear. We are not in the early innings of a payoff curve. We are watching the distance between investment and return grow (Sequoia Capital, 2025).

This is the $600 billion question facing enterprise leaders in 2026. Where is the value?

The Scale of the Investment

The numbers are no longer speculative. Goldman Sachs estimates that hyperscaler capital expenditure on AI infrastructure reached $400 to $423 billion in 2025 alone. Bank of America's analysis found that at some companies, AI-related capital expenditure now consumes up to 94 percent of operating cash flows. These are not venture-stage bets. This is operational capital allocation at a scale that reshapes balance sheets and constrains investment in other strategic priorities (Goldman Sachs, 2025; Bank of America, 2025).

Enterprise organizations outside the hyperscaler tier are following the same pattern at smaller scale. AI platform licenses, data infrastructure, model fine-tuning, integration engineering, security and compliance architecture. The total cost of bringing AI capability into an enterprise technology stack has increased every quarter for three years.

The expectation behind this spending is clear: AI will produce measurable productivity gains, revenue growth, or cost reduction. The evidence that it is doing so at a rate proportional to the investment is far less clear.

The Bottleneck Is Not Compute

The conventional explanation for the gap is timing. AI infrastructure is a long-term investment. Returns will materialize as adoption matures. Give it time.

Research from Harvard Business School suggests the explanation is more structural. In a 2023 experiment with 758 consultants at Boston Consulting Group, Fabrizio Dell'Acqua and colleagues tested what happens when skilled professionals use frontier AI models on real consulting tasks. The results revealed a pattern that should concern every enterprise leader spending heavily on AI infrastructure.

On tasks that fell within AI's capability frontier, consultants using GPT-4 improved their performance by 40 percent. The tools worked as advertised, and the humans using them got better results. On tasks that fell outside that frontier, consultants using AI performed worse than those working without it. They deferred to confident-sounding but incorrect AI outputs rather than applying their own expertise. The AI did not just fail to help. It actively degraded the quality of human judgment (Dell'Acqua et al., 2023).

The variable that determined whether AI improved or degraded performance was not the model's capability. It was the human's capacity to evaluate AI outputs and exercise judgment about when to rely on them and when to override them. More capable tools without more capable humans did not produce better outcomes. They produced worse ones.

The Human Readiness Deficit

This finding from Dell'Acqua's team is the key to understanding the $600 billion revenue gap. The bottleneck to AI value is not compute capacity, model quality, or integration architecture. It is human capacity: the ability of people across the organization to identify where AI creates genuine value, integrate it meaningfully into their workflows, and exercise judgment when AI outputs are unreliable.

Consider the internal audit conducted by a major consulting firm that invested $50 million in AI infrastructure across its practice. When leadership assessed the results, only approximately 25 percent of teams had genuinely integrated AI into workflows that improved client outcomes. Another 40 percent had adopted the tools but were using them in ways that did not measurably improve productivity. The remaining 35 percent had access to the same tools and the same training and showed no clear outcome improvements.

The infrastructure was identical across all three groups. The models were the same. The integration was the same. The difference was at the human layer: whether the people using the tools had the readiness to use them in ways that produced value.

This pattern repeats across industries and geographies. The organizations that are closing the revenue gap are the ones investing in structured, ongoing AI capacity development for the people who actually use the tools. The organizations where the gap continues to widen are the ones that treat infrastructure as the strategy and assume human adaptation will follow.

The Reframe for Enterprise Leaders

AI infrastructure is a necessary condition for AI value. It is not a sufficient one. The sufficient condition is human readiness, and the data on what that requires is now clear.

It requires evaluation skill: the ability to distinguish between AI outputs that are reliable and those that sound reliable but are wrong. It requires workflow integration: the ability to redesign how teams work so that AI amplifies rather than replaces human judgment. It requires what researchers call "appropriate reliance," a calibrated understanding of when to trust AI and when to override it. And it requires ongoing development, because the tools, the models, and the best practices for using them are evolving on a timeline measured in months.

The organizations seeing real returns from AI investment are the ones that have recognized this and allocated their readiness budget accordingly. They are spending on human development with the same seriousness they bring to infrastructure procurement. They are measuring capability, not just adoption. They are treating readiness as a continuous investment, because in a technology environment that changes every quarter, one-time training is a depreciating asset.

The Decision in Front of You

If you are responsible for AI strategy, AI program delivery, or the budget that funds either, the question to ask before your next investment cycle is this: what percentage of your total AI spend is allocated to the readiness of the people who will determine whether the infrastructure produces value?

In most organizations, the honest answer is low single digits. The infrastructure gets funded. The licenses get approved. The human development work gets treated as a secondary initiative with a fraction of the budget and none of the urgency.

The $600 billion revenue gap is the result. It will not close with more infrastructure. It will close when organizations invest in the human capacity to turn that infrastructure into outcomes.

This is the central question of the 2026 MIT Conference on AI Leadership Readiness (September 12-13, Kresge Auditorium). More than 1,200 enterprise leaders will convene around the research, the frameworks, and the organizational strategies for closing the gap between AI spending and AI value. Register here to be part of the conversation.

References

Sequoia Capital. (2025). AI's $600B Question. Sequoia Capital Research.

Goldman Sachs. (2025). Global hyperscaler AI capital expenditure analysis.

Bank of America. (2025). AI capex as percentage of operating cash flows, enterprise analysis.

Dell'Acqua, F., McFowland, E., Mollick, E., Lifshitz-Assaf, H., Kellogg, K., Rajendran, S., Krayer, L., Candelon, F., & Lakhani, K. R. (2023). Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality. Harvard Business School Working Paper No. 24-013.

Dr. Felicia Newhouse is the founder of AI-Powered Women and convener of the 2026 MIT Conference on AI Leadership Readiness. The AI Leadership Readiness Program helps enterprise organizations close the gap between AI infrastructure investment and AI-driven business value.