Why Corporate America Is Seeing Zero Financial Return on Massive AI Spending

2026-10-24 economy

New York, Sunday, 25 October 2026.
Apollo’s analysis reveals enterprise AI investments deliver zero bottom-line gains outside tech, with 80% of top spending driven circular by venture capital rather than organic earnings.

Apollo’s Stark Finding: AI Spend Fails to Move the Needle Outside Tech

A sweeping new analysis from Apollo Global Management, the $653 billion alternative asset manager, has delivered a sobering verdict to corporate America: enterprise-wide investments in artificial intelligence have produced virtually no measurable financial return for industries beyond the technology sector in 2026. The findings, published in Apollo’s Daily Spark series on 21 October 2026, indicate that despite multi-billion-dollar capital expenditures and relentless industry hype, corporate earnings, productivity metrics, and profit margins outside Silicon Valley remain flat, offering no discernible boost to the bottom line[1]. This assessment comes at a pivotal moment as chief executives and boardrooms face mounting pressure to justify escalating enterprise software and automation budgets amid broader macroeconomic headwinds, including rising interest rates and elevated energy costs[6].

The Circularity Trap: How Venture Capital Fuels AI Spending Without Real Payoff

Underlying Apollo’s analysis is a deeper structural issue: AI spending outside the technology sector is increasingly circular and unsustainable. According to Apollo’s research, 80% of the top 1% of AI spenders are venture capital-backed firms, compared to just 13% of median spenders[4][5]. This dynamic suggests that much of the capital flowing into AI infrastructure—particularly compute token purchases—is being subsidized by venture funding rather than generated from organic earnings[4]. The phenomenon, described by Apollo as “circularity in AI token spending,” raises critical questions about the long-term viability of such investments. Financial commentators such as Mike Zaccardi, CFA, CMT, have noted that if AI returns continue to lag, the burden of unproductive capital will ultimately fall on venture returns, potentially weakening demand for AI infrastructure tokens and exacerbating sectoral imbalances[5].

Silicon Valley’s Outsized Gains Mask Broader Economic Stagnation

The discrepancy between AI’s financial impact in the technology sector and its lack of payoff elsewhere is stark. Apollo’s analysis indicates that the so-called “Mag 7” tech giants—comprising firms like Nvidia, Microsoft, and Alphabet—are capturing the vast majority of incremental revenues and profit margins associated with AI deployment[4][7]. Torsten Slok, Apollo’s chief economist, has observed that while these companies report substantial earnings growth, a significant portion is attributable to inter-sectoral revenue recognition, where one firm’s capital expenditure becomes another’s revenue stream—an effect consistent with the Kalecki-Levy income equation[7]. Meanwhile, outside this elite cohort, corporate profit margins remain largely unchanged, with no evidence of productivity gains that AI proponents have long promised[4][7].

Data Center Bottlenecks and Regulatory Headwinds Complicate the AI Narrative

Despite these sobering findings, AI remains a critical driver of economic growth in 2026, albeit unevenly distributed. Slok has emphasized that AI infrastructure—particularly data center construction—accounts for approximately 1 percentage point of U.S. GDP growth, even as overall growth remains below pre-pandemic norms[8]. However, this expansion is threatened by mounting operational and political challenges. In the second quarter of 2026, 45 major data center projects valued at $68 billion were delayed due to public opposition and regulatory constraints, and experts warn that only 25% to 50% of planned facilities may ultimately be completed because of power availability limitations[8]. Upcoming ballot measures in California—where voters will decide on proposals to prohibit or impose moratoriums on data center construction—highlight the growing bipartisan backlash against the industry’s environmental and infrastructure footprint[8].

Why AI Productivity Gains Remain Elusive Outside Software

The absence of measurable financial returns from AI outside the tech sector is not merely a function of poor execution; it reflects deeper structural constraints inherent to non-software industries. Industry observers point to a fundamental mismatch between AI’s current capabilities and the operational realities of sectors such as healthcare, law, and manufacturing. In white-collar environments, critical context is often scattered across fragmented systems or resides in employees’ tacit knowledge, making it resistant to automation without costly organizational restructuring[9]. Unlike software development—where AI excels due to verifiable, deterministic tasks such as code generation and static analysis—non-tech industries face high variability, regulatory complexity, and the need for physical capital investment, processes that historically require decades to yield measurable productivity gains[9]. This dynamic echoes the historical transition from water power to electrification, which took over a decade to fully manifest in productivity statistics[9].

A Cautionary Tale or a Temporary Lag? The Debate Among Economists

The interpretation of Apollo’s findings has sparked intense debate. While some, like Doug Simons, caution that the current AI investment cycle may resemble a 2000-style dot-com bubble, others argue that it is far too early to dismiss AI’s long-term potential[7]. Critics such as Haseeb Qureshi contend that Apollo’s methodology—focusing on aggregate profit margins rather than isolating returns between high-AI adopters versus low-AI adopters—risks obscuring the true economic impact of the technology[3]. Qureshi notes that rising interest rates, elevated oil prices, and weak consumer sentiment may be masking AI’s stabilizing influence on corporate margins in a challenging macroeconomic environment[3]. Meanwhile, Apollo’s own data center analysis underscores the dual nature of AI’s economic footprint: while it drives growth in capital-intensive sectors, its benefits remain concentrated and unevenly distributed[8].

Implications for Corporate Strategy and Capital Allocation

For corporate leaders navigating this landscape, Apollo’s findings carry significant strategic implications. The report suggests that many non-tech enterprises may be overestimating the near-term financial returns of AI deployments and underestimating the hidden costs of integration, including employee deskilling, technical debt, and the risk of “AI psychosis”—a phenomenon where AI-generated misinformation and unverified outputs lead to costly missteps[9]. Some organizations, however, are achieving tangible gains by targeting narrowly defined, high-verifiability use cases. For instance, AI has transformed medical transcription in Bay Area hospitals, enabling faster documentation and reducing administrative burdens—though concerns persist about accuracy and patient safety[9]. The contrasting experiences highlight a broader truth: AI’s value is highly context-specific, and indiscriminate adoption without clear metrics of success may yield diminishing returns.

The Road Ahead: Balancing Hype with Realism

As of October 2026, the AI investment narrative remains bifurcated: a thriving tech ecosystem fueled by AI revenues and a lagging broader economy struggling to translate capital expenditures into productivity gains. Apollo’s analysis suggests that without fundamental improvements in AI’s applicability to non-software domains or a sustained period of benign macroeconomic conditions, the current investment cycle risks becoming unsustainable. The firm’s data center projections, regulatory headwinds, and circular spending dynamics all point to a maturing phase in the AI economic cycle—one that demands greater discipline, clearer ROI frameworks, and more realistic expectations from corporate stakeholders[1][4][8]. The question now is whether this moment represents a necessary correction or the beginning of a longer, more painful reckoning for the AI economy.

Sources

Artificial intelligence Corporate earnings