Corporate America has poured extraordinary sums into artificial intelligence infrastructure over the past two years, treating the technology as an existential competitive necessity. But a striking new analysis from Goldman Sachs throws cold water on the return-on-investment narrative that has driven much of that spending: the earnings impact, at least as reported by companies themselves, is nearly invisible — and where it is visible, the news is often unwelcome.
According to the Goldman Sachs findings, cited by Seeking Alpha on August 16, 2026, just 2% of companies in the S&P 500 actually quantified the effects of artificial intelligence in their second-quarter earnings reports. That figure alone is arresting. Hundreds of chief executives have spent the better part of two years invoking AI in investor calls, strategy presentations, and annual letters as a transformational growth engine. Yet when it came to attaching hard numbers to that transformation in a formal earnings context, the overwhelming majority declined or were simply unable to do so.
More pointed still is what happened among the small cohort that did try to measure AI's financial footprint. Of that 2% of S&P 500 companies willing to quantify AI's effects, 11% reported that AI spending had produced a negative impact on their earnings. In other words, a meaningful share of the companies disciplined enough to actually count the cost found that the technology, at this stage, is subtracting from the bottom line rather than adding to it. This is the kind of data point that reframes an entire investment narrative — quietly, without fanfare, but with considerable consequence for capital allocation decisions across the market.
The Gap Between Narrative and Numbers
The divergence between AI's rhetorical prominence and its measurable financial contribution is not, in isolation, surprising to anyone who has studied previous waves of enterprise technology adoption. The commercial internet, cloud computing, and enterprise resource planning software all passed through prolonged periods during which capital expenditure raced well ahead of demonstrable productivity or earnings gains. Economists and technology historians sometimes refer to this as the productivity paradox — the lag between the deployment of transformational technology and its appearance in economic statistics.
What is notable in the Goldman Sachs analysis is the sheer scale of the gap. When fewer than one in fifty S&P 500 companies can quantify AI's contribution to earnings — after two-plus years of aggressive, often highly publicized investment — it suggests that either the technology is still in a pre-commercial maturation phase for most enterprise applications, or that the tools companies need to measure AI's impact on their operations remain underdeveloped. Possibly both are true simultaneously.
From a banking and financial services perspective, this matters acutely. Institutions ranging from regional lenders to global investment banks have committed substantial resources to AI-driven underwriting, fraud detection, customer service automation, and regulatory compliance tools. Those deployments represent real capital expenditure hitting real income statements. If the broader corporate sector cannot yet demonstrate earnings accretion from AI, financial services firms face precisely the same pressure to justify their AI budgets to boards, regulators, and shareholders — armed with very little standardized methodology for doing so.
Disclosure Standards Are Now the Battlefield
The Goldman Sachs finding implicitly raises a disclosure question that financial regulators and standard-setters have yet to resolve. If AI spending is material to a company's competitive strategy and cost structure — and for a growing number of firms it plainly is — then the absence of quantified AI impact disclosures in earnings reports starts to look less like a measurement problem and more like a governance gap. Investors allocating capital on the basis of AI-optimism deserve to know whether that optimism is grounded in measurable outcomes or in aspiration.
The U.S. Securities and Exchange Commission has been active in pushing for enhanced disclosure around climate-related financial risks; there is a reasonable argument that AI-related capital expenditure and its earnings implications warrant similar scrutiny, particularly given the multibillion-dollar commitments some companies have made. Without standardized reporting frameworks, investors are left comparing incomparable management commentaries — a fog of optimism with no agreed-upon unit of measurement.
What This Means for Investors and Institutions
The Goldman Sachs analysis should serve as a discipline-restoring moment for both markets and management teams. It does not argue that AI is without commercial promise — that would be an overreaction to a dataset representing a single earnings quarter. What it does argue, with quiet authority, is that the promise has not yet arrived in the form that investors can actually count. Companies that have leaned heavily on AI as a forward earnings growth story now face a credibility test: at what point do disclosures need to catch up with expenditure?
For financial institutions in particular, the pressure is compounding. Regulatory expectations around AI governance are rising across jurisdictions, while internal CFOs are being asked to justify AI budgets that, by Goldman's own reckoning, the majority of companies cannot even express in earnings terms. The 2% figure is not just a statistical curiosity — it is a benchmark of corporate accountability for the most consequential technology investment cycle of the decade. The market will eventually demand that benchmark rises, and the institutions that build rigorous AI return-on-investment frameworks ahead of that demand will be better positioned than those that wait.
Written by the editorial team — independent journalism powered by Codego Press.
