This is not an isolated forecast. Gartner's broader research on AI project failures points to consistent, non-technical reasons driving the pullback :

• Escalating costs with no clear ROI model — AI infrastructure costs (compute, APIs, integration, ongoing maintenance, and human oversight) are significant and often spiral. 90% of CIOs report that AI costs are already limiting their ability to maximize value .
• Unclear business value — Many projects are abandoned after proof-of-concept because organizations cannot demonstrate a measurable impact on the bottom line .
• Poor data quality and inadequate risk controls — Weak governance, data issues, and insufficient risk management cause projects to collapse . Gartner predicts that through 2026, organizations will abandon 60% of AI projects unsupported by AI-ready data .
• "Agent washing" and use-case drift — Organizations pursue AI for its own sake without a disciplined business case, leading to failure. Gartner analyst Anushree Verma notes that many vendors have rebranded existing chatbots as AI agents without delivering meaningful outcomes .

These findings align with Gartner's separate April 2026 estimate that 57% of IT infrastructure and operations managers have at least one AI project failure behind them, and that one in five AI projects in that domain collapses entirely .

The Spending-ROI Disconnect: Evidence from Multiple Sources

Multiple major sources corroborate the growing disconnect between massive AI investment and measurable business outcomes:

Goldman Sachs (August 2026): Strategist Ben Snider reports that AI spending is surging sharply, but the technology has yet to produce a meaningful improvement in earnings for most companies. Only 2% of S&P 500 companies have quantified AI's impact on their earnings, and most report little to no immediate benefit . Analyst Jessica Rindels estimates U.S. AI investment will total almost $600 billion in 2026, equivalent to nearly 2% of GDP, yet evidence of matching economic gains remains thin .

Goldman Sachs (May 2026): Analyst Jim Covello notes that most enterprises have yet to generate any returns from their AI spending, citing an MIT study that found 95% of enterprise AI pilots delivered no measurable P&L impact . Covello argues the economics of AI are "more questionable today than two years ago" .

Goldman Sachs (July 2026): Economist Jan Hatzius and analyst David Peng offer a reality check — AI adoption and its productivity payoff may take far longer than markets expect. Their analysis shows productivity impact follows a J-curve: a modest drag for the first four years, statistically significant gains only after eight years, and a peak impact of roughly 0.6 percentage points in year 12. If ChatGPT's 2022 launch is the equivalent of the PC's 1981 debut, the payoff is still years away .

McKinsey (2026): Fewer than 20% of enterprise AI deployments show measurable business impact .

Sequoia Capital: Has flagged a roughly $600 billion revenue gap between AI investment and AI return .

Hyperscaler cash flow risk: Goldman Sachs warns that the four dominant cloud vendors (Amazon, Alphabet/Google, Microsoft, Meta) are collectively on course to spend more than their entire operating cash flow on AI infrastructure, with no proven enterprise ROI to justify the outlay . Combined 2026 capex for these four companies is projected at roughly $650 billion .

Why Projects Fail: It's Not the Technology

A consistent finding across all major analyses is that AI project failures are organizational and strategic, not technical. The model almost never fails — but the business case, data foundation, and cost controls do.

Gartner identifies three primary failure modes driving its predictions: escalating costs with no clear ROI model, unclear business value, and inadequate risk controls . As one Gartner analyst put it, "most agentic AI projects right now are early-stage experiments or proof of concepts that are mostly driven by hype and often misapplied" .

The consequences are already visible. Gartner predicts that by 2028, 70% of AI initiatives will be decommissioned due to unmanaged cost explosions and value drift in organizations lacking a dedicated AI financial management practice .

What This Means for Business Leaders

The pattern across Gartner, Goldman Sachs, McKinsey, MIT, and Sequoia is consistent: AI spending is at historic highs, but the vast majority of organizations cannot yet demonstrate measurable, bottom-line returns from that spending. Many are already pulling back or failing to get projects past the pilot stage.

For organizations investing in AI, the research points to a clear path forward: identify a specific operational bottleneck with measurable cost or quality impact, build an AI solution for that specific problem, measure the before/after impact rigorously, and scale only what works . Projects that fail to establish clear ROI metrics from day one are unlikely to survive the coming reckoning.

This is not an isolated forecast. Gartner's broader research on AI project failures points to consistent, non-technical reasons driving the pullback :

• Escalating costs with no clear ROI model — AI infrastructure costs (compute, APIs, integration, ongoing maintenance, and human oversight) are significant and often spiral. 90% of CIOs report that AI costs are already limiting their ability to maximize value .
• Unclear business value — Many projects are abandoned after proof-of-concept because organizations cannot demonstrate a measurable impact on the bottom line .
• Poor data quality and inadequate risk controls — Weak governance, data issues, and insufficient risk management cause projects to collapse . Gartner predicts that through 2026, organizations will abandon 60% of AI projects unsupported by AI-ready data .
• "Agent washing" and use-case drift — Organizations pursue AI for its own sake without a disciplined business case, leading to failure. Gartner analyst Anushree Verma notes that many vendors have rebranded existing chatbots as AI agents without delivering meaningful outcomes .

These findings align with Gartner's separate April 2026 estimate that 57% of IT infrastructure and operations managers have at least one AI project failure behind them, and that one in five AI projects in that domain collapses entirely .

The Spending-ROI Disconnect: Evidence from Multiple Sources

Multiple major sources corroborate the growing disconnect between massive AI investment and measurable business outcomes:

Goldman Sachs (August 2026): Strategist Ben Snider reports that AI spending is surging sharply, but the technology has yet to produce a meaningful improvement in earnings for most companies. Only 2% of S&P 500 companies have quantified AI's impact on their earnings, and most report little to no immediate benefit . Analyst Jessica Rindels estimates U.S. AI investment will total almost $600 billion in 2026, equivalent to nearly 2% of GDP, yet evidence of matching economic gains remains thin .

Goldman Sachs (May 2026): Analyst Jim Covello notes that most enterprises have yet to generate any returns from their AI spending, citing an MIT study that found 95% of enterprise AI pilots delivered no measurable P&L impact . Covello argues the economics of AI are "more questionable today than two years ago" .

Goldman Sachs (July 2026): Economist Jan Hatzius and analyst David Peng offer a reality check — AI adoption and its productivity payoff may take far longer than markets expect. Their analysis shows productivity impact follows a J-curve: a modest drag for the first four years, statistically significant gains only after eight years, and a peak impact of roughly 0.6 percentage points in year 12. If ChatGPT's 2022 launch is the equivalent of the PC's 1981 debut, the payoff is still years away .

McKinsey (2026): Fewer than 20% of enterprise AI deployments show measurable business impact .

Sequoia Capital: Has flagged a roughly $600 billion revenue gap between AI investment and AI return .

Hyperscaler cash flow risk: Goldman Sachs warns that the four dominant cloud vendors (Amazon, Alphabet/Google, Microsoft, Meta) are collectively on course to spend more than their entire operating cash flow on AI infrastructure, with no proven enterprise ROI to justify the outlay . Combined 2026 capex for these four companies is projected at roughly $650 billion .

Why Projects Fail: It's Not the Technology

A consistent finding across all major analyses is that AI project failures are organizational and strategic, not technical. The model almost never fails — but the business case, data foundation, and cost controls do.

Gartner identifies three primary failure modes driving its predictions: escalating costs with no clear ROI model, unclear business value, and inadequate risk controls . As one Gartner analyst put it, "most agentic AI projects right now are early-stage experiments or proof of concepts that are mostly driven by hype and often misapplied" .

The consequences are already visible. Gartner predicts that by 2028, 70% of AI initiatives will be decommissioned due to unmanaged cost explosions and value drift in organizations lacking a dedicated AI financial management practice .

What This Means for Business Leaders

The pattern across Gartner, Goldman Sachs, McKinsey, MIT, and Sequoia is consistent: AI spending is at historic highs, but the vast majority of organizations cannot yet demonstrate measurable, bottom-line returns from that spending. Many are already pulling back or failing to get projects past the pilot stage.

For organizations investing in AI, the research points to a clear path forward: identify a specific operational bottleneck with measurable cost or quality impact, build an AI solution for that specific problem, measure the before/after impact rigorously, and scale only what works . Projects that fail to establish clear ROI metrics from day one are unlikely to survive the coming reckoning.