The AI investment paradox: Genuine transformation or FOMO at scale?

As Microsoft, Alphabet, Amazon and Meta plan to invest a combined $320 billion in AI technologies in 2025 based on the findings of Ropes & Gray LLP, the technology industry faces a critical question: Are we witnessing a transformative productivity revolution, or inflating the most anticipated bubble in market history?

The question isn’t academic. For CTOs and CIOs making budget decisions in 2026, the stakes couldn’t be higher. Bet too conservatively and risk competitive obsolescence as AI-powered competitors surge ahead. Bet too aggressively on unproven technology and risk joining the growing list of organizations that have poured millions into AI initiatives without measurable returns.

The answer reveals a market simultaneously delivering measurable productivity gains while exhibiting concerning characteristics of speculative excess. Understanding the dynamics driving investment in your organization and industry may determine whether AI becomes a transformative catalyst or an expensive cautionary tale.

The scale of the bet

AI investment has reached unprecedented levels, dwarfing previous technology cycles. Gartner, Inc. estimates that global spending is projected to hit $1.5 trillion in 2025, climbing to over $2 trillion by 2026. Organizations increased AI infrastructure spending by 166% year over year in Q2 2025, reaching $82 billion, according to the International Data Corporation. AI-related capital expenditures accounted for 1.1% of GDP growth in H1 2025, surpassing the U.S. consumer as the primary driver of economic growth documented by J.P. Morgan Asset Management.

Goldman Sachs Research notes consensus estimates have underestimated AI capex growth by more than 50% for two consecutive years. Wall Street analysts, known for aggressive growth projections, have consistently been too conservative. RSM US LLP estimates that these financial commitments extend far into the future, with Big Tech expected to dedicate $300-400 billion annually over the next eight years.

The velocity and magnitude of this investment cycle set it apart. During the dot-com bubble, investment ramped up over several years. With AI, we’ve seen near-vertical acceleration within months of ChatGPT’s November 2022 launch.

The case for genuine value

Enterprise adoption has accelerated dramatically, with 78% of organizations now using AI, up from 55% in 2023, according to Stanford HAI. This isn’t superficial experimentation but core workflow integration. Research done by Fullview estimates that early adopters report a $3.70 value per dollar invested, with top performers achieving returns of $10.30. Wharton research found 72% of enterprises formally measure AI ROI, with three-quarters seeing positive returns.

The productivity gains are substantial and measurable. Workers save 40-60 minutes daily, resulting in a 10% productivity improvement based on research by ALM Corporation. For a 1,000-person organization with average labor costs of $100,000, this represents $10 million in annual value. IBM has documented $4.5 billion in productivity savings through internal AI initiatives.

Specific use cases demonstrate concrete impact. Software development AI spending reached $4 billion, with 50% of developers using AI coding tools daily, based on Menlo Ventures findings. Menlo also discovered healthcare AI solutions captured $1.5 billion in 2025, up from $500 million the previous year. Administrative burden has become unsustainable for many healthcare organizations, and AI-powered clinical documentation tools are delivering immediate relief.

Fullview noted that financial services firms report average productivity gains of 20%, with 57% of AI leaders in finance reporting ROI exceeding expectations. Loan processing accuracy has increased by 90%, while processing times have fallen by 70%. IBM noted that across industries, 66% of surveyed enterprises reported significant productivity gains.

The pattern is clear: organizations with disciplined implementation, rigorous measurement and fundamental workflow redesign are capturing substantial value.

The warning signs mount

These genuine achievements coexist with alarming indicators of speculative excess. Jamie Dimon, head of JPMorgan, warns that while “AI is real,” much money being invested will be wasted. Coming from one of the most influential voices in global finance, this isn’t casual skepticism but a considered assessment of systemic risk.

The structural concerns fall into three categories. First, circular financing has become endemic. NVIDIA’s $100 billion investment in OpenAI, in which NVIDIA funds a customer to purchase its own products, exemplifies this pattern. Michael Burry, who famously predicted the 2008 housing crisis, observes: “True end demand is ridiculously small. Almost all customers are funded by their dealers” based on an investigation by NPR. These circular arrangements make it nearly impossible to assess genuine market demand.

Second, companies employ increasingly complex financial engineering to keep AI debt off balance sheets. Special-purpose vehicles now represent at least $100 billion in off-balance-sheet debt. Meta’s Louisiana data center deal involves a $27 billion loan that never appears on Meta’s balance sheet, according to NPR. The structure is reminiscent of special-purpose entities that obscured risk in previous financial crises.

Third, valuations badly mismatch realistic revenue trajectories. OpenAI, valued at $500 billion, loses over $11.5 billion quarterly while projecting only $13 billion in annual revenue in 2025 according to The American Prospect. The company simultaneously commits to $300 billion in computing spending with Oracle over five years.

An MIT study finding 95% of generative AI initiatives are getting zero return reflects genuine monetization challenges. NPR noted that only 3% of customers currently pay for AI services. If users won’t pay directly, the business model depends entirely on indirect value capture, neither of which has been demonstrated at the scale required to justify current valuations.

The concentration risk

Market concentration has reached concerning levels. In late 2025, 30% of the S&P 500 was held by just five companies, the greatest concentration in half a century. AI-related stocks accounted for 75% of S&P 500 returns since ChatGPT’s launch, based on Yale Insights research.

Harvard’s Andy Wu notes that Big Tech’s hedging strategies, Microsoft outsourcing to OpenAI, Amazon supporting any model and Meta building open-source suggest these companies “don’t really think that core AI technology is a meaningful business in and of itself” in an open article in the Harvard Gazette.

The downstream customers of AI infrastructure face acute risk. Wu observes, “there’s no short-term scenario in which they are economically viable given how costly it is today”. These companies lack immediate paths to profitability.

Distinguishing signal from noise

FOMO and genuine value coexist and interact in complex ways. Fullview noted the challenge for technology leaders is separating sustainable transformation from speculative mania. Success correlates strongly with execution discipline rather than investment magnitude. Organizations achieving impact commit 20%+ of digital budgets to AI, invest 70% of AI resources in people and processes rather than technology alone, and implement rigorous oversight.

The size advantage is real but not deterministic. IBM determined that large enterprises report productivity gains (72%) more frequently than small businesses (55%). This gap reflects not only resource availability but also organizational maturity, change management capability and the presence of standardized processes that can benefit from automation. Small organizations can succeed, but must be even more disciplined in their use-case selection and implementation rigor.

The measurement gap remains the most concerning indicator separating winners from losers. Research performed by Larridin noted that while 89% of enterprises use AI, only 23% measure ROI. Without rigorous metrics, organizations cannot distinguish between genuine transformation and expensive experimentation. They’re flying blind, making continued investment decisions based on enthusiasm rather than evidence.

Organizations with rigorous metrics report impressive returns: 27% productivity improvement, 11.4 hours saved per knowledge worker per week and $8,700 per employee annually reported by Larridin. These organizations didn’t achieve better results by accident. They established clear baselines, systematically tracked usage and outcomes, and made data-driven decisions about scaling or pivoting. The measurement discipline itself drives better outcomes by forcing clarity about objectives and accountability for results.

The path forward for technology leaders

For technology leaders navigating 2026’s investment decisions, several principles emerge:

• Measure relentlessly. The divide between success and disappointment correlates directly with measurement rigor. Connect AI usage to revenue per employee, cost per transaction or other metrics that matter to the business. Create dashboards that make AI’s impact visible.
• Redesign workflows fundamentally. Organizations achieving enterprise-wide impact are three times more likely to fundamentally redesign workflows rather than automate existing processes, based on research performed by McKinsey & Company. AI’s value emerges from transformation, not incremental efficiency. The organizations seeing 10x improvements have stopped asking “how can AI help with this task” and started asking “if we could do this perfectly, what would it look like.”
• Build organizational capabilities systematically. Formal AI training programs achieve 2.7x higher proficiency and 4.1x higher satisfaction than self-guided learning, according to Larridin. The most successful organizations treat AI capability building like any other critical skill, with a structured curriculum and regular assessment.
• Maintain healthy skepticism. Federal Reserve Chair Powell notes AI differs from previous bubbles because corporations generate substantial revenue. However, risk remains significant. For corporate technology leaders, this means being selective about vendor dependencies.
• Expect longer payback periods. Fullview noted that most organizations achieve ROI within 2-4 years, substantially longer than typical technology payback periods. Early wins matter for maintaining momentum, but sustainable transformation requires patience and persistent iteration.

Both…and neither

Current AI investment is driven by both FOMO and genuine value operating simultaneously at a massive scale. This dual reality makes navigation particularly treacherous. The technology delivers measurable productivity gains when implemented with discipline. Organizations with the right approach are capturing real value that compounds over time. Simultaneously, speculative excess, circular financing and questionable valuations suggest significant correction risk that could affect even well-managed initiatives.

Organizations that will thrive through this period can maintain a clear-eyed assessment of both opportunities and risks. They must invest aggressively enough to capture productivity gains and avoid competitive disadvantage while remaining disciplined enough to avoid the financial engineering and speculative commitments that characterize bubble-era excess. This requires what might seem like contradictory stances: ambitious about possibility, skeptical about hype, patient with timelines and rigorous about measurement.

NPR noted in a recent story that 2024 Nobel Prize winner Daron Acemoglu observes: “I have no doubt that there will be AI technologies that will come out in the next ten years that will add real value and add to productivity, but much of what we hear from the industry now is exaggeration”. This balanced view captures the essential truth. AI is neither the solution to all problems nor a complete mirage. It’s a powerful set of tools that will transform many aspects of work, but transformation takes time, requires hard work and delivers unevenly across use cases and organizations.

Technology leaders must bet boldly on AI’s transformative potential while maintaining rigorous measurement and risk management that distinguishes sustainable transformation from speculative mania. The stakes are high, the uncertainties substantial and the need for balanced, evidence-based decision-making has never been greater. The winners won’t be those who invested most or least, but those who invested most wisely, with clear objectives, rigorous measurement and the organizational discipline to learn and adapt as this technology continues to evolve.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?