NextFin News - The multi-year honeymoon between Wall Street and artificial intelligence appears to be entering a period of cooling, as a record share of investors and the American public express growing unease over the scale of corporate spending on the technology. According to a Bank of America global fund manager survey released on Tuesday, February 17, 2026, 35% of institutional investors now believe companies are overinvesting in AI—a historic high that signals a fundamental shift in market psychology. This sentiment is echoed by the broader public; a YouGov/The Economist survey conducted over the past week found that 54% of Americans believe corporate AI investment has become excessive, with many citing fears of job displacement and a lack of trust in the technology.
The impact of this skepticism is already manifesting in equity valuations. For years, announcements of increased AI capital expenditure (capex) served as a reliable catalyst for stock price appreciation. However, recent earnings reports from industry leaders suggest this "AI tailwind" has turned into a headwind. Amazon, for instance, saw its stock price tumble nearly 16% after announcing plans to spend $200 billion on capex in 2026—a 60% increase from the previous year. Similarly, Microsoft shares have declined approximately 9% since late January, following a report detailing $37.5 billion in quarterly capex, representing a 65% year-over-year surge. This decoupling of spending and stock performance suggests that the market is no longer willing to grant tech giants a blank check for AI infrastructure without clear evidence of near-term profitability.
The underlying cause of this shift is a growing concern over the "opportunity cost" of AI. As Elyas Galou, a senior global investment strategist at Bank of America, noted, the massive capital requirements for AI infrastructure—including specialized chips and energy-intensive data centers—are siphoning funds away from stock buybacks. Since the 2008 financial crisis, buybacks have been a primary driver of the tech-led bull market. With Morgan Stanley recently raising its 2026 hyperscaler capex forecast to $740 billion, investors are increasingly wary that the pursuit of "compute supremacy" is coming at the expense of shareholder returns and balance sheet stability.
Beyond the boardroom, the AI boom is colliding with the realities of the U.S. energy grid and shifting federal policy. U.S. President Trump has recently moved to shield residential ratepayers from the soaring electricity costs associated with the AI buildout. In January 2026, U.S. President Trump stated on Truth Social that he "never wants Americans to pay higher electricity bills because of data centers." This has led to a Department of Energy proposal for "Emergency Capacity Auctions," which would require tech companies to sign 15-year "take-or-pay" contracts to fund new power plants. While this $15 billion strategy aims to add 7.5 gigawatts of baseload power, it places a direct financial burden on hyperscalers, further squeezing the margins of AI-heavy business models.
From an analytical perspective, the market is transitioning from a "visionary phase" to an "execution phase." The initial surge in AI valuations was driven by the promise of transformative productivity gains; however, the current phase is defined by the staggering costs of the physical layer—energy, land, and hardware. The fact that demand for compute still exceeds supply, as Morgan Stanley analysts noted, does not negate the financial risk of overcapacity. If the anticipated revenue from AI software and services fails to materialize at a scale that justifies $740 billion in annual infrastructure spending, the industry faces a potential "AI bubble" burst reminiscent of the dot-com era.
Looking ahead, the most significant indicator to watch will be the first major capex cut from a leading hyperscaler. Such a move would likely trigger a broader re-rating of the tech sector, as it would signal that even the wealthiest companies have reached the limits of their risk tolerance. Furthermore, as the 2026 midterms approach, AI is emerging as a potent political fault line. With 70% of participants in a recent FGS Global study believing AI will destroy more jobs than it creates, the political pressure to regulate or tax AI development may increase, adding another layer of uncertainty for investors who were once unequivocally bullish on the silicon-led future.
Explore more exclusive insights at nextfin.ai.
