Investor anxiety over the artificial intelligence spending binge has reached its highest point on record, according to Bank of America’s latest Fund Manager Survey. A record share of roughly 35% of fund managers now say corporations are overinvesting in capital expenditures, while about 25% label an “AI bubble” as the single biggest tail risk facing markets. The survey also reveals that approximately 30% of respondents view hyperscaler AI capital spending as the most likely trigger for a credit crisis, a finding that puts the tech sector’s infrastructure arms race under sharper scrutiny than at any point since the generative AI boom began.
Record Overinvestment Warnings in the BofA Survey
The February 2026 edition of the BofA Fund Manager Survey marks a new high-water mark for concern about corporate spending discipline. Roughly 35% of surveyed managers now believe companies are overcommitting capital, the largest share ever recorded in the monthly poll. That figure has climbed steadily over the past year as hyperscalers have announced progressively larger data-center buildouts and compute procurement deals. The acceleration suggests that professional allocators are no longer debating whether AI spending is aggressive, they are debating how badly it could end.
Equally telling is the risk taxonomy the survey produces. About 25% of respondents singled out an “AI bubble” as the top tail risk, placing it ahead of geopolitical conflict, inflation resurgence, and trade disruption. Meanwhile, roughly 30% identified hyperscaler AI capital expenditure specifically as the most probable origin of a credit event. That distinction matters because it moves the concern from a vague market-sentiment worry to a concrete credit-channel fear: if the largest technology companies lever up or misallocate capital on AI infrastructure that fails to generate proportional revenue, the ripple effects could reach bond markets, lending standards, and broader corporate financing conditions.
Hyperscaler Filings Show the Scale of the Bet
The survey results gain weight when read alongside the companies’ own regulatory disclosures. Meta Platforms’ 10-K for the fiscal year ended December 31, 2025, details billions directed toward data centers and AI-related infrastructure, with risk-factor language acknowledging uncertainty around payback timelines. Meta’s filing does not frame its spending as speculative, but the sheer magnitude of committed capital, combined with forward-looking language about investment payback, aligns with the concerns fund managers are voicing.
Microsoft’s most recent quarterly filing, accessible through its EDGAR page, shows a similar pattern of surging cloud and compute expenditures. Amazon and Alphabet round out the group. Amazon’s SEC filings document expanding data-center leases and infrastructure commitments tied to AWS and AI workloads, while Alphabet’s disclosures outline capital directed at Google Cloud and custom AI chips. Taken together, these four companies are pouring capital into AI infrastructure at a pace that dwarfs prior technology investment cycles, and none of their filings offers a firm timeline for when the spending will translate into commensurate earnings growth.
Why Credit Risk Is the Real Flashpoint
The survey’s credit-crisis finding deserves close attention because it signals a shift in how investors frame AI risk. For most of 2024 and 2025, the dominant worry was valuation: were AI-linked equities priced too richly? The new concern is structural. When roughly 30% of fund managers point to hyperscaler capex as the likeliest source of a credit event, they are describing a scenario in which overbuilt capacity forces write-downs, tightens lending to the broader tech sector, and potentially spreads stress through high-yield bond markets that have financed adjacent infrastructure plays.
This is not an abstract thought exercise. Large-scale capital misallocation has triggered credit dislocations before. The fiber-optic overbuild of the late 1990s left telecom companies with crushing debt loads and stranded assets, contributing to a wave of defaults in 2001 and 2002. The current AI buildout differs in important ways: the hyperscalers financing it are far more profitable and cash-rich than the telecom carriers of that era. But the BofA survey suggests that even well-capitalized companies can alarm creditors if spending growth consistently outpaces visible demand. The gap between capital deployed and revenue realized is the variable fund managers are now watching most closely.
The Counter-Argument: Spending as Strategic Moat
Not everyone reads the data as a warning sign. The bull case holds that massive upfront investment in AI compute creates durable competitive advantages that will pay off over a multi-year horizon. Each hyperscaler is effectively racing to lock in GPU supply, secure energy contracts, and build proprietary model-training capacity that smaller rivals cannot replicate. From this perspective, the spending is not reckless but rational: the cost of underinvesting, and ceding the AI platform layer to a competitor, could be far greater than the cost of overbuilding.
There is also a structural argument about efficiency gains. If investor pressure from surveys like BofA’s encourages hyperscalers to pursue open-source collaborations or shared infrastructure models, the industry could reduce redundant buildouts without sacrificing capability. Early signs of this dynamic are visible in the proliferation of open-weight large language models and cross-licensing agreements for specialized hardware. The question is whether these efficiency channels can scale fast enough to close the gap between capital outlay and revenue that the survey respondents find so troubling. For now, the filings show spending accelerating, not moderating.
Broader Capital Allocation Tensions
The AI capex debate does not exist in isolation. Across the economy, large-scale capital commitments are drawing scrutiny for their uncertain returns. Los Angeles, for instance, is pursuing a $25 billion subway project designed to ease freeway congestion, a bet on infrastructure whose payoff depends on ridership projections that may or may not materialize. In a different arena, the NFL’s Houston Texans plan to relocate their headquarters to a new suburban development, tying the franchise’s future to real-estate assumptions about suburban growth.
These examples illustrate a broader pattern: when capital is cheap or abundant, institutions of all kinds tend to commit to large, long-duration projects whose returns depend on optimistic forecasts. The AI buildout is the most concentrated version of this pattern because it involves a handful of companies spending at a scale that can move credit markets. Fund managers appear to be drawing a direct line between the hyperscalers’ capital plans and systemic financial risk, a connection that was largely absent from the conversation even a year ago.
Global Signals of Risk Repricing
The concerns highlighted in the Bank of America survey are emerging alongside other signs that investors are reassessing how they price long-term risk. In New Zealand, for example, net migration has fallen to its lowest level in a decade as more citizens depart, according to reporting on the country’s migration slowdown. That demographic shift complicates infrastructure and housing investment plans that had assumed steady population growth, underscoring how quickly macro assumptions can change beneath capital-intensive projects.
In the United States, a recent airspace closure in Texas and its political fallout have highlighted another dimension of risk: operational and regulatory shocks that can disrupt activity with little warning. For hyperscalers building AI data centers and cloud regions, similar disruptions (whether tied to energy supply, local permitting, or national security reviews) could alter project timelines and cost structures. The BofA survey’s focus on credit risk implicitly incorporates these uncertainties, since unexpected delays or policy shifts can turn otherwise sound investments into sources of balance-sheet strain.
What Shifts If Returns Lag
The practical question for investors and corporate strategists is what happens if AI revenue growth fails to keep pace with the infrastructure investment already locked in. The BofA survey does not offer a timeline, but the logic of the concern points toward a critical window over the next 12 to 18 months. By that point, the hyperscalers will have completed or nearly completed many of the data-center expansions currently under construction. If utilization rates disappoint, or if enterprise customers adopt AI tools more slowly than projected, the financial pressure will show up first in capital-efficiency metrics and then in credit-spread widening for the broader technology sector.
A pullback scenario would not necessarily mean AI itself has failed. The underlying technology could still prove enormously valuable even if the initial infrastructure investment turns out to be oversized. But for fund managers allocating capital today, the distinction between “AI works” and “AI spending earns an adequate return on capital” is crucial. If the latter does not materialize, boards may force management teams to curb capex, prioritize buybacks or dividends, and renegotiate supplier contracts. That process could cool the AI arms race without collapsing it, yet still deliver the kind of market volatility and credit repricing that the survey flags as a top tail risk.
How Investors and Companies Might Respond
One likely response to mounting anxiety is greater transparency. Hyperscalers may feel compelled to provide more granular disclosures about AI-related revenue, utilization rates, and project-level returns, going beyond the aggregate capex lines in their current filings. Independent outlets such as Bloomberg’s newsroom are already parsing these numbers closely, and more detailed reporting could help investors distinguish between sustainable investment and speculative overreach. Clearer metrics would not eliminate risk, but they could narrow the range of plausible outcomes that currently fuels bubble fears.
On the investor side, portfolio managers are likely to refine their exposure, favoring companies with demonstrably high returns on incremental AI investment and more conservative balance sheets. That could mean rotating within the tech sector rather than exiting it entirely, or using credit derivatives to hedge against the specific scenario the BofA survey highlights—a funding squeeze tied to overbuilt AI infrastructure. In that environment, the winners may be firms that can show, with data, that each additional dollar of AI capex is generating measurable productivity gains or recurring revenue rather than simply adding to an already vast pool of underutilized compute.
More From The Daily Overview
*This article was researched with the help of AI, with human editors creating the final content.
Grant Mercer covers market dynamics, business trends, and the economic forces driving growth across industries. His analysis connects macro movements with real-world implications for investors, entrepreneurs, and professionals. Through his work at The Daily Overview, Grant helps readers understand how markets function and where opportunities may emerge.
