Bank of America warns investors of diminishing returns from AI spending

The AI trade made plenty of investors look brilliant. Bank of America strategists think the effortless part of that story is ending.

BofA’s strategists say investors will have a harder time pulling easy gains out of AI-related investments. The warning comes as the biggest spenders on AI infrastructure keep writing larger checks while the market grows less certain those checks will pay off.

The numbers behind the caution

Start with the spending. BofA projected US hyperscaler capital expenditure at around $795 billion in 2026 and nearly $1.08 trillion in 2027.

Hyperscalers are the giant cloud operators, names like Microsoft, Alphabet and Amazon. Capex is the money they pour into physical infrastructure, the data centers and hardware that AI models need to run.

The stock market has not been cheering. By June 2026, US hyperscalers were trailing the S&P 500 by nearly 15% year-to-date.

BofA read that underperformance as a sign of doubt about whether the expected returns on all that investment will materialize. A BofA equity strategy note from that same month warned that the AI-driven stock rally might be losing momentum. The note also pointed out that corporates are rationing their AI usage and shifting toward cheaper models.

That detail carries more weight than it first appears. If paying customers are trading down, the landscape is more competitive than many expected, and premium pricing becomes harder to defend.

Fund managers are getting nervous

The unease extends well beyond one strategy desk. In BofA’s August 2026 Global Fund Manager Survey, 38% of respondents identified AI hyperscaler capital spending as a likely source of systemic credit risk.

By September, AI spending was taking up 80% of discussion time in BofA’s client meetings. Those conversations featured analogies to historical tech booms, a comparison that rarely gets made when everyone feels relaxed.

The financing side adds another layer. Hyperscalers are funding their spending through debt, reportedly reaching a capex-to-free-cash-flow ratio of approximately 94%.

That ratio compares what a company spends on building with the spare cash its operations generate. At approximately 94%, there is very little daylight between the two, which leaves a thin cushion if returns arrive later or smaller than planned.

What this means for investors

The nearly 15% year-to-date underperformance by June 2026 shows that recalibration may already be underway. Investors appear less willing to reward spending on its own and more interested in evidence that it produces revenue and margins.

The credit angle deserves close attention. With hyperscalers leaning on debt and the 38% survey reading flagging systemic credit risk, any stumble in AI monetization could matter to bond investors and lenders, not just shareholders.

For investors, a few signals stand out over the coming quarters. Watch whether corporate customers keep rationing AI usage, whether pricing resistance deepens, and whether hyperscaler capex guidance continues to climb toward the roughly trillion-dollar levels BofA has modeled.