‘A major risk to the global economy’: BofA warns about the AI boom the day before a midterm vote framed around data-center backlash

BofA warns that a sustained correction in the AI trade could weigh on global growth through weaker investment and a reversal in stock-driven wealth effects, while AI-related borrowing is contributing to interest-rate pressure. Data-center construction spending reached a preliminary annualized $84.95B in August, up 73% year over year; major tech companies are expected to spend about $800B on capex this year and $1.2T next year. The article also cites AI-linked labor-market pressures, local resistance to data-center projects, and softer analyst-reported orders, alongside uncertainty over post-election tariffs and policy.

Analysis

The key risk is not a near-term local-vote shock; it is a deterioration in the conversion of AI capex into durable cash flows. If infrastructure spending keeps rising while enterprise adoption fails to produce measurable productivity or revenue, investors may mark down the capex-heavy leaders and tighten financing assumptions across data-center projects. That can feed through to long yields and wealth-sensitive consumption. Meanwhile, construction hiring does not necessarily offset job displacement in professional services: skills and geography differ, so the consumption impact may be asymmetric.

Election outcomes are a weak proxy for a national buildout stop. The more plausible policy channel is narrower: permitting friction, tax-credit changes, or federal standards that raise project costs and delay power connections. That is a 1–3 month watch; the 6–18 month structural test is whether hyperscaler spending generates returns that justify continued investment. Grid equipment and power-infrastructure suppliers may retain order visibility, but labor, transformer, and interconnection constraints can cap volume and delay revenue recognition.

Contrarian angle: broader market breadth could absorb an AI-led correction better than headline indices imply, while local opposition may slow marginal projects rather than reverse committed spending. The thesis weakens if major buyers reaffirm capex while reporting accelerating AI-linked monetization, or if broad earnings and order indicators recover. It strengthens with capex cuts, project cancellations, weaker cloud demand, or a sustained rise in long yields.

Key Decisions for Investors

• Consider a measured long RSP / short QQQ pair as a relative-value expression of broadening earnings participation versus concentrated AI-capex exposure; scale in rather than trade the election headline. Falsify if mega-cap technology earnings breadth and AI-linked revenue reaccelerate while equal-weight relative performance rolls over.
• Avoid shorting data-center construction or power-infrastructure beneficiaries solely on local ballot results. Track permitting decisions, tax-credit proposals, interconnection delays, and order/backlog commentary; these are more relevant catalysts than campaign rhetoric.
• Treat the next 1–3 months as a capex-quality test: monitor hyperscaler spending guidance alongside cloud growth and evidence of customer monetization. A spending increase without corresponding revenue evidence is a reason to reduce AI infrastructure exposure; a clear monetization step-up invalidates that caution.
• Watch long-term Treasury yields and consumer-facing earnings for second-order confirmation. A sustained yield rise alongside weaker discretionary demand would support reducing broad equity beta; improving orders and stable rates would argue against extrapolating the current slowdown.