Why BofA says AI is not a big factor in near-term monetary policy

BofA estimates AI adds roughly 0.4 percentage points to U.S. GDP growth this year but is only modestly inflationary, mainly via higher utility/energy costs for data centers and a wealth effect from equity markets. Because adoption and diffusion remain low and productivity gains are not yet widespread, BofA says central banks are unlikely to change policy near-term on AI alone; in the longer run AI could raise productivity and the neutral rate, complicating policy trade-offs.

Analysis

AI capex is creating concentrated, idiosyncratic demand shocks rather than broad-based demand inflation — think localized spikes in electricity, copper, and real estate in key data‑center corridors rather than a national wage spiral. That pattern favors asset owners of immovable infrastructure (data‑center REITs, utilities, transmission/equipment OEMs) where pricing power and contract duration turn lumpy capex into multi‑year cashflow optionality; in constrained grids we should expect mid‑single‑digit percentage uplifts to peak load within 12–24 months in the most affected zones, amplifying utility earnings seasonally.

There is a multi‑year macro channel where faster productivity growth lifts neutral real rates and steepens the curve, benefiting banks via wider NIMs and pressuring long duration assets. But the transition is non‑linear: near term the capex cycle is financed and concentrated in mega‑caps (keeping broad CPI muted), while a later, broader adoption wave would shift labor markets and rents — a 2–5 year pivot that would force policy recalibration rather than a next‑quarter move.

Key tail risks are a macro slowdown that truncates capex (turning winners into inventory/lease risk), grid permitting or carbon‑policy constraints that cap data‑center expansion, and faster‑than‑expected diffusion that actually accelerates disinflation via productivity. Watch cross‑market signals: widening commercial real estate spreads in data‑center hubs, copper backwardation, and a steepening 2s10s as early indicators that the medium‑term monetary re‑pricing is starting to matter.

Key Decisions for Investors

• Buy Digital Realty (DLR) 12‑month call spread (buy $90 / sell $120) to play multi‑year contracted cashflows from data‑center expansion; risk = premium (limited), target = 2x premium if leasing/spot rents reprice upward within 9–18 months.
• Overweight NextEra Energy (NEE) (5–12 month horizon) via 1–2% portfolio overweight or buy 9‑month calls as a high‑conviction play on grid demand and long duration regulated returns; downside scenario: permitting/regulatory pushback compresses near‑term upside — cap losses to ~premium/dividend foregone.
• Long Freeport‑McMoRan (FCX) equity (12–18 months) to capture higher copper intensity from data‑center and grid buildouts; set stop at 20% drawdown and target 30–50% upside if copper tightens; tail risk is cyclical demand collapse that can halve the target.
• Pair trade: Long Bank of America (BAC) (6–12 months) vs short long duration Treasuries (e.g., small TLT position) to express a gradual rise in neutral rates from productivity gains; reward = wider NIMs and higher ROE for banks, risk = recession or rapid policy easing which would hurt BAC while rallying TLT (limit net exposure so drawdown capped to ~10%).