BofA says AI productivity boost visible in narrow tasks, not yet economy-wide

BofA Global Research estimates AI is adding only about 0.1% per year to aggregate macro productivity today, with potential upside to 1.0% annually over the next decade if adoption broadens and costs fall. The report frames the current economic impact as limited by implementation lags, skills gaps, and organizational constraints, while still suggesting long-term GDP growth could reach 4.5% annualized in an optimistic case. Separately, BofA kept U.S. Q2 GDP tracking at 2.6% saar after April housing starts data, while final May University of Michigan consumer sentiment came in at 48.2 versus 48.0 expected.

Analysis

The near-term market implication is less about AI as a productivity boom and more about AI as a capex and margin-reinvestment cycle. If aggregate output is only inching up while adoption is already broad, the first beneficiaries are still the infrastructure layer: compute, power, networking, and software vendors that monetize “pick-and-shovel” demand before enterprise P&L uplift shows up. That argues for continued relative strength in the AI supply chain even if the broader economy does not yet re-rate on AI optimism.

The second-order risk is that slow macro transmission creates a valuation gap between beneficiaries and the rest of tech. As long as AI productivity gains remain task-specific, investors will keep paying for revenue tied to model usage and inference, while discounting end-demand improvements in labor-heavy industries. That is bearish for stocks whose AI story depends on downstream labor savings showing up in margins within 1-2 quarters; those savings may instead be delayed into 2026-27.

On the macro side, a stable GDP tracker and slightly firmer sentiment reduce urgency for immediate easing, which is supportive for cyclicals that need no policy rescue but also caps duration-sensitive multiple expansion. The bigger contrarian angle is that low measured productivity now may be setting up a later breakaway: once workflow redesign and workforce retraining catch up, the market may need to price a second-stage margin expansion across software, services, and some industrial automation names faster than consensus expects. The risk to that view is that organizational friction remains stubborn, making AI a long-duration theme with intermittent disappointment rallies rather than a clean straight-line trend.

Key Decisions for Investors

• Stay long AI infrastructure vs. broad software: long SMCI/NVDA or a semis basket against XLK equal-weight for 1-3 months; upside comes from continued capex spend even if macro productivity stays muted, while the risk is multiple compression if hyperscaler spending guidance rolls over.
• Buy on weakness in power/grid beneficiaries: initiate longs in VRT and ETN over 6-12 months as the market keeps underpricing data-center electricity and cooling bottlenecks; risk/reward is attractive because the demand driver is physical and less dependent on AI hype cycles.
• Fade near-term beneficiaries that require rapid enterprise ROI: consider shorting a basket of lower-quality application software names with rich AI narratives versus MSFT or ORCL for the next 1-2 quarters; the thesis is that monetization lags adoption, pressuring names with the most aggressive AI expectations.
• Use a call spread on IWM vs. QQQ only if rates start easing: the stable growth print lowers recession odds but does not yet justify a small-cap leadership shift; this is a tactical trade, not a structural one, and should be paired with a catalyst on policy pivot.
• Maintain a tail hedge against AI disappointment: buy 3-6 month put spreads on a high-beta AI proxy basket into strength; if enterprise AI ROI narratives slip another quarter, the unwind can be sharp even without a broad macro downturn.