Is AI really boosting US economic growth?

Capital Economics finds the AI-driven boost to US growth so far concentrated in ICT industries—computer hardware, software and data processing contributed roughly 0.9 percentage points to GDP growth in H1 2025, with computer-hardware and data-center investment up about 40% year‑over‑year and software investment adding ~0.5 percentage points—while tech goods/service price declines have subtracted about 0.5% from the GDP deflator. Outside the ICT core adoption remains under 15% and there is little sign of AI-led productivity gains in major service sectors; ICT output is rising even as employment falls, signaling productivity gains but not broad-based job creation. Capital Economics cautions these gains may be cyclical (post‑COVID labor market effects) rather than structural, so claims of a decade‑long AI productivity miracle are premature and the economy remains in the early stages of the AI boom.

Analysis

Capital Economics finds the AI-driven boost to U.S. growth to date narrowly concentrated in ICT industries: computer hardware, software and data processing together contributed almost 0.9 percentage points to U.S. GDP growth in H1 2025, investment in computer hardware and data centers rose roughly 40% year‑over‑year, and software investment contributed about 0.5 percentage points to annualized GDP growth. Falling prices for tech goods and services have subtracted an unusually large 0.5% from the GDP deflator so far this year, underscoring tech-driven deflation within the sector.
The productivity uplift is sector-specific: adoption of AI in non‑ICT industries remains below 15%, ICT output is rising even as employment is falling, and overall payroll growth is running below 50,000 per month, which implies productivity gains in tech have not translated into broad employment or cross‑sector gains. Capital Economics cautions much of the improvement may be a cyclical response to post‑COVID labor tightness rather than a structural, economy‑wide transformation.
Market signals are mixed and cautious: the aggregated sentiment score is 0.05 with a modest market‑impact score of 0.25, and per‑ticker sentiment tilts mildly positive for NVDA, SMCI and APP. The firm concludes the U.S. remains in the early stages of an AI boom and that forecasts of a decade‑long productivity miracle are premature, so investors should treat tech‑capex beneficiaries as tactical opportunities rather than proofs of durable, economy‑wide outperformance.

Key Decisions for Investors

• Favor tactical overweight to ICT hardware and cloud/capex beneficiaries (e.g., firms aligned with GPU, data‑center and software investment) rather than broad market bets, maintain position sizing discipline
• Avoid assuming economy‑wide productivity gains; refrain from large, long‑duration exposure predicated on a decade‑long AI miracle and prefer shorter time‑horizon trades
• Monitor key indicators—non‑ICT AI adoption rates (noting the current <15% level), monthly payroll growth (current <50k), software/hardware capex trends, and GDP deflator moves—for signs the cycle is broadening
• Use hedges or diversified structures to protect against a reversion if AI gains prove cyclical (e.g., options protection or reducing concentrated tech long exposure), keep risk controls tight
• Watch company‑level revenue and margin reports for NVDA, SMCI and software vendors as leading signals of whether ICT capex strength is translating into sustainable earnings