Is AI driving US growth? (am.jpmorgan.com)

AI-related capital spending is emerging as a measurable engine of U.S. growth: in H1 2025 AI-directed capex contributed roughly 1.1 percentage points to GDP growth and in Q2 tech-related categories added 4.3 points to investment growth. Hardware is leading the charge—spending on computers and related equipment is up about 41% year‑over‑year, data center construction hit a record $40 billion annual rate (≈ +30% YoY), and hyperscalers (Meta, Alphabet, Microsoft, Amazon, Oracle) are projected to spend about $342 billion on capex in 2025. Private AI firms such as OpenAI and Anthropic are also investing heavily to scale servers, GPUs and networking for frontier model training. For the AI/ML community, this signals durable infrastructure tailwinds—more data centers, chips and networking capacity that enable larger models and faster experimentation—but with important caveats. Current official data mainly capture the first-wave hardware buildout; the next phase (power, grid upgrades, reshoring) will take years and faces permitting, supply-chain and import-/GDP accounting limits. Data centers also have lower employment multipliers, and hyperscaler capex can be volatile if demand forecasts soften. Practitioners and investors should therefore view AI spending as a meaningful growth cushion and catalyst for tooling, MLOps and model scale, while remaining mindful of macro fragility and execution risks.