AI Optimism Drives Concentrated Corporate Investment

The Federal Reserve Bank of San Francisco Economic Letter, authored by Aakash Kalyani and Huiyu Li, reports that U.S. business spending related to artificial intelligence expanded markedly in 2025 among publicly traded firms. According to BEA data cited in the Economic Letter, spending on information processing equipment, software, and data center construction accounted for one-third of all business investment in the third quarter of 2025, the highest share since 1947. The Economic Letter states that observed increases in capital spending and research and development funding were concentrated entirely in the largest public firms that register positive sentiment about AI. The authors also report that data center investment remains roughly 1% of private nonresidential fixed investment, citing BEA figures.

The Economic Letter uses sentiment measures from quarterly earnings calls to infer evolving optimism about AI. Industry-pattern observations: researchers often use earnings-call text as a near-real-time indicator of corporate sentiment, but these measures are biased toward public firms and can be dominated by high-volume speakers. Sentiment signals capture tone and attention, not precise project budgets, so linking call-level positivity to specific capital expenditures requires careful aggregation and robustness checks.

Editorial analysis: Concentration of AI-linked investment within the largest firms, as reported by the Economic Letter, amplifies questions about the distributional effects of AI on aggregate demand and productivity growth. In macro statistics, large-firm swings in technology capex can materially affect headline investment series because these firms account for a disproportionate share of total spending. At the same time, the Economic Letter highlights that narrowly defined data center investment remains a small share of fixed investment, indicating substantial portions of information-processing spending reflect broader hardware and software upgrades rather than pure cloud capacity buildout.

Editorial analysis: Observers should track a few indicators to assess persistence and breadth of AI investment:

• •BEA revisions to industry-level investment and capex shares for 2025-2026
• •Firm-level R&D and capital spending by decile, to confirm whether diffusion beyond top firms occurs
• •Trends in earnings-call sentiment toward AI for midcap and small-cap public firms
• •Data center construction starts and cloud-provider capex disclosures, to reconcile the small BEA share with reported hyperscaler investments

Industry observers and practitioners will gain the most clarity by combining official BEA aggregates with firm-level disclosures and text-based sentiment metrics, as the Economic Letter demonstrates.