Goldman Sachs AI Adoption Tracker: US Enterprise AI Adoption Rises to 21.5% in July 2026

Overview

This report updates Goldman Sachs' AI Adoption Tracker through July 2026, tracking AI investment, enterprise adoption, labor-market effects, and productivity impacts. The core conclusion is that AI-related investment remains strong, particularly across the semiconductor and hardware chains; US enterprise AI adoption continues to rise; labor-market impacts have emerged in some technology, marketing, design, and customer-service occupations but have not yet generated broad unemployment pressure; and evidence of productivity gains is strong in specific deployment settings.

Core views

AI adoption and investment momentum continue to strengthen. US enterprise AI adoption reached 21.5%, with large enterprises significantly ahead; the information, professional services, and finance industries remain at the forefront, while subsectors such as computing, broadcasting, and web search have adoption rates of 50% or higher. On capital expenditure, semiconductor revenue, Taiwanese AI hardware shipments, US AI hardware imports, and AI hardware investment in the US national accounts all indicate expanding demand. On labor, employment drag from AI remains concentrated and is partially offset by growth in data-center-related construction employment.

Analysis framework

The report combines the US Census Bureau Business Trends and Outlook Survey (BTOS), industry revenue forecasts, US national accounts, Taiwanese hardware shipments, US import data, job postings, keywords from corporate earnings calls, layoff announcements, academic research, and corporate case studies to establish four tracking lines covering AI investment, adoption, labor, and productivity.

Methodology notes

• Macro and industry trackingAI Adoption Tracker

  Multidimensional tracking of AI investment, adoption, labor, and productivity

  High-frequency surveys, official statistics, industry forecasts, and corporate disclosures are used to observe the pace of AI diffusion and its economic impact.

• Enterprise surveyCensus Bureau Business Trends and Outlook Survey (BTOS)

  US enterprise AI adoption rate

  The report uses the BTOS to measure AI use in daily business functions among US institutions. The wording of the survey questions changed on December 4, 2025, causing a jump in the level; Goldman Sachs continues to track the series using the new definition.

• Automation exposureGS AI Automation Exposure Scores

  AI automation exposure by subsector

  Based on the methodology of the March 2023 report “The Potentially Large Effects of Artificial Intelligence on Economic Growth,” Goldman Sachs estimates the share of work in each subsector exposed to AI automation and uses it to explain differences in AI adoption rates.

• Productivity evidenceAcademic studies and company anecdotes

  Generative AI productivity gains

  The report compiles academic studies and corporate cases, showing average efficiency or productivity gains of approximately 23% and 32%, respectively, although the evidence primarily comes from limited settings where AI has already been deployed.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Global semiconductors

  Core beneficiary chain of expanding AI investment

  Strengths

  Global revenue is expected to reach an annualized $834 billion by the end of 2026, with significant upward revisions to memory-manufacturer revenue forecasts.

  Weaknesses

  The report provides no company-level earnings, valuation, or rating views.

  Comparison

  The upward revision to memory-related forecasts is particularly notable within the AI hardware chain.

  Risks

  A slowdown in AI capital expenditure, order pull-forward, the inventory cycle, or overly high expectations could weaken revenue realization.

• AI hardware and Taiwanese shipment chain

  Reflects demand for AI infrastructure construction

  Strengths

  Taiwan AI-related hardware shipments have grown 72% since 2022, while US net AI hardware imports rose to $55.7 billion in May.

  Weaknesses

  Hardware demand may be affected by the capital-expenditure pace of large cloud providers.

  Comparison

  Compared with software, hardware investment is more directly reflected in current macroeconomic data as capital-expenditure expansion.

  Risks

  Supply-chain bottlenecks, trade policy, and demand volatility could affect the sustainability of shipments.

• Information, professional services, finance, computing, broadcasting, and web search industries

  Industries leading in AI adoption

  Strengths

  Information, professional services, and finance remain leaders; adoption rates in computing, broadcasting, and web search subsectors have reached or exceeded 50%.

  Weaknesses

  Industries with high adoption rates are also more likely to face pressure from job displacement and workflow restructuring.

  Comparison

  Adoption among large enterprises is significantly higher than among small enterprises, indicating uneven diffusion.

  Risks

  Higher adoption rates may not immediately translate into higher margins; the report finds no current relationship between company-level AI exposure and margin growth.

• Data-center-related construction employment

  Employment beneficiary of AI infrastructure investment

  Strengths

  Related employment has increased by approximately 290,000 workers since 2022, with the trend over the latest six months showing an increase of approximately 14,000 workers per month.

  Weaknesses

  Employment growth depends on the data-center construction cycle, and its persistence depends on subsequent capital expenditure.

  Comparison

  This increase partially offsets employment drag in industries affected by AI.

  Risks

  If AI infrastructure investment cools, the boost to construction employment may weaken.

• AI-affected occupations and technology employment

  Potentially pressured area from AI substitution and efficiency gains

  Strengths

  Evidence of productivity gains is relatively strong, and efficiency may improve in specific workflows.

  Weaknesses

  Employment drag has been observed in marketing, graphic design, customer service, and some technology roles; the share of technology employment remains below its long-term trend.

  Comparison

  The correlation between broad labor slack indicators and AI adoption remains limited, but early signs of a negative relationship have emerged between employment growth, hours growth, and AI adoption.

  Risks

  Accelerating AI-related layoffs, volatility in unemployment among young technology workers, and corporate automation could create structural employment pressure.

Key data

• US enterprise AI adoption rate21.5%Up 0.9 percentage points from June; expected to reach 24.3% over the next six months.
• AI adoption rate among large institutions38.3%Institutions with more than 250 employees, using the average of the latest six surveys.
• Large- versus small-enterprise adoption gap18.2 percentage pointsInstitutions with more than 250 employees relative to those with fewer than 10 employees; the gap has widened from 13.2 percentage points since February.
• Share of surveyed US employees working at organizations that have adopted AI47%Up from 41% in the previous quarter.
• Global semiconductor revenue forecast$834 billion annualizedEquity analysts expect this level to be reached by the end of 2026.
• Upward revision to memory-manufacturer revenue forecasts$174 billion annualizedThe cumulative upward revision in consensus forecasts since ChatGPT was publicly launched in November 2022.
• Incremental US AI hardware investment$463 billionThree-month annualized average in the US national accounts, approximately 1.4% of GDP above 2022 levels.
• Growth in Taiwan AI-related hardware shipments72%Growth since 2022.
• US net AI hardware imports$55.7 billionContinued to rise in May.
• Incremental data-center-related construction employment290,000 workersThe increase relative to the broader construction-employment trend since 2022; the trend over the latest six months is approximately 14,000 workers per month.
• AI-related layoffs14,000 in June; 102,000 in the first half of 2026The number of employees attributed to AI in corporate layoff announcements.
• Active US AI-related job postingsMore than 80,000The number of US AI-related job postings.
• Share of Russell 3000 companies mentioning AI and labor keywords25%The share of earnings calls for the second quarter of 2026 recorded to date.
• Evidence of productivity gainsApproximately 23% in academic studies; approximately 32% in corporate casesReflects average productivity or efficiency gains from generative AI in deployed fields.

Impact & implications

For investment implications, continued AI demand supports the semiconductor, hardware, data-center, and related infrastructure chains, while the diffusion of AI may drive productivity improvements in some industries. From a macro perspective, labor-market impacts have not yet become widespread, but technology, customer service, marketing, and design roles face structural pressure. Investors need to distinguish between aggregate demand expansion driven by AI, redistribution of market share, and potential employment-substitution risks.

Risks

• The wording of the BTOS survey questions changed on December 4, 2025, causing a jump in the level; historical comparability requires caution.
• No statistically significant and comprehensive relationship has yet emerged between AI adoption and broad unemployment or employment growth, leaving macro conclusions uncertain.
• Expectations for semiconductor and AI hardware investment may be affected by cloud-provider capital expenditure, supply chains, inventories, and changes in macroeconomic demand.
• Evidence of productivity gains comes primarily from limited deployment settings and may not extrapolate linearly to all industries.
• AI-related layoffs and job displacement may intensify structural pressure on specific occupations and young technology workers.
• The report provides no individual-stock ratings or target prices; an industry theme is not equivalent to a recommendation for any single security.

What to watch

• Whether US enterprise AI adoption can rise from 21.5% to the expected 24.3% over the next six months.
• Whether the AI adoption gap between large and small enterprises continues to widen.
• Whether global semiconductor revenue forecasts, memory-manufacturer forecast revisions, Taiwanese AI hardware shipments, and US net AI hardware imports continue to rise.
• Marginal changes in AI-related hiring, layoff announcements, employment growth, hours growth, and unemployment among young technology workers.
• Whether data-center-related construction employment can continue to offset employment drag in industries affected by AI.
• Whether official productivity data in high-AI-adoption industries continues to improve and whether corporate margins begin to reflect AI-related benefits.