AI-powered growth: GenAI as a new engine of US economic performance

Accurately measuring AI's overall economic impact is challenging, but we can gain valuable insight by focusing on the GDP categories most likely to reflect AI advancements such as software, R&D, information processing equipment, data centers and power infrastructure. AI-related business investment surged at an impressive annualized rate of 18% during the first half of 2025 and fueled a substantial 1ppt increase in GDP growth in the second quarter, following a 1.3ppt boost in the first quarter.

This AI-driven capex surge aligns with the baseline scenario we outlined within our 2023 series on the economic impact of AI for business leaders, “Catalyzing economic growth through AI investment.” That scenario showed a sustained acceleration in AI-related capital investment – growing 25% faster than the 2017–2022 trend – and lifting AI investment from 5.2% of real GDP in 2022 to 8.1% by 2028. Although AI investment came in below expectations in 2023 and 2024, in part due to macroeconomic challenges like higher interest rates and tighter financial conditions, 2025 is set to exceed those projections. Overall, the strong trajectory remains in place with GenAI investment now estimated to account for 6% of GDP in 2025.

While the headline numbers are encouraging, it’s important to note that not all AI-driven investment translates directly into GDP gains. Some software and hardware, including chips, are imported, which reduces their net contribution to domestic output. For instance, after accounting for trade in intellectual property and computer equipment, GenAI’s net contribution to GDP averaged a more modest 0.4ppt during the first half of 2025.

Contribution to q/q US real GDP growth from AI-related components

Q1 2022 - Q2 2025

Importantly, much of the AI-driven transformation remains understated in official statistics, owing to the intangible nature of many effects as well as the lag before productivity gains are fully realized. Efficiency improvements embedded in intermediate processes – such as automation or enhanced decision-making – only register in GDP once they translate into increased output of final goods and services.

This dynamic aligns with our earlier expectations: while headline GDP figures are starting to capture the AI investment surge, the full scale and speed of economic transformation will only become evident over time, as complementary innovations emerge and workforce capabilities and business processes evolve.

AI adoption has been accelerating at a strong rate across firms according to the biweekly Business Trends and Outlook Survey¹ (BTOS) from the US Census Bureau. The share of businesses reporting they are currently using AI to produce goods and services has risen from 3.7% to 10% since late 2023, with growth in adoption ranging from 44% annualized to nearly 98% across industries. While this surge reflects widespread enthusiasm for AI technologies, the rate of AI adoption is not uniform across sectors.

Overall share of US businesses reporting AI usage¹

September 2023 - September 2025

Leading the charge are industries such as information, professional, scientific and technical services, and finance where the integration of AI tools has become a strategic priority. These sectors benefit from robust digital infrastructure and have established high demand for automation and data-driven insights.

Industries like real estate, education and healthcare are steadily increasing their AI adoption, although adoption rates are more moderate due to factors such as legacy systems or regulatory hurdles.

In comparison, sectors like construction, accommodation and food services, transportation and traditional retail continue to lag, largely because their operations depend heavily on in-person and physical activities that are less conducive to AI integration at this time.

Share of US businesses reporting AI usage by industry¹

September 2023 vs. September 2025

Optimism around AI adoption has propelled US equity markets to record highs, even amid macro headwinds from tariffs, elevated policy uncertainty, high interest rates and immigration constraints. A few tech companies now account for over one-third of the total value of the S&P 500, and hyperscalers are pouring unprecedented amounts into building out AI infrastructure. The combined capital expenditures by the largest cloud providers more than tripled between 2019 and 2024, reaching $237b last year. Capital spending is projected to rise 39% this year and approach $330b.

Aggregate total capital expenditures of four of the largest hyperscalers

2011 - 2025F

The same optimism driving record equity valuations is translating into hard investment on the ground, as hyperscalers race to build the infrastructure that powers AI. As of July 2025, data center construction has surged to a record $41b annualized, marking a 30% increase compared to 2024. To put things into perspective, whereas office construction (general and financial) was eight times larger than data center construction in 2019, data center construction is set to overtake traditional office construction in the coming year.

Business investment captured in GDP data tells a similar story. US businesses are pouring billions into AI-enhanced hardware such as high-performance servers and data storage systems, with total investment in information processing equipment up 35% annualized in the first half of 2025.

US nominal construction spending (annualized)

Q1 2015 – Q4 2027F

Beyond physical structures, intellectual property investment is accelerating, underscoring a structural shift toward intangible capital as a driver of competitiveness. Business investment in software, including AI-related applications, cloud platforms and enterprise digital tools, advanced at an impressive 23% annualized pace in the first half of 2025. R&D spending is also picking up, as firms race to develop AI models, semiconductor technologies and advanced software. Overall, business investment that captures AI technology has surged by 48% since 2020, whereas traditional business investment has been roughly flat over the same period.

US real nonresidential fixed investment

Q1 2020 - Q2 2025 (2020 Q1=100)

AI’s impact on the labor market is unfolding in nuanced ways — reflected in a surge in AI-related job postings, slightly slower employment growth in sectors leading AI adoption and marked improvements in workplace efficiency. According to Indeed, the share of US job postings referencing AI has nearly doubled since mid-2023, though such listings still account for only about 3% of total openings.

Share of US job postings mentioning AI terms

January 1st, 2019 – September 30th, 2025

At the same time, US sectors reporting the highest AI adoption, such as the information or finance and insurance sectors, have tended to show somewhat slower job creation over the past two years. The correlation isn’t strong, but it highlights how current GenAI tools overlap with tasks common in these fields, making AI integration more likely, and potentially showing up in sector hiring and employment readings as a result.

AI adoption rate and change in total US payroll employment by industry¹

Hyperscalers offer a similar example: while capital spending on AI infrastructure has surged to record levels, employment growth at these firms has largely plateaued since 2021, underscoring the potential for a positive supply shock to translate into structurally reduced labor demand.

Anecdotal evidence also suggests that GenAI is beginning to influence labor productivity. A recent survey by the St. Louis Fed estimates that workers using GenAI saved an average of 5.4% of their work hours, translating into a 1.1% aggregate productivity boost if scaled across the economy.

Among frequent users, time savings were even more pronounced, with 33.5% of users saying it saved them four hours or more, compared to only 11.5% for users who used it only one day a week. While these gains may not yet fully appear in official productivity measures, they signal that AI is already enhancing efficiency at the firm level, laying the groundwork for broader economic impact.

This observation is consistent with our prior analysis in “The productivity potential of GenAI,” where we highlighted the lagged nature of productivity gains following major capital investment in new technologies. Just as the Information and Communication Technology (ICT) revolution of the 1990s required years of complementary innovation, workforce adaptation and process transformation before its full productivity benefits were realized, we expect the current wave of AI-driven investment to set the stage for a similar, and potentially even greater, productivity acceleration in the years ahead.

As adoption broadens and matures, and as organizations integrate AI more deeply into their operations and workflows, the productivity impact is likely to become increasingly visible in macroeconomic data – reinforcing our baseline scenario that GenAI-driven productivity is set to provide a substantial lift to the economy worth $650b over the next decade and lifting real GDP by nearly 2.5% by 2033.

Summary 

This is the seventh installment of the EY-Parthenon macroeconomic article series on the economic impact of AI. The series aims to provide insights on the economic potential of generative AI (GenAI), including new developments and actionable insights to arm companies’ decision-makers. The seventh article in this series covers how GenAI is fueling significant AI-powered growth in the US economy — with a rapid rise in AI adoption across industries.