According to Goldman Sachs, corporate America is changing due to artificial intelligence, yet government growth statistics do not fully reflect this surge.
The size of the surge was highlighted by Goldman analysts in a report published on Saturday: Revenue at US firms that supply AI infrastructure has increased by $400 billion since 2022, which first appears to indicate that AI has been a significant contributor to economic growth in recent years.
However, official figures provide a different picture.
According to the analysts, since 2022, AI technology has increased real US economic activity by around $160 billion, or 0.7% of GDP. However, official data only show about $45 billion, or 0.2% of GDP, of growth fostered by AI. Analysts estimate that this leaves about $115 billion unaccounted for.
The gap emphasizes the divergence between what businesses report and what the government measures as a result of the Commerce Department’s Bureau of Economic Analysis technique to calculate growth.
According to the Goldman analysts, the BEA’s approach for determining GDP sees semiconductors as intermediate inputs that are only taken into account when the goods (such consumer laptops) that they allow are sold, meaning that the estimated impact of AI on GDP is probably considerably less.
Thus, intermediate inputs include high-performance semiconductors, which are the chips that drive AI training. The value of their imports is deducted from GDP, and their usage in the development of AI systems is not considered an investment.
But according to the analysts, the chips created in recent years are being utilized to train and assist AI models, thereby creating an intangible asset whose final production value has not yet been fully capitalized or quantified in GDP.
The anticipated $75 billion spent on cloud-based corporate solutions and AI model development has not been included in investment figures, according to Goldman’s analysts.
The situation was further complicated by new import regulations.
Business investment in information-processing equipment seemed to increase in the first half of 2025, primarily as a result of enterprises rushing to acquire networking equipment and servers before President Donald Trump imposed import taxes.
The pattern likely represents one-time frontloading prior to tariffs, which in turn exaggerates the typical demand for AI investment, analysts noted. Imports are deducted from GDP, hence the advantage of investment was somewhat offset.
It is difficult to determine AI’s influence on other crucial metrics. Businesses are also having difficulty demonstrating it in their financial results.
According to a separate survey released earlier this month by Goldman Sachs, a record number of S&P 500 companies discussed artificial intelligence (AI) on earnings calls in the second quarter of the year, although the proportion of companies measuring the impact of AI on profitability today is still limited.
