Here’s why AI is NOT boosting US GDP - despite billions pouring into the technology. Goldman Sachs says a surge in corporate AI spending has yet to translate into meaningful economic growth, thanks to inflation, overseas sales, and accounting quirks that downplay its true impact.
While public companies exposed to AI have seen their annualised revenues jump by more than $340bn since 2022 - and are projected to grow by a further $580bn by the end of 2025 - official data suggests that real AI-related investment in US GDP terms has only increased by $42bn over the same period.
This sharp contrast, according to Goldman Sachs, can be explained by inflation, overseas revenue, and the way national accounts classify AI-related spending.
Inflation and overseas sales distort the numbers
A significant portion of AI spending has gone toward semiconductors, where prices have risen sharply due to supply constraints. Goldman Sachs estimates that margin expansion alone accounts for $30bn of the reported AI revenue increase.
In addition, around $130bn of AI-related revenue growth has come from sales outside the US, meaning it does not contribute to domestic GDP figures.
Accounting rules are holding back AI's impact on GDP
There is also a technical issue with how AI investment is recorded in national accounts. The Bureau of Economic Analysis (BEA), which compiles US GDP data, classifies semiconductor purchases as intermediate inputs rather than investments, because they are traditionally embedded in products that are later sold.
This means that while AI firms are spending billions on high-end chips to train machine learning models, much of this spending is not captured in GDP.
Cloud computing faces a similar classification issue. The computing power used to train AI models is treated as an intermediate input rather than an investment, unless it is continuously used in production for over a year. Goldman Sachs argues this approach likely underestimates the real impact of AI spending on the economy by as much as $100bn.
A bigger impact could be seen in 2025
Despite these distortions, the bank sees potential for AI-related investment to make a more visible impact on US GDP in 2025.
Spending is expected to broaden beyond semiconductors and cloud computing to areas like data centres, networking hardware, and utilities, which are more likely to be captured as real investment in GDP accounts.
However, Goldman Sachs warns that unless the BEA updates its methodology, a significant portion of AI spending will continue to go unmeasured.
While AI is reshaping industries and driving stock market gains, its impact on broader economic growth remains difficult to quantify. Investors and policymakers hoping for a productivity boom may need to wait until AI investment is more widely distributed across sectors—or until statistical agencies adjust how they measure it.