AI, GDP, and Corporate America - September 29, 2025

The boom in artificial intelligence spending in the United States is growing at a remarkable rate. But is it already contributing significantly to US GDP? And is there a risk of overinvestment that could weaken Corporate America?

The revenues of US companies providing AI infrastructure have increased by around $400 billion since 2022. As this figure represents almost 10% of the increase in US GDP over the period, one might think that AI is already having a significant impact on US growth.

However, this is not the right way to look at it. As Goldman Sachs experts point out in a recent study, this figure needs to be adjusted for the revenues of entities that sell and produce outside the US ($150 billion), and take into account inflation and imports of inputs. This leaves an impact of $160 billion on GDP in volume terms, representing a contribution of 0.7% to US growth since 2022 (+0.3% per year).

However, only part of this impact appears in the official statistics. The Bureau of Economic Analysis (BEA) still considers semiconductors (including Nvidia chips) as mere intermediate inputs, rather than investments. Similarly, new investment items – such as cloud computing dedicated to training AI models – are not accounted for as such. As a result, official US GDP has only been boosted by $45 billion, or +0.1% per year since 2022 (Figure 1).

This situation brings to mind economist Robert Solow's famous remark in the 1980s: “You see computers everywhere, except in productivity statistics.” If Goldman Sachs' figures are accurate, a similar gap between the real economic impact of AI and its statistical measurement could be emerging.

One thing seems certain: investment in AI will continue to grow in the coming quarters, driven by spectacular announcements from major players in the sector. The comparison with the Internet and mobile phone boom of the late 1990s is tempting. At the time, the investment frenzy caused a sharp increase in Corporate America's financing gap: companies had to raise massive amounts of funds through loans and bond issues.

Today, the situation is different. AI spending is mainly financed by the colossal cash flows of hyperscalers (Microsoft, Google, Amazon, etc.), whose profits are more than comfortable. Data published in the second quarter show that, overall, US companies continue to invest less than they save, and their profit margins remain high (Figure 2). In other words, while the risk of overinvestment may exist in the medium term, it does not yet appear to be an immediate danger.

In our global funds, AI remains a key component of our investments, with an equally weighted basket (known as “new capex”) of 40 global companies—including 20 US companies—that are directly or indirectly involved in the artificial intelligence value chain.