AI’s Surprisingly Limited Macro Footprint (So Far)

Summary

• The AI revolution’s footprint has so far been much bigger in markets than in the real economy.
• The Magnificent Seven’s extremely large capex spending has a high import component that limits its GDP impact.
• IT employment is gently falling but represents only about 2% of private employment.
• As AI adoption spreads to the whole economy long term, the 1990s precedent suggests further declines in labour’s income share.

Market Implications

• I still expect two-three 2025 cuts and at most one 2026 cut against markets pricing two and three cuts, respectively.

Early Stages of New Wave of Technological Innovations

Here, I discuss the AI revolution’s current macro footprint.

The new generation of AI models and their LLMs (large language models) subsets have unlocked new capabilities in the handling of unstructured data such as text, images or music and in content creation. As a result, they have the potential to automate more complex tasks and to handle more complex datasets than previous generations of machine learning programs. They could also deeply change the way people interact with computers.

AI stands to become a revolutionary general-purpose technology – a pervasive, productivity enhancing technology with broad applications across sectors. Previous examples include electricity, digitalisation, or the internet.

In an economy as complex and financialised as the US, adoption and diffusion of AI is unlikely to be a linear process, as current market gyrations show. AI’s economic potential has triggered a gold rush, with just about every business claiming AI use and capabilities as well as an exponential growth in AI users (Chart 1). Yet the evidence of sustainable, profit-generating AI-based business models remains limited, which is to be expected at an early stage in such a deep technological revolution (Chart 2). For now, the strongest business case seems to be coding though many more profitable ideas are no doubt being developed.

That AI has captured investors dreams can be seen in the extreme outperformance of Magnificent Seven stocks (Chart 3). Their earnings dwarf the rest of the markets, though it is unclear whether what share of these earnings is derived from AI and whether the outperformance is sustainable. The continuation of current trends would see the Magnificent Seven become the only profitable corporates in the US!

Regardless, current Magnificent Seven capex plans are so large that they are impacting GDP data.

Chart 1: We Are All Using AI…	Chart 2: …But Is There a Profit-Making AI-Based Business Model Yet?
Chart 3: Magnificent Seven’s Extreme Stock Outperformance	Chart 4: Will Magnificent Seven Capture All US Profits?

Magnificent Seven Capex Has High Import Content

US GDP growth is slowing (Chart 5). It averaged 1% QoQ SAAR in H1 2025, compared with 2024’s 2.5% average (I averaged Q1 and Q2 2025 GDP to remove the distortion created by tariff frontrunning). In addition, the drivers of growth have changed. Consumption typically drives US growth, which represents 70% of GDP. But consumption has been flat since end-2024, largely because households want to stabilise consumer debt. By contrast, capex has taken off. It contributed twice as much to growth as consumption in H1 2025, even though it represents only 15% of GDP.

A decomposition of capex spending shows it was driven by software and IT equipment spending (Chart 6). It is very likely these high numbers represented the impact of Magnificent Seven’s extremely large capex spending, representing nearly $600bn in the year to June 2025 (Chart 7). Compared, in the year to June 2025 GDP accounts final spending in total and on IT equipment was $1.6tn and $0.6tn, respectively.

However, IT equipment capex has a high import content. Excluding net imports, IT equipment capex has been falling since 2024 (Chart 8). This suggests the large IT equipment contribution to growth was more than offset by higher net imports (Charts 5 and 6).

This implies that even if the Magnificent Seven’s extremely large capex plans are unsustainable, possibly due to a market selloff, the actual impact on GDP will be limited.

Chart 5: Capex Lifted H1 Growth	Chart 6: Software, IT Equipment Are Up
Chart 7: Extreme Magnificent Seven Capex Spending	Chart 8: IT Capex Import Intensive

AI impact on the labour market has also been limited so far.

Will AI Lower Labour Income Share?

Despite the AI revolution and its labour saving potential, the labour market has remained in balance, though this partly reflects slower labour supply growth caused by tighter immigration policies.

Also, the AI revolution’s impact on the structure of employment seems limited (Chart 9). Since the pandemic:

• The biggest source of job growth has been in health and education, unrelated to AI.
• Manufacturing and trade transportation and utilities have continued falling relative to total employment, aligning with pre-pandemic trends and largely due to productivity improvements.
• Businesses’ services employment share has contracted but this largely reflects a decline in temporary employment and to a lesser extent IT employment (Chart 10). The decline in IT employment could reflect AI, especially since one of the proven business cases for AI is computer programming. However, IT employment represents less than 2% of total private employment.

Furthermore, post pandemic, the hourly wages of IT workers have struggled to keep up with private sector wages, which could also reflect AI’s labour-saving impact (Chart 11). AI raises productivity and eliminates jobs, with an uncertain impact on overall labour income.

Overall, the share of IT workers’ wages in total private sector wages has been falling since 2022, the first time since the 1990s IT bubble burst. This more likely than not reflects AI’s impact (Chart 12).

As the broader economy adopts AI, risk exists of a further decline in workers income share and increase in inequalities. The precedent is the internet revolution of the late 1990s that accelerated the decline in workers’ income share. This time policy interventions could be needed to prevent further declines.

Chart 9: Fastest Job Growth in Health and Education	Chart 10: Gentle Decline in IT Employment
Chart 11: IT Wages Slower Post-Covid	Chart 12: Will AI Eat Up Workers’ Share?

Market Consequences

At this stage in the revolution, AI has only a limited impact on growth and inflation. Magnificent Seven capex spending has a high import component that limits its overall impact on GDP. Similarly, there are early signs AI could weaken the labour market but so far this has been limited to IT employment that is a small share of total employment.

Therefore, this analysis does not change my expectations on Fed policy. I still expect two-three 2025 cuts and at most one 2026 cut against markets pricing two and three cuts respectively.

The commentary contained in the above article does not constitute an offer or a solicitation, or a recommendation to implement or liquidate an investment or to carry out any other transaction. It should not be used as a basis for any investment decision or other decision. Any investment decision should be based on appropriate professional advice specific to your needs. You are not permitted to publish, transmit, or otherwise reproduce this information, in whole or in part, in any format to any third party without the express written consent of Macro Hive. This includes providing or reproducing this information, in whole or in part, as a prompt.)