Assenagon Study: AI Between an Investment Boom and Earnings Pressure

- Investment boom creates new dependencies: Billions spent on AI infrastructure support the US economy but leave it more vulnerable to weakening demand.
- Falling AI costs open up opportunities: New applications are becoming more economically viable. Broad-based productivity gains, however, have yet to materialise.
- Financial ties amplify risks: When suppliers help finance their customers, sales and the value of equity stakes can come under pressure simultaneously.
Nvidia’s latest quarterly results and Anthropic’s expected stock market debut today bring the central question of the AI boom into focus: When will the billions invested translate into sustainable earnings?
A new study by independent asset manager Assenagon examines AI’s economic potential and the dependencies arising from investment and financial ties. Its authors, Thomas Romig, CIO Multi Asset, and Sebastian Schmider, Head of AI Solutions & Macro Analytics, assess the implications for the economy and investors.
Adoption Is Growing Faster Than Measurable Benefits
AI is becoming increasingly affordable and capable. A comparison of models shows that using AI today can cost as little as one-seventieth of what it did at the end of 2024. This makes applications that were previously barely profitable more economically attractive. Lower costs can, however, also boost demand: according to the authors, more intensive use could drive up total spending over the medium term, even as costs per application decline.
AI adoption is advancing rapidly, outpacing productivity gains across the economy. In the euro area, the proportion of employees using AI at work rose from 26 per cent in 2024 to 52 per cent in 2026. Yet the vast majority of companies surveyed in an international study reported no productivity impact over the preceding three years.
Investment Boom Runs Ahead of Expected Earnings
The AI build-out is already supporting economic activity: equipment, software and new data centres are boosting investment demand. This alone, however, does not establish that lasting productivity gains will follow.
At the same time, the scale of investment is creating an economic dependency. Even slower spending growth could weaken the demand stimulus. If companies scale back their expansion plans, suppliers and other business partners could also come under pressure. Bottlenecks in energy, construction and equipment could further constrain the delivery of planned capacity.
Heavy Investment Raises Pressure on Future Earnings
The financial scale is particularly evident among the major cloud providers Alphabet, Amazon, Microsoft, Meta and Oracle. In 2025, their capital expenditure averaged around 32 per cent of revenue. In June, the Bank for International Settlements (BIS) estimated that this ratio would reach around 48 per cent in 2026. It also estimated that gross debt issuance by the five companies would total around USD 175 billion in 2026.
Financing and Sales Can Depend on the Same Customer
Close ties within the industry add another layer of complexity. Under so-called circular financing arrangements, a chip or cloud provider might supply capital to a customer while also selling it hardware or computing capacity. In the BIS sample of 1,246 AI companies analysed, such structures accounted for around 16 per cent of deals between AI companies from 2021 to 2025. Their share of disclosed deal value was 46.4 per cent.
One example is the partnership between Microsoft and Anthropic announced in November 2025: Microsoft committed to investing up to USD 5 billion, while Anthropic committed to purchasing USD 30 billion worth of Azure computing capacity.
“Such agreements can enable expansion and do not, in themselves, constitute evidence of unsustainable business arrangements,” says Schmider. These ties nevertheless carry risks: “Sales and the value of equity stakes depend on the success of the same customer.” If that customer’s earnings disappoint, orders and equity valuations could fall simultaneously, while financing could become more expensive. Such feedback effects could amplify the investment cycle.
Sustainable Earnings Matter for Investors
For equities, the authors argue that the key questions are which companies can generate lasting profits from AI and what expectations are already reflected in share prices. “The investment and productivity data examined do not support a blanket assessment of the attractiveness of AI stocks,” says Romig. “For corporate bonds, the focus is on the sustainability of cash flows, financing needs and obligations to business partners.”
In this environment, diversification requires an understanding of shared economic dependencies. Different companies and asset classes can be exposed to the same investment cycle—for example, when an equity investment in an equipment supplier and a bond issued by a cloud provider both benefit from continued capacity expansion.
The full study is available in the latest edition of Assenagon Perspectives.
Munich/Frankfurt, 19 November 2026
