Futurum Equities CEO Daniel Newman expressed his views on the return on investment for the largest infrastructure buildout in history.
Newman took to X on Thursday to emphasize the importance of patience, stating that expecting massive returns in the initial two years is “idiotic”. He said the AI boom is still in its early stages, arguing that the most valuable applications of the technology have yet to emerge and that its full potential remains largely untapped.
He urged a longer-term perspective, preferably a 5-year horizon, as more reasonable, particularly in relation to the development of artificial intelligence (AI).
“We've barely scratched the surface of AI. The highest value use cases are in their infancy,” Newman wrote.
The Math Behind AI’s Big Bet
Newman referenced a chart from a Financial Times article in May, wherein Joachim Klement, the managing director at Panmure Liberum suggested that AI companies are planning unprecedented capital spending growth of about 20% annually over the next five years, while revenues are projected to rise 15% a year.
Even assuming unrealistically that there are no operating costs and all additional revenue becomes profit, the expected return on investment from AI data center spending appears strongly negative for most companies, with Amazon.com (NASDAQ:AMZN) being the notable exception, as Microsoft Corporation (NASDAQ:MSFT), Meta Platforms (NASDAQ:META) , Alphabet Inc. (NASDAQ:GOOGL) (NASDAQ:GOOG) and Oracle Corporation (NYSE:ORCL) stayed in the negative.
There are two potential ways for AI companies to justify their massive spending, according to Klement: either AI generates far more revenue than currently expected, or planned investments are scaled back. However, achieving a reasonable return would require an additional $2 trillion to $5 trillion in annual revenue, a daunting target for companies that currently generate about $1.5 trillion combined. Alternatively, data center and chip investments may not fully materialize if investor enthusiasm wanes or financing becomes more difficult.
AI Expansion Tests Supply Chain Capacity
The ongoing debate around the massive capital expenditure (capex) being poured into AI, intensifies, meanwhile. In May, Goldman Sachs lifted its 2026 forecast for U.S. business investment to 7.8%, driven by AI spending, which is tracking towards more than $800 billion by year-end.
However, there are concerns about the physical-world constraints behind the AI buildout. As Jordi Visser of 22V Research points out, the AI boom may run out of infrastructure before it runs out of money. Visser said only 12% to 18% of the expected $8 trillion AI infrastructure buildout has been completed, yet supply-chain strains are already emerging.
He highlighted shortages in key components such as memory chips, cooling systems, copper, fiber and power equipment, warning that companies with large order backlogs could face revenue-recognition risks if products are delayed despite investor expectations.
Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.
Image via Shutterstock
Login to comment