Despite pessimistic forecasts regarding the return on investment of artificial intelligence, the cloud divisions of Amazon, Google, and Microsoft have demonstrated staggering growth. In the second quarter of 2026, the combined revenue of AWS, Google Cloud, and Azure increased by 48% year-over-year. For comparison, this figure stood at 39% in the previous quarter, and in 2024–2025, the companies remained in the range of 24–33%.
This dynamic breaks the usual logic of corporate growth. Typically, when a business reaches a turnover of hundreds of billions of dollars, its expansion rate slows down due to the high base effect. Here, however, we observe the opposite process: the larger the players become, the faster they grow. This is direct proof that the demand for AI infrastructure not only exists but is becoming explosive in nature.
Numbers That Speak for Themselves
A breakdown of the reports reveals impressive results for each segment. AWS revenue reached $42.2 billion, showing a 37% increase—the fastest pace in the last 18 quarters. The division's operating margin meanwhile stood at a solid 36.8%.
Google Cloud demonstrates even more telling dynamics: $24.8 billion in revenue and 82% growth. The division's operating profit more than tripled over the year, and the margin jumped from 20.7% to 35.6%. Azure, in turn, crossed the significant milestone of $100 billion in annual revenue, showing a 43% increase in constant currency.
Market forecasts are also being revised upward. The combined revenue of Amazon, Google, Meta, and Microsoft over a two-year horizon grew from approximately $1.6 trillion in 2024 to $2.5 trillion. Average annual growth rates relative to revenue over the last 12 months increased from 11% to 18%.
Skeptics' Arguments and Counterarguments
Pessimists, of course, have their reasons. According to PIMCO estimates, over the next two years, capital expenditures of the largest cloud capacity providers will absorb about 94% of their operating cash flow. In 2023, this share did not exceed 40%. However, Fidelity specialists view the situation differently: the ratio of capital expenditures to free cash flow across the broad market is currently below one, whereas at the peak of the dot-com era in 2000, it approached four.
This leads to a fundamental conclusion: the current cycle is financed by real profits, not by borrowed funds based on future expectations. Critics of AI's return on investment are mixing up two different things. The reports clearly indicate that the infrastructure layer has already proven its viability. Sales of computing itself, data storage, and AI services are growing, and profitability is growing along with them. The only open question remains how the entire corporate sector will earn from AI.
I see Oracle as the next beneficiary of this wave. The company's backlog of contractual obligations reached $638 billion—up 363% year-over-year. Contracts are signed in advance, although revenue from them has not yet appeared in the reports. When the data centers reach full capacity, the same profitability growth scenario will repeat, just with a time lag. Investors should closely monitor this case.