Wall Street has finally had an epiphany: Is the moat of AI giants in the "cloud"?

Over the past few months, investors have been fretting over how big tech firms can recoup returns from their massive, sweeping investments in AI.

Now, cloud computing seems to have emerged as that "ultimate answer".

Amazon, Microsoft, and Google under Alphabet all run cloud computing businesses that are seeing rapid growth amid the AI boom. The latest financial reports show that clients including AI labs and large enterprises are rushing to rent chips and computing equipment from them — to the point that what is limiting business expansion is no longer market demand, but the limit of production capacity.

Unlike those previous business ideas with uncertain monetization paths, such as chatbot ads or model subscription systems, the profit model of cloud computing has long been fully verified by the market and is extremely easy to understand:

These tech giants purchase or lease data centers and computing equipment, then rent them out through their cloud businesses, recouping their costs within a few years. While this is a capital-intensive business, it boasts extremely high profit margins: Amazon's cloud division AWS announced an operating margin of as high as 39% in the second quarter last Thursday.

Andy Jassy, CEO of Amazon, is currently trying to build AWS into a "perfect image" that is both stably risk-resistant and has extremely strong explosive growth potential. He revealed that the company only needs an average of less than three years to recoup costs for its computing power equipment purchases, and most of the computing power contracts signed with AI customers last for more than five years.

This means that the second half of these long-term lease contracts is all pure profit "cash cows". "The resulting revenue, free cash flow and return on invested capital are very attractive," Jassy said.

Wall Street's "Enlightenment"

While being in this frantic period of heavy investment in computing power also means that Amazon's free cash flow is currently negative — the market has long been extremely sensitive to the high spending of hyperscale cloud service providers recently, to the point that investors will sell off their shares as soon as they see cash outflows.

However, more Wall Street figures seem to have finally come to their senses last week:

As long as there is a cloud computing business behind it that can steadily absorb the dividends, the crazy spending of tech giants on AI is completely justifiable.

After AWS delivered a stunning performance report with a 37% year-on-year surge in quarterly revenue — far exceeding analysts' expectations of 31%, Amazon's stock price skyrocketed last Friday. Microsoft also ushered in a capital carnival last Wednesday, with its Azure cloud computing business achieving an unexpected 43% growth, driving its stock price to a sharp rise. The combined market value of just these two companies surged by about 950 billion U.S. dollars after the release of their financial reports.

Alphabet, Google's parent company, briefly suffered a stock price slump last month after it sharply raised its full-year capital expenditure guidance, but its cloud computing business's impressive 82% growth quickly calmed the market's panic afterwards. Its share price rebounded by about 10% in the past week, fully recouping all previous losses.

AI may be the most important driving force behind cloud computing growth — enterprises and AI developers need stronger computing power, and renting computing resources via the cloud is both simple and often the fastest option. Huge deals with large customers are also boosting revenue and backlog growth — for example, the 10-year contract worth more than 100 billion U.S. dollars that AWS signed with Anthropic in April this year.

In fact, long before the AI boom, another trend had been quietly and continuously unfolding.

Enterprises are shifting more computing workloads to the cloud and reducing their reliance on self-owned equipment. A survey of IT executives conducted by Piper Sandler in June this year shows that they expect their spending to increase by about 5% this year. Most people plan to increase their investment in cloud computing, while relatively few plan to increase spending on self-owned computing equipment.

A New Rival Rises in the Cloud Track

Given these two major trends, it is not hard to understand why Jassy said last Thursday that AWS is expected to eventually grow into an enterprise with an annual revenue of 1 trillion U.S. dollars — although this figure may sound a bit outrageous at present. According to data from FactSet, analysts expect AWS's revenue to rise to about 170 billion U.S. dollars this year.

Other companies are also trying to grab a large share of this expanding cake.

For some time, the cloud businesses of Microsoft and Google have been growing faster than Amazon, even posing a real threat to AWS's dominant position in the industry that it practically pioneered 20 years ago. If their growth rates continue to outpace AWS, it is not hard to foresee that they will overtake this larger rival by the end of this decade.

Of course, if the AI bubble bursts, all these companies will face difficulties — if Anthropic, OpenAI and other big buyers of AI computing power run into trouble, huge contracts are likely to be renegotiated, and order backlogs will shrink.

However, for cloud computing giants, there is still a fallback: at the very least, they can fall back on the time-tested business model that supported their development before the rise of AI.

But for pure AI players that lack the moat of cloud business, there is no such fallback — the most notable example is Meta. After the social media giant released its financial report last Wednesday and slightly raised the median of its 2026 capital expenditure range, its share price fell by about 5% immediately.

Although Mark Zuckerberg stated last Wednesday that Meta is considering building its own cloud computing business, facing the three giants that have already established absolute barriers, it may be destined to struggle to catch up from far behind.

It can be said that in this ultimate choice to distinguish AI winners from losers, Wall Street finally found that crucial measuring scale last week.

This article is from the WeChat official account "CLS AI Daily", author: Xiaoxiang, published with authorization from 36Kr.