When investors think about artificial intelligence, their minds tend to jump to chipmakers or buzzy startups promising the next big breakthrough. But this earnings season revealed a quieter, potentially more profitable story playing out among the tech giants building the infrastructure behind it all. Amazon, Alphabet, and Microsoft demonstrated that they are already converting AI demand into real revenue and profit, not just promises. And behind that performance sits a staggering number that explains why these companies may still be the smartest plays in the space: a combined cloud backlog of more than $2.3 trillion, up from roughly $800 billion just one year ago.
What makes that figure so compelling is that it does not represent hopeful projections or speculative forecasting. These are signed customer commitments stretching years into the future, and many of them were negotiated before AI infrastructure became scarce and computing prices began their steep climb. Over the next two years, a large portion of those contracts will come up for renewal in a market where GPU capacity has become one of the world’s most valuable commodities. Rental prices for Nvidia’s H100 processors have surged about 63 percent over the past year despite newer chips entering the market, an almost unheard-of dynamic in technology where hardware typically gets cheaper over time.
Both Microsoft and Google acknowledged during their earnings calls that they remain capacity constrained, meaning customers want more AI computing than the companies can currently deliver. That is a luxury few capital-intensive businesses ever experience, and it helps explain why combined 2026 capital spending plans for the major hyperscalers now total somewhere between $740 billion and $770 billion, nearly double what they invested the prior year. Normally such aggressive spending would make investors nervous, but this cycle looks different because these companies are expanding against contracts that have already been signed.
Oracle has benefited from some of the same demand trends, but its heavier reliance on debt to finance infrastructure buildout sets it apart from its larger peers and introduces greater financial risk if growth slows or construction timelines slip. Its stock has been nearly cut in half over the past year. By contrast, Amazon, Alphabet, and Microsoft generate enormous free cash flow that funds expansion internally, giving them far more flexibility. Their cloud businesses already produce tens of billions in annual operating income, yet rising compute prices paired with supply-constrained demand suggest those profits could accelerate faster than most investors anticipate.
None of this means AI leadership is guaranteed or that competition will not intensify. But unlike many AI investments built on what might someday be possible, these three hyperscalers are pouring capital into serving demand that customers have already committed to buy. For long-term investors looking past the hype cycle toward durable returns, that distinction may make all the difference.
