NEW YORK, NY, NOVEMBER 7, 2026 — This year, the largest U.S.-based technology firms have made a record-breaking amount of capital expenditures for new data centers. In total, the four largest cloud providers are estimated to spend nearly $745 billion on their infrastructure between now and 2026. A large portion of these investments will go toward supporting the processing needs of artificial intelligence (AI) applications, reflecting trends in enterprise cloud computing.
The Scale of the Spending
In addition to being responsible for much of the money that is funding the build out today, Microsoft, Amazon, Alphabet and Meta have seen their total capital expenditure rise by 84 percent in just one year (2024-2025). The growth rate of their total capital expenditure has continued to grow since then, based on their latest financial reports.
The total amount spent by these four companies in the second quarter was over $171 billion. This amount was 78.6 percent greater than what was spent by the same four companies during the second quarter of the previous year. All of the four companies have now revised their projected spending for the entire year upward through their most recent earning reports, tracked via equity market reports.
In addition, Microsoft is expected to invest approximately $190 billion into data center development and other related infrastructure during the current fiscal year. In terms of overall global AI-related infrastructure investments, industry experts anticipate that total investment in this area could potentially increase up to $500 billion by the end of 2026. This projection is significantly higher than previous years’ investment totals.
Where the Money Is Going
A significant amount of money being spent is on chips and custom designed AI accelerator cards instead of on traditional server hardware. It has been estimated that around $180 billion of what is expected to be spent this year by large cloud service providers is for GPU’s only. The remaining funds are used to fund power systems, network infrastructure and the buildings themselves where the hardware resides, analyzed in corporate financial filings.
Construction of the massive Abilene, Texas data center campus linked to a significant cloud partnership is ongoing. That last building on the campus is projected to be complete before the end of the current year. Other similarly sized projects have begun across many U.S. states as well as in other nations around the world.
One company will be investing over 33 billion euros into Spain’s cloud and artificial intelligence (AI) infrastructure by 2035. A second company has invested $69 million in an AI-ready facility that opened in India this year. According to executives, the proportion of their total budget being spent on international expansion is increasing.
Why Demand Keeps Climbing
Enterprise customers have been demanding cloud capacity for both AI training and real-time inference workloads. AI training occurs periodically when a company develops a new model, whereas real-time inference produces continuous income. This has led cloud providers to increase their capacity designed exclusively for AI applications, highlighted in tech industry forecasts.
A leading provider is rapidly expanding its cloud infrastructure business with a partner-based growth model. The provider’s infrastructure division expects to nearly double its computing capacity by the end of 2022. Small-scale providers are also competing to acquire power and property to build additional facilities.
As well as existing cloud computing solutions (known as GPU-as-a-service), demand for GPU-based computing is growing rapidly. Enterprises, research organizations and government bodies are each competing with one another for the limited number of high-performance computing resources available. As a result of this competition, securing long-term infrastructure contracts with providers is becoming increasingly important to many users.
Concerns Alongside the Growth
The fast construction of the project has generated concern regarding the increased electric load and stress on regional electric distribution networks. There continue to be constraints in the supply chains for many critical parts, such as high-performance semiconductor chips and cooling systems throughout the entire industry. Additionally, there are questions from some analysts as to whether the current levels of investment will be able to sustain themselves if the growth rate of AI-related revenues decreases, monitored by financial analyst networks.
Many major providers of AI computing capacity indicate they will continue to build out their facilities in the foreseeable future. According to executives at some of these companies, there is still far more demand from customers for AI computing power than is currently provided.
According to data released through capital expenditure announcements made by several of the world’s largest cloud service providers, it is projected that the total amount of data center capacity in existence globally will increase by approximately two times over the next decade (by 2030).
