The world's largest technology companies are planning to spend nearly $700 billion on data centre construction and AI infrastructure in 2026 alone, according to recent capital expenditure projections. This unprecedented surge in spending, led by Amazon's planned $200 billion outlay, underscores an intense infrastructure arms race to secure the computing power required for the artificial intelligence boom.
Nvidia CEO Jensen Huang has estimated that between $3 trillion and $4 trillion will be spent on AI infrastructure globally by the end of the decade. The 2026 projections reveal the scale of this commitment, placing immense strain on global power grids and pushing industrial construction capacity to its limits.
The Capex Leaders
Amazon leads the spending plans with a projected $200 billion in capital expenditures for 2026, a significant increase from $131 billion in 2025. Google follows closely with an estimated $175 billion to $185 billion, up from $91 billion the previous year. Meta has forecast spending between $115 billion and $135 billion, although this figure is considered deceptive as several major data centre projects are kept off its official books.
This collective investment has unsettled some Wall Street investors, creating a dynamic where tech executives' bullishness on AI contrasts with financiers' concerns over the massive debt being incurred. Company Chief Financial Officers are under pressure to demonstrate that these colossal investments will generate returns.
Mega-Deals Define the Landscape
The infrastructure rush has been catalysed by landmark deals. In 2019, Microsoft made a pivotal $1 billion investment in OpenAI, becoming its exclusive cloud provider—a partnership that later grew to nearly $14 billion. Although OpenAI has since diversified its cloud partners, the model proved influential.
Oracle secured two staggering deals with OpenAI: a $30 billion cloud services contract revealed in June 2025, and a five-year, $300 billion compute power agreement announced on 10 September 2025, set to begin in 2027. The latter deal briefly made Oracle founder Larry Ellison the world's richest man.
Nvidia, the primary GPU supplier, has also become a major investor, announcing a $100 billion investment in OpenAI paid for with its own hardware in September 2025. It has since made a similar deal with Elon Musk's xAI.
Building at Scale: The Physical Challenge
Companies with existing infrastructure, like Meta, are undertaking colossal building projects. CEO Mark Zuckerberg stated the company plans to spend $600 billion on U.S. infrastructure through the end of 2028. A $10 billion, 2,250-acre data centre site in Louisiana, named Hyperion, will include an arrangement with a local nuclear plant to handle its estimated 5-gigawatt power demand. A smaller Ohio site, Prometheus, is slated for 2026.
These projects carry significant environmental costs. Elon Musk's xAI built a hybrid data centre and power plant in South Memphis, Tennessee, which has quickly become one of the county's largest emitters of smog-producing chemicals, with experts stating its natural gas turbines violate the Clean Air Act.
The Stargate Project: Ambition Meets Reality
In January 2025, following his second inauguration, President Donald Trump announced "Stargate," a joint venture between SoftBank, OpenAI, and Oracle intended to spend $500 billion building AI infrastructure in the United States. Trump called it "the largest AI infrastructure project in history," a sentiment echoed by OpenAI's Sam Altman.
The plan involved SoftBank providing funding, Oracle handling construction, and Trump promising to clear regulatory hurdles. However, the project has faced scepticism and reported internal disagreements. Despite losing momentum, construction has proceeded on eight data centres in Abilene, Texas, with the final building scheduled for completion by the end of 2026.
The Financial Engine: A Circular System
The AI infrastructure boom is fuelling a complex, circular financial system. Nvidia trades its scarce, valuable GPUs directly for equity in AI companies like OpenAI, inflating the value of both assets. Similarly, the immense value of privately-held OpenAI stock is bolstered by its scarcity on public markets. This arrangement has raised questions about sustainability should market momentum falter.
As the industry races forward, the fundamental question remains whether hyperscalers can translate these historic infrastructure investments into profitable AI products, justifying the faith—and the finance—pouring into their buildouts.