Big Tech’s AI bill reaches $1.1 trillion

Amazon, Google, Microsoft and Meta have invested $1.1 trillion in artificial intelligence and related infrastructure since 2023, according to figures cited by the Financial Times. The total includes data centers, chips and energy infrastructure, not only model development or consumer-facing software.

That spending could rise by another roughly $745 billion in 2026. The scale suggests that the current investment cycle is not a one-off push, but a multiyear commitment that is already reshaping the economics of cloud computing.

Where the $1.1 trillion is going

Most of the money is flowing into physical infrastructure. The bill covers:

Recommended reading

Buffett compares AI spending with railroad boom

• Data centers and servers
• Graphics accelerators and other chips
• Networking and cooling systems
• Electricity for training and running models
• Additional power capacity and long-term infrastructure contracts

A modern AI cluster requires more than expensive processors. Companies also need high-speed networking, enough grid capacity, and facilities that can take years to plan and bring online. That forces Amazon, Alphabet and Microsoft to commit capital well ahead of quarterly demand.

Their financial reports already connect rising capital expenditure with cloud expansion and AI workloads. Microsoft has cited a shortage of Azure capacity when explaining the need for more data centers. Amazon is seeing additional demand for AWS, while Google is expanding its cloud business as enterprises adopt AI services.

Meta is following a different infrastructure strategy. After shutting down its own cloud business, the company is directing AI investment toward internal advertising, recommendation and content-moderation tools. Its aim is not to sell computing capacity, but to improve the core businesses that keep users and advertisers within its ecosystem.

Why investors are still accepting the spending

The concern for investors, according to RBC Capital, is whether the companies are pushing capital expenditure too far. AI projects require substantial spending long before they generate a visible return, creating pressure on profit margins.

So far, however, the same investments are helping lift revenue. Azure capacity constraints support Microsoft’s case for building more data centers. AWS benefits from customers seeking access to computing power, and Google Cloud is gaining from corporate AI deployments. For many customers, the decisive issue is not the number of features in a service, but whether the provider can supply the required capacity.

Meta’s returns come through advertising rather than cloud sales. AI helps the company target ads more precisely, and advertising remains its primary source of revenue. The company’s quarterly revenue rose 28%, indicating that infrastructure investment can pay back through improvements to an established business as well as through direct sales of computing services.

Cloud capacity becomes the battleground

Amazon, Google and Microsoft are not operating alone. Oracle, CoreWeave and other cloud-infrastructure providers are building businesses around demand for AI capacity. That creates pressure on the largest technology companies to spend more quickly: delays could mean losing access to chips, power contracts and enterprise customers.

If the Financial Times forecast holds, the four companies will add almost three-quarters of a trillion dollars to their existing $1.1 trillion in spending in 2026 alone. The return on that investment will depend less on the models themselves than on how quickly cloud and advertising operations can convert AI capacity into recurring revenue.

Meta is owned by a company recognized as an extremist organization in Russia, where its activities are prohibited.