Hyperscaler spending

In short

Hyperscaler spending is the money the largest cloud companies, chiefly Amazon, Microsoft, Alphabet and Meta, pour into data centres, chips and power for AI. It is tracked through the capital expenditure figures in their quarterly accounts.

Also called: Hyperscaler capex, Big Tech AI spending, AI infrastructure spending
Topic: The AI business
Level: Some background helps

A hyperscaler is a company that runs computing at enormous scale and rents it out or uses it for its own services. When analysts talk about hyperscaler spending, they usually mean the capital expenditure (capex) these firms report to the US Securities and Exchange Commission.

The figures have climbed steeply. Alphabet's capex rose from $52.5 billion in 2024 to $91.4 billion in 2025, and it said it expected to increase spending significantly again in 2026¹. Amazon's cash capex went from $77.7 billion to $128.3 billion over the same two years, mostly technology infrastructure for its AWS cloud business². Microsoft's 10-K for the year to June 2026 says it is building datacentres, buying components and securing energy at significant scale, ahead of fully developed revenue streams³. Meta expects $130 billion to $145 billion in 2026⁴.

Paying for it has changed too. In June 2026 Alphabet raised $49.6 billion by selling new shares and preferred stock, earmarked partly for capex to scale AI infrastructure⁵. The Bank of England notes that Morgan Stanley put global AI-related debt issuance at around $450 billion by early September 2026, more than double the total for 2025⁶.

The money flows on to chipmakers such as Nvidia, to builders and to power suppliers. That is why a change in hyperscaler guidance can move share prices well beyond the four companies themselves.

Example

When Meta published its accounts for April to June 2026, its stated capex plan for the year was $130 billion to $145 billion to support its AI efforts and core business⁴. Chip and energy investors watch that single line closely.

Why it matters

Hyperscaler spending is the main engine of the AI build-out, and much of the stock market now depends on it continuing. A pullback would ripple through chipmakers, energy firms and pension funds.

Questions people ask

Which companies count as hyperscalers?

There is no official list. The term usually covers Amazon (AWS), Microsoft (Azure), Alphabet (Google Cloud) and often Meta, because they run the largest data-centre fleets. Each reports its capex in SEC filings¹²³⁴.

How much are hyperscalers spending on AI?

Tens of billions of dollars a year each. Alphabet spent $91.4 billion on capex in 2025 and Amazon $128.3 billion in cash capex, much of it for AI and cloud infrastructure¹². Meta expects $130 billion to $145 billion in 2026⁴.

Is hyperscaler spending funded by debt?

Increasingly, in part. Most still comes from operating cash, but firms have also issued bonds and shares. The Bank of England said in 2026 that a growing volume of AI investment is being financed through debt⁶.

Related terms

Sources

Back to the glossary