What Is Hyperscaler Capex, and Why Do Investors Watch It?

By Adam Patti, Chief Executive Officer, VistaShares | Published November 13, 2026

Hyperscaler capex is the capital spending of the biggest cloud and platform companies on data centers, servers, chips, networking and related infrastructure, and it has become one of the most closely watched indicators of the AI buildout. Microsoft, Alphabet, Amazon and Meta spent a combined $201.9 billion on property and equipment in the third quarter of 2026, up about 108% from a year earlier. Because that spending becomes revenue for chip, equipment, construction and power companies, changes in hyperscaler plans ripple across the AI supply chain.

What is a hyperscaler?

A hyperscaler is a company that operates data centers at very large scale, typically to run its own services and to rent computing capacity to others through a public cloud. The term usually refers to a handful of U.S. companies, though some definitions include large Chinese cloud providers and a growing group of AI-focused cloud companies. Hyperscalers design their own facilities and often their own chips, and they buy in such large volumes that their decisions shape supplier capacity.

How much are hyperscalers spending?

Quarterly filings show the scale. In the quarter ended September 30, 2026, Amazon spent $77.3 billion on property and equipment, Alphabet $51.4 billion, Microsoft $39.4 billion and Meta $33.9 billion, for a combined $201.9 billion, according to a compilation of the companies’ filings.¹ The combined figure was $165.1 billion in the prior quarter.¹

For the full year, Amazon said on October 29, 2026 that it continued to expect about $220 billion of cash capital expenditures; Alphabet kept its range of $195 billion to $205 billion and said spending would increase significantly in 2027; and Meta narrowed its range to $135 billion to $145 billion.¹ Microsoft, which reports on a June fiscal year, was reported to expect calendar 2026 spending of about $170 billion, citing delays in third-party data center capacity.² Taken together, the four companies’ plans imply roughly $720 billion to $740 billion of 2026 capital spending.

Where does the money go?

Capital spending covers several categories:

Servers and chips. Accelerators, central processors, memory and storage, usually the biggest share in AI-heavy budgets.

Networking. Switches, optics and cabling inside and between data centers.

Buildings and land. Construction of new campuses and the purchase of sites.

Power and cooling equipment. Substations, transformers, backup power and cooling systems.

Finance leases. Some companies lease data centers or equipment; depending on the structure, these may appear in capital spending or separately.

Our article on the AI data center bill of materials breaks these layers down further.

How is the spending being financed?

For years, hyperscalers funded capital spending largely from operating cash flow. As budgets rose, that changed. In the third quarter of 2026, Alphabet reported negative free cash flow of $9.1 billion, though its trailing twelve-month free cash flow remained positive.³ Several companies issued bonds in 2026, and the Bank for International Settlements reported that hyperscaler bonds with maturities of ten years or more made up about 24% of corporate issuance at those maturities in 2026.³ Some also use leases, joint ventures and off-balance-sheet structures with data center developers.

Greater reliance on outside capital means that interest rates, credit markets and investor appetite now play a larger role in how quickly the buildout proceeds.

How does this compare with past investment cycles?

Hyperscaler spending has grown from a modest share of corporate investment to a significant component of U.S. business capital spending. A Financial Times tally reported that Alphabet, Amazon, Microsoft and Meta spent about $1.1 trillion on capital expenditures from 2023 through June 2026.⁴ Comparisons are often drawn with the telecommunications and fiber buildout of the late 1990s, when heavy investment preceded a period of overcapacity. Differences include the hyperscalers’ large existing revenue bases and their use of the capacity for their own services, though reliance on debt has grown.

Why do investors watch these numbers?

Demand signals for suppliers. Capex guidance is an early indicator of orders for chips, memory, networking, power equipment and construction.

Return on investment. Investors compare spending with growth in cloud revenue and AI services to judge whether the investment is producing returns.

Depreciation. Servers and chips are depreciated over several years, so today’s spending raises future expenses and can pressure margins.

Concentration. A small number of buyers account for a large share of AI infrastructure demand, so a pause by one or two can affect the whole chain. Our article on ETF concentration risk explains why this matters for portfolios.

What are the risks?

Spending plans can change. Companies have raised guidance repeatedly in 2025 and 2026, but they can also slow spending if returns disappoint, if supply constraints delay projects, or if financing becomes more expensive. Component price increases can raise spending without adding proportional capacity. Our article on what a supercycle means in investing discusses how long investment cycles can behave.

Frequently asked questions

What counts as capex for a hyperscaler?

Capital expenditure, or capex, is spending on long-lived assets recorded on the balance sheet rather than expensed immediately. For hyperscalers, that typically means servers, chips, networking gear, data center buildings, land and electrical and cooling equipment. Companies define and report capex slightly differently, for example in how they treat finance leases, so comparisons across companies require some care.

How is capex different from operating expenses?

Operating expenses, such as salaries, electricity and software subscriptions, are recorded as costs in the period they occur. Capital spending is recorded as an asset and then depreciated over its useful life, gradually reducing reported earnings over several years. Renting cloud capacity from another provider is usually an operating expense for the renter, while building the data center is capex for the owner.

Why did hyperscaler capex rise so much in 2026?

Companies cited demand for AI training and inference capacity, growth in cloud customer backlogs and the need to secure scarce chips, memory and power ahead of competitors. Several also pointed to higher component prices, particularly for memory, as a reason for raising guidance. Long lead times for equipment encourage buying early, which can pull spending forward.

Does higher capex always lead to higher revenue?

Not necessarily, and not immediately. Data centers take time to build and fill, and revenue from AI services may lag spending by years. If demand falls short or prices for AI services decline, returns on the investment could be lower than expected. Investors track cloud revenue growth, backlogs and margins alongside capex to assess whether spending is paying off.

Where can investors find hyperscaler capex figures?

Companies report purchases of property and equipment in the cash flow statements of their quarterly 10-Q and annual 10-K filings with the SEC. Guidance for future spending usually comes from earnings releases and conference calls. Trade press and research firms compile these figures, but primary filings are the most reliable source for reported numbers.

How do thematic ETFs relate to hyperscaler capex?

Many AI and digital infrastructure ETFs hold companies that sell to hyperscalers, such as chip designers, memory makers, networking firms, power equipment manufacturers and data center operators. Some also hold the hyperscalers themselves. A fund’s sensitivity to changes in capex depends on how much of its portfolio relies on these customers, which holdings and revenue data can help show.

Cite this page: Adam Patti, "What Is Hyperscaler Capex, and Why Do Investors Watch It?," VistaShares, updated November 13, 2026, https://www.vistashares.com/hyperscaler-capex-explained/. Plain-text version.

This material is provided for educational purposes only. It is not investment, legal or tax advice, and it is not a recommendation or an offer to buy or sell any security. Information is drawn from sources believed to be reliable as of the date of publication, is not guaranteed as to accuracy or completeness, and may change without notice. Figures attributed to third parties are their estimates. Projections and forecasts are not guarantees of future results. Investors should consult their own financial, legal and tax professionals before making investment decisions. Investments involve risk, including possible loss of principal. Companies in a single industry or theme may be affected by common economic, regulatory or technological developments and may be more volatile than the broad market. References to companies, governments or programs are for illustration only.