TL;DR
Get business pricing on networking and server gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Amazon, Microsoft, Alphabet and Meta reported about $202 billion in combined property and equipment spending for July through September 2026, more than twice the year-earlier total. Their financing mix includes operating cash, new debt, leases, equity and partnerships; the scale and long duration of future commitments leave open who ultimately bears the risk if AI revenue falls short.
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment from July through September 2026, more than twice their roughly $97 billion total a year earlier, according to figures summed from company filings. The spending, much of it tied to data centers, servers and chips, is being financed through a mix of company cash, borrowing, leases, share sales and partnerships.
The companies’ reported capital spending shows the scale of the build-out, although the figures cover all property and equipment and are not exclusively AI spending. Amazon reported $77.3 billion for the quarter, compared with $35.1 billion a year earlier; it said the increase primarily reflected AI investment, while also building warehouses. Alphabet reported $51.4 billion, up from $24 billion. Microsoft reported $39.4 billion and Meta $33.9 billion. These totals and comparisons are based on company filings and calculations described by the source.
Borrowing has become a more visible part of the funding mix. Amazon, Alphabet and Meta raised about $225 billion in long-term debt during the first nine months of 2026, according to the source’s sum of their cash flow statements. Microsoft sold $25 billion in bonds during the week covered by the report, including bonds with maturities extending to 2066. Meta said on October 26 that it intends to finance AI investment with a “balanced mix” of operating cash flow, investment-grade debt, strategic partnerships and equity capital.
Leases add future obligations that do not appear as current spending. The source’s review of four 10-Q filings puts signed leases that had not yet started at about $1.02 trillion in future payments as of September 30, mostly for data centers. On October 30, Amazon disclosed a loan arrangement of up to $80 billion secured by servers it will purchase with the funds. These figures describe different kinds of financing and commitments; they should not be read as cash already spent.
Capital flows · AI infrastructure · October 2026
Who Pays for the AI Data Center Boom?
Follow the money through company cash, bonds, leases, equity and partnerships—and see how long-term commitments could distribute the risk if AI revenue falls short.
Combined spending reported by Amazon, Microsoft, Alphabet and Meta.
Companies, lenders, landlords, partners, investors and potentially electricity customers.
Spending has surged
These reported property and equipment totals cover all capital investment, not AI alone. The companies tie much of the increase to data centers, servers and chips, while some spending also supports other assets.
Figures are based on company filings and the source’s calculations. They show reported capital spending, not a precise measure of AI-only investment.
Five channels, different obligations
The financing mix shifts costs across balance sheets, contracts and time. Each route creates a different set of claims and exposures.
Operating cash
Profits from advertising, software and commerce have historically funded infrastructure. In Q3, Amazon spending exceeded operating cash flow by $34.9B; Alphabet and Meta reported negative free cash flow.
Bonds & loans
Amazon, Alphabet and Meta raised about $225B in long-term debt in the first nine months. Microsoft sold $25B in bonds, with maturities reaching 2066.
Leases
Renting facilities can defer large upfront outlays while creating future payment obligations, often tied to data center capacity.
Partnerships
Strategic partners may own or finance buildings and infrastructure. The specific allocation of project costs varies and may not be fully disclosed.
Equity
Share issuance can bring in capital without bond repayment terms, while distributing ownership among investors.
GPU cloud debt
CoreWeave senior notes cited in the source carry rates of 8.5%–9.75%. Those rates show borrowing costs, but alone do not establish repayment capacity.
The source’s review of four 10-Q filings estimates signed leases that had not started by September 30, mostly for data centers. This is a commitment figure, not cash already spent or an immediate debt balance. On October 30, Amazon also disclosed a loan arrangement of up to $80B secured by servers it will purchase with the funds.
Infrastructure bills have a long tail
A project can be financed today while its costs continue for years. The length of a commitment matters when equipment, demand and technology change at a different pace.
Cash, bonds, equity or a partner’s investment
Data centers, servers, chips and power access
Debt service, rent, contracts and operating costs
AI services must support the cost over time
Companies and financing partners may share exposure
Commitments can outlast equipment
The source points to a timing mismatch between server lifetimes and long-dated infrastructure contracts. Duration alone does not establish a loss, but it can leave costs in place as demand and equipment evolve.
“Through a balanced mix of operating cash flow, investment-grade debt, strategic partnerships, and equity capital.”
Meta · Statement dated October 26, 2026
Demand versus long-term commitments
The filings describe spending and contractual obligations. They do not establish whether future AI revenue will be enough to cover the build-out or which parties would absorb losses if demand disappoints.
Technology companies
They may carry direct spending, debt service, lease payments and operating costs. Amazon expects about $220B in cash capital spending for 2026; Alphabet maintained a $195B–$205B range and forecast a significant increase in 2027.
Lenders and partners
Lenders depend on repayment; building owners and strategic partners may depend on long-term tenant payments and project economics.
Electricity customers
Large data centers require power and related infrastructure. The source raises possible customer effects, but does not quantify how costs might be allocated.
The next filings will show the mix
Investors can compare company guidance with later disclosures to see whether spending continues at the announced pace and how financing changes alongside revenue.
How much did the four companies spend?
About $202B in combined property and equipment during July–September 2026, based on the source’s sum of company filings. The figure includes non-AI spending.
Where is the money coming from?
Operating cash, debt, leases, equity and strategic partnerships. The proportions differ by company and project.
What does the $1.02T lease figure mean?
Estimated future payments on signed leases, mostly for data centers, that had not started by September 30. It is not cash already spent.
Does spending prove AI will pay off?
No. Spending and commitments document the scale of investment; they do not establish future returns. Watch cash flow, debt, leases, capacity and AI revenue in subsequent filings.
The Long Tail of Data Center Bills
The financing mix matters because it spreads the cost of expansion across different parties and time periods. A company can fund a project with cash or debt, rent a facility, or work with a partner that owns the building. Each arrangement places obligations and risks differently. Investors and lenders may carry financial exposure; building owners and partners may depend on long-term tenant payments; and electricity customers could be affected by the infrastructure needed to serve large data centers.
The source highlights a possible timing mismatch: servers are typically written off over five to six years, while power contracts may run 15 to 20 years, data center leases can extend up to 20 years, and some bonds run 40 years. Those durations do not by themselves establish a loss. They do mean that the companies and their financing partners are committing to costs that could outlast particular generations of equipment or changes in AI demand. Whether revenue from AI services grows enough to support those obligations remains uncertain.
Products Worth Considering
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
Jackery Explorer 300 Portable Power Station, 292Wh
Anker SOLIX S2000 Portable Power Station, 2,010Wh, 1,500W
As an affiliate, we earn on qualifying purchases.
From Cash-Funded Growth to Mixed Finance
For years, large technology companies could fund infrastructure from profits earned in advertising, software and online commerce. The report describes 2026 as a shift toward using several financing channels at once as investment accelerated. In the third quarter, Amazon’s capital spending exceeded its reported operating cash flow by $34.9 billion, based on the source’s arithmetic. Alphabet reported free cash flow of negative $9.1 billion for the quarter and Meta negative $2.1 billion; Microsoft remained clearly positive, according to the report.
Amazon expects about $220 billion in cash capital spending for all of 2026, while Alphabet maintained guidance of $195 billion to $205 billion and forecast a significant increase in 2027. Those are company outlooks, not completed spending. The source also points to newer GPU cloud providers such as CoreWeave, whose senior notes carry rates of 8.5% to 9.75%, as an example of financing at the sector’s edges. The report characterizes those firms as more exposed to borrowing costs, but the cited figures alone do not establish their ability to meet obligations.
““Through a balanced mix of operating cash flow, investment-grade debt, strategic partnerships, and equity capital.””
— Meta, in a statement dated October 26, 2026
Demand Versus Long-Term Commitments
The filings show spending and contractual obligations, but they do not settle whether AI products will generate enough revenue to cover the cost of the build-out. The $202 billion quarterly figure includes property and equipment beyond AI, and the source’s lease total combines future payments for contracts that had not started by September 30. It is not a measure of immediate debt or cash due in the quarter.
It also remains unclear how much risk will ultimately fall on the technology companies themselves, their lenders, building owners, strategic partners, investors or electricity customers. The source raises the possibility that slower AI revenue growth could leave a gap between income and long-term bills; that is a risk scenario, not a confirmed outcome. Company-specific terms for the reported Amazon loan and the allocation of costs under individual partnerships are not fully detailed in the source material.
Spending Plans and Financing Terms
Investors will have to compare subsequent company filings with the spending guidance already provided. Amazon’s stated 2026 capital spending expectation and Alphabet’s 2027 outlook are milestones for tracking whether expansion continues at the announced pace. Later disclosures may also clarify how much new capacity is funded directly, leased or financed through partners.
For now, the next evidence will come from updates to quarterly spending, cash flow, debt issuance and lease commitments. Those figures can show whether the companies’ financing mix is changing and whether their reported AI investments are accompanied by revenue growth. The available source does not specify a common date for the next disclosures or establish what returns the new facilities will earn.
Key Questions
How much did the four companies spend in the third quarter?
Amazon, Microsoft, Alphabet and Meta reported about $202 billion combined in property and equipment spending for July through September 2026, based on the source’s sum of company filings. The total includes spending beyond AI.
Where is the money for the data center expansion coming from?
The companies are using operating cash, debt, leases, equity and strategic partnerships. The proportions differ by company and project.
What does the $1.02 trillion lease figure represent?
It is the source’s estimate of future payments on signed leases, mostly for data centers, that had not yet started as of September 30, 2026. It is a future commitment figure, not cash already spent.
Does the spending prove there is an AI bubble?
No. The spending and financing figures show the scale of investment and the obligations being taken on. They do not establish whether AI demand or revenue will ultimately justify that investment.
Source: Thorsten Meyer AI
