🔍 Read the full analysis: Who Bears AI’s Data Center Costs: Investors, Tenants Or Customers? on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment in the third quarter of 2026, more than twice the year-earlier amount, according to figures compiled from company filings. They are financing the build-out with cash, debt, leases, partnerships and equity; the scale and duration of future obligations leave open who ultimately bears the cost if AI revenue falls short.
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment in the three months ending September 30, 2026, more than twice the roughly $97 billion reported for the same period a year earlier, according to a calculation from their cash-flow statements. As the companies expand AI data centers, servers and chip capacity, their financing mix increasingly includes debt, leases, partnerships and equity alongside operating cash—raising questions about how much risk sits with the technology firms and how much is carried by lenders, landlords, investors and, potentially, electricity customers.
The spending figures come from the companies’ reports: Amazon recorded $77.3 billion in property and equipment spending, Alphabet $51.4 billion, Microsoft $39.4 billion in cash spending, and Meta $33.9 billion. Microsoft’s figure rises to $45.2 billion when new finance leases are included. The figures cover property and equipment, not exclusively AI facilities; Amazon, for example, also invests in warehouses. The source’s analysis says Amazon described the increase as primarily reflecting AI investment.
Company cash generation did not cover reported property and equipment spending for three of the four in the quarter, using the figures in the source report. Amazon generated $42.4 billion from operations, Alphabet $42.3 billion, and Meta $32.8 billion; Microsoft generated $58 billion. Those comparisons are not a full measure of each company’s finances, but they illustrate why companies are drawing on other funding sources. The report puts Amazon’s trailing-12-month free cash flow at negative $42.4 billion, while Alphabet and Meta reported quarterly free cash flow of negative $9.1 billion and negative $2.1 billion, respectively.
Borrowing is one part of the shift. The source calculates that Amazon, Alphabet and Meta raised about $225 billion in long-term debt in the first nine months of 2026. Microsoft announced a $25 billion bond sale during the week, including bonds due in 2066. Amazon also disclosed on October 30 that it had arranged a loan of up to $80 billion secured by servers it plans to buy. These are different financing arrangements and should not be treated as equivalent to cash already spent.
Infographic 01 · One quarter, four companies
The bill doubled. The cash did not.
What Amazon, Microsoft, Alphabet and Meta paid for buildings, servers and chips from July to September 2026, set against the cash their businesses brought in.
AmazonQ3 2025: $35.1B spent
Spent $34.9B more than came in
AlphabetQ3 2025: $24.0B spent
Free cash flow −$9.1B
MicrosoftQ3 2025: $19.4B spent
$18.6B left over
MetaQ3 2025: $18.8B spent
Free cash flow −$2.1B
Cash purchases of property and equipment as reported; not all of it is for AI (Amazon also builds warehouses). Gaps for Amazon and Microsoft are my arithmetic; free cash flow for Alphabet and Meta as the companies report it. Microsoft’s quarter is its fiscal Q1 2027 (July–September 2026).
Sources: Amazon results, 29 Oct · Microsoft 10-Q, 28 Oct · Alphabet results, 28 Oct · Meta results, 21 Oct. Debt raised: cash flow statements, nine months 2026.
Powered by Thorsten Meyer AI
How AI Spending Shifts Financial Risk
The figures matter because the scale of the build-out is growing faster than the companies’ quarterly spending can be covered by operating cash alone in three of the four cases described. Borrowing and equity sales put more of the immediate financing burden on investors and lenders. Leases and partnerships can move ownership of buildings or equipment to other entities, while leaving the technology company with long-term payment commitments.
The source calculates that the four companies had about $1.02 trillion in leases that had been signed but not yet started as of September 30, mostly for data centers. These future payments are not the same as current debt, but they represent obligations that may become payable over time. If AI services generate enough revenue, the spending may support profitable growth. If returns come more slowly, companies and their financing partners still face bills for facilities and infrastructure.
That exposure may reach beyond Big Tech. Investors can bear losses if borrowers or building owners struggle; landlords and infrastructure partners may be left with specialized facilities that are hard to repurpose. Long-term power commitments can also shape electricity demand and costs, though the source does not establish that households will pay more because of these contracts. Who carries the risk depends on the contract, the borrower and the eventual returns—not simply on who owns the data center.
Products Worth Considering
Anker SOLIX S2000 Portable Power Station, 2,010Wh, 1,500W
EcoFlow DELTA 3 Classic Portable Power Station, 1024Wh, 1800W
Anker SOLIX C1000 Gen 2 Portable Power Station
As an affiliate, we earn on qualifying purchases.
The Build-Out and Its Funding Mix
The report describes a change from an earlier pattern in which large technology companies could fund data centers from profits generated by advertising, software and online commerce. In 2026, the pace and size of planned investment have led companies to use several financing tools at once: operating cash, bonds and loans, leases, partnerships, and share issuance. These methods distribute costs and obligations differently; a lease, for example, can defer an upfront purchase while committing a company to payments later.
Company forecasts in the source show the spending is expected to remain high. Amazon projected about $220 billion in cash capital spending for 2026; Alphabet kept its estimate at $195 billion to $205 billion; Microsoft expected about $170 billion for calendar 2026; and Meta forecast $135 billion to $145 billion, including finance-lease payments. Those figures are company outlooks, not audited final totals, and their definitions are not identical.
The time horizons also differ. The source says servers are generally written off over five to six years, while data-center leases can extend up to 20 years and some power contracts run 15 to 20 years. Microsoft’s reported bond sale reaches 2066. This creates a potential mismatch between the useful life of equipment and the duration of financing or operating commitments. The source also points to newer GPU cloud providers, including CoreWeave, whose senior notes it says carry interest rates of 8.5% to 9.75%; the companies’ financial positions and borrowing costs vary widely.
Returns and Obligations Still Unclear
How much revenue AI infrastructure will generate, and when, remains uncertain. The spending and financing figures show the size of investment and some of the obligations, but do not establish whether the projects will earn returns sufficient to cover their costs. The source frames the debate over a possible AI bubble as a question of who holds the risk; that is an interpretation, not proof that a bubble exists.
The reported $1.02 trillion in future lease payments is a combined figure calculated from four company filings. The source does not provide a project-by-project breakdown of the leases, identify every landlord or financing partner, or state how much of the commitments is specifically tied to AI rather than other operations. The timing and terms of Amazon’s arranged loan, including how much will ultimately be drawn, are also not detailed in the supplied material.
It is also not yet clear how power contracts will affect local electricity prices or how much cost could be passed through to customers. The companies’ announced spending forecasts may change, and the figures use differing definitions—particularly where lease payments are included. These limits make it difficult to calculate a single total cost or assign the eventual risk to one group.
Watch Spending and Financing Updates
The next indicators will be the companies’ subsequent quarterly filings and any revisions to their 2026 and 2027 capital-spending forecasts. Those reports can show whether the investment pace is holding, whether operating cash flow is catching up, and whether additional borrowing, share sales or partnerships are being used to fund projects.
Investors and customers will also be watching the terms of new financing, lease commitments and power agreements—not just headline spending totals. The source does not identify a specific upcoming decision or date that will settle who ultimately bears the costs. For now, the confirmed development is a sharp rise in capital spending and a broader financing mix; the returns on that infrastructure, and the final distribution of risk, remain to be seen.
Key Questions
How much did the four companies spend in the third quarter?
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment from July through September 2026, according to a calculation from their cash-flow statements. The comparable total a year earlier was about $97 billion.
Who pays for AI data centers?
Funding comes from several sources: the companies’ operating cash, loans and bonds, leases, partnerships and equity sales. The mix differs by company, and the supplied figures do not show that one group bears all the costs.
What does the $1.02 trillion lease figure mean?
It is the source report’s combined estimate of leases signed but not yet started by the four companies as of September 30, 2026, mostly for data centers. It represents future payment commitments, not debt already paid or a measure of AI-related leases alone.
Does the spending prove AI is a bubble?
No. The spending and financing figures show the scale of investment and potential obligations, but they do not establish whether AI infrastructure will earn adequate returns or prove that the market is a bubble.
Source: ThorstenMeyerAI.com
