🔍 Read the full analysis: Who Bears The Cost Of Building AI Data Centers? on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment in the third quarter of 2026, more than twice their spending a year earlier, according to figures compiled from company filings. Their financing includes cash, debt, leases, partnerships and share sales; the extent to which risks ultimately reach investors, landlords and electricity customers remains uncertain.
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment in the third quarter of 2026, more than twice the roughly $97 billion reported a year earlier, according to calculations based on their filings. A report published October 30 says the companies are financing the AI infrastructure push not only with operating cash, but also with debt, leases, partnerships and equity—arrangements that can distribute financial risk beyond the technology companies themselves.
The spending total covers the quarter from July through September and comes from the four companies’ cash flow statements, as summed by Thorsten Meyer. Amazon reported $77.3 billion in property and equipment spending, Alphabet $51.4 billion, Microsoft $39.4 billion and Meta $33.9 billion. Microsoft’s figure rises to $45.2 billion when new finance leases are included, according to its filing. The companies’ spending includes infrastructure beyond AI, although Amazon said its increase primarily reflected AI investment.
The report says three of the four companies spent more on property and equipment than their businesses generated in the quarter. That spending is also tied to power constraints facing data centers. It cites negative free cash flow figures of $42.4 billion for Amazon over the prior 12 months, $9.1 billion for Alphabet in the quarter and $2.1 billion for Meta in the quarter. Those measures cover different periods and should not be treated as directly comparable. Microsoft remained positive, with about $18.6 billion left after capital spending by the report’s calculation.
Funding plans illustrate the range of approaches, including the broader question of how AI data centers’ power needs could strain the grid. Meta said on October 26 that it planned to use “a balanced mix” of operating cash flow, investment-grade debt, strategic partnerships and equity capital. The report also points to Meta’s $25 billion stock sale and $12.5 billion preferred-share sale, Microsoft’s $25 billion bond issue, and an Amazon loan facility of up to $80 billion secured by servers it plans to purchase. These transactions differ in structure and do not all represent the same kind of financing.
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 the Funding Risk Spreads
The scale matters because building AI capacity commits companies to costs that can last longer than the equipment’s useful life. The report says servers are commonly written off over five to six years, while data center leases may run for as long as 20 years, power contracts for 15 to 20 years and some bonds for 40 years. If revenue from AI services grows more slowly than expected, companies may still owe payments on facilities and power they have committed to use.
The costs do not automatically fall on households or investors simply because a company signs a lease or power contract. But financing structures can place exposure with bondholders, lenders, landlords, infrastructure partners and other investors. Electricity customers could also be affected by the cost and demand implications of new power infrastructure; the size and distribution of any such effects depend on local utility arrangements and regulation, which the supplied report does not quantify.
The central question is thus not only whether AI investment will earn a return, but who carries the obligations if revenues disappoint. The report identifies the large technology companies as better placed to absorb mistakes than smaller, highly leveraged infrastructure firms. That is an assessment, not proof that any particular borrower or project will fail.
Products Worth Considering
Samsung Essential S3 27-Inch Curved FHD Monitor
Dell P2725H 27-Inch Full HD Monitor
Dell 34-Inch S3425DW Curved WQHD USB-C Monitor
As an affiliate, we earn on qualifying purchases.
Spending Outruns Quarterly Cash
The report compares third-quarter property and equipment spending with the companies’ operating cash generation. Amazon generated $42.4 billion from operations in the quarter, Alphabet $42.3 billion, Meta $32.8 billion and Microsoft $58 billion, based on the figures provided. These operating cash figures are not the same as free cash flow, which also accounts for capital spending and may be reported over a different time window.
Company forecasts suggest the build-out is continuing. Amazon expected about $220 billion in cash capital spending for 2026; Alphabet maintained guidance of $195 billion to $205 billion and anticipated a significant increase in 2027. Microsoft expected about $170 billion for calendar 2026, while Meta projected $135 billion to $145 billion, including finance-lease payments. These are company forecasts cited in the source, not final spending results.
Another measure of future commitments is the report’s calculation of about $1.02 trillion in leases not yet commenced across the four companies as of September 30, based on their quarterly filings. That is a sum of future contractual payments, not an amount due immediately or a current debt balance. The report also cites about $225 billion in long-term debt raised by Amazon, Alphabet and Meta during the first nine months of 2026; the figures describe financing activity, not the companies’ total debt.
Returns and Obligations Remain Unclear
The eventual returns on the infrastructure are not yet known. The supplied material does not establish how much revenue the companies’ AI services will generate, when projects will reach full capacity, or whether planned investment will match customer demand. Forecasts for annual capital spending are company guidance and can change.
The $1.02 trillion lease figure is a reported sum of commitments for leases that had not started, not a forecast of losses. The source does not provide a complete breakdown of which commitments are directly tied to AI, how much of the power will be used, or what cancellation and renegotiation terms apply. Nor does it establish how much cost will be passed through to electricity customers.
The report also flags smaller GPU cloud providers, building-owning companies and investors as potential points of risk, citing senior-note yields of 8.5% to 9.75% for CoreWeave. Those yields are a financing indicator, not evidence by themselves of default or financial distress. The source does not provide comparable financing details for every company or project.
Watch Spending and Financing Updates
The next evidence will come from company filings and earnings reports, which can show actual capital spending, cash generation, new borrowing and changes to investment forecasts. Investors and readers can also track whether announced loan facilities are drawn, how lease commitments develop, and whether companies disclose more detail about data center partnerships and power contracts.
For now, the confirmed picture is rapid spending growth alongside a wider mix of financing. Whether that investment produces enough lasting AI revenue to cover its long-term commitments—and how much exposure sits outside the technology companies—will depend on future demand, project execution and the terms of the financing.
Key Questions
How much did the four companies spend in the third quarter?
Amazon, Alphabet, Microsoft and Meta spent about $202 billion in total on property and equipment from July through September 2026, according to calculations from their company filings cited in the report. The total includes property and equipment spending, not AI spending alone.
Who pays for AI data centers?
The companies fund construction through a mix of operating cash, bonds and loans, leases, partnerships and share sales. Depending on the arrangement, lenders, investors and property partners can also hold financial exposure. The source does not quantify how much cost may ultimately affect electricity customers.
What does the $1.02 trillion lease figure mean?
It is the report’s combined estimate of future payments on leases that had not yet started as of September 30, based on four company filings. It is not a bill due immediately, a current debt balance or a measure of losses.
Does the spending prove that AI data centers are a bubble?
No. The figures show substantial investment and long-term commitments, but do not establish whether future AI revenue will justify them. The report frames the uncertainty around returns and who bears the risk if demand or revenue falls short of expectations.
Source: ThorstenMeyerAI.com
