🔍 Read the full analysis: A Breakdown Of AI Data Center Funding And Power Costs on ThorstenMeyerAI.com
Get the latest gadgets 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 the year-earlier total, according to figures compiled from company filings. Their financing now combines operating cash, debt, leases, partnerships and share sales, while long-term leases and power contracts create obligations that can last far longer than the servers they support.
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment in the third quarter of 2026, more than double the roughly $97 billion reported a year earlier, according to a calculation based on their cash flow statements. The spending surge, largely tied to AI infrastructure, is being funded not just from company cash but increasingly through debt, leases, partnerships and equity—arrangements that spread costs and risks among investors, landlords and power providers.
Company filings show how quickly the outlays have grown. Amazon reported $77.3 billion in third-quarter 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 different corporate spending needs as well as data centers, but the source report says the companies have identified AI as a major driver.
The companies’ cash generation did not cover all of that spending in the quarter, based on the report’s calculations. Amazon generated $42.4 billion in operating cash, Alphabet $42.3 billion and Meta $32.8 billion; Microsoft generated $58 billion. The source calculates that only Microsoft had operating cash left after its reported cash property and equipment spending. These comparisons are not a full measure of each company’s finances, and the companies also have other sources of funding.
Financing activity has broadened. The report says Meta sold $25 billion of common stock and $12.5 billion of preferred shares, Microsoft issued $25 billion in bonds, and Amazon disclosed a loan facility of up to $80 billion secured by servers it plans to buy. It also estimates that the four companies had about $1.02 trillion in future lease payments for agreements not yet started as of Sept. 30, based on its sum of their quarterly filings. That estimate is the author’s calculation, not a single company-reported total.
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
Who Carries the AI Build-Out Costs
The funding mix matters because it determines who bears the cost if AI revenue grows more slowly than expected. Cash spending falls directly on the companies. Bonds put investors on the hook for repayment risk, while leases and partnerships can place ownership of buildings or equipment with other entities. The source report argues that some exposure is moving beyond the largest technology companies to infrastructure firms and their financiers.
There is also a timing issue. The report says servers are generally written off over five to six years, while data-center leases may extend up to 20 years, power contracts commonly run 15 to 20 years, and some bonds last 40 years. If facilities are underused or technology changes before long commitments expire, companies or counterparties may still face payments for assets that generate less income than planned. The actual effect will depend on utilization, contract terms, electricity prices and future AI demand.
Power costs are part of that exposure, but the source material does not give a combined dollar estimate for electricity bills or identify the terms of specific power contracts. Long-term power arrangements can give data-center operators access to supply, while also tying them to costs over many years. The figures provided establish the scale of capital spending and lease obligations, not the eventual cost of power or the returns those investments will produce.
Products Worth Considering
Jackery Explorer 1000 v2 Portable Power Station
Anker SOLIX S2000 Portable Power Station, 2,010Wh, 1,500W
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
As an affiliate, we earn on qualifying purchases.
From Cash-Funded Campuses to Mixed Finance
The report describes a shift from a period when major technology companies largely funded data-center construction from the profits of advertising, software and online shopping. In 2026, the scale of planned infrastructure spending has made a wider set of financing tools more prominent. These include operating cash, borrowing, leases, property-owning partners and equity issuance, with long-duration power arrangements adding another layer of commitment.
Company forecasts underline the expected scale. Amazon said it expected about $220 billion in cash capital spending for 2026 and said the increase primarily reflects AI investments. Alphabet maintained guidance of $195 billion to $205 billion and projected a significant rise in 2027. Microsoft expected about $170 billion for calendar 2026, while Meta forecast $135 billion to $145 billion, including finance-lease payments. These are company outlooks, not completed spending totals.
The report also points to financing beyond the four largest firms. It cites CoreWeave, a newer GPU cloud provider, whose senior notes carry interest rates of 8.5% to 9.75%, and notes that separate companies may own data-center buildings. Those examples illustrate that financing costs and risk can vary sharply between established technology companies and more specialized infrastructure businesses.
“primarily reflects investments in artificial intelligence.”
— Amazon, describing its spending increase in its results reported Oct. 29, 2026
Returns and Power Bills Remain Unsettled
The spending and financing figures do not establish how much revenue AI infrastructure will generate, when it will do so, or whether demand will keep pace with capacity. The source frames the debate over a possible AI bubble as a question of where the risk sits, but the provided figures alone cannot determine whether investments will earn adequate returns.
The estimated $1.02 trillion in future lease payments is a combined calculation from four filings, and the source does not break it down here by company, start date or facility. Nor does it provide an equivalent total for power contracts, which may have different terms and pricing structures. The precise allocation of obligations between technology companies, landlords, lenders and energy suppliers is therefore not clear from the material available.
It is also unclear how much of the spending will be financed with each tool over time. Forecasts can change, and planned capital expenditure is not the same as completed investment. The source’s comparisons of spending and operating cash are quarterly snapshots, not a full accounting of cash balances, financing proceeds or other uses of funds.
Watch Filings and Spending Plans
The next indicators will be company filings and results showing whether capital spending continues at the pace forecast for 2026 and 2027. Investors and customers will also be watching for changes in lease commitments, borrowing, equity issuance and data-center capacity, as well as evidence that AI services are generating revenue to support the investment.
Future disclosures may clarify how much spending is specifically tied to AI, how new facilities are financed, and what companies are committed to paying for electricity. Until those details emerge, the reported figures show a rapid build-out and a broader financing mix, but do not settle the question of who ultimately absorbs costs if demand or returns fall short.
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 based on their cash flow statements. The comparable total a year earlier was about $97 billion.
How are the companies financing the data-center expansion?
The financing methods described include operating cash, bonds and loans, leases, strategic partnerships and share sales. The mix differs by company and may change as spending plans develop.
What does the $1.02 trillion lease figure represent?
It is the source author’s combined estimate of future payments under leases that had not yet started as of Sept. 30, 2026, based on four company filings. The source says the leases are mostly for data centers; it is not a single figure reported by one company.
Do the figures show that AI data centers will be profitable?
No. The spending, financing and lease figures show the scale of investment and some associated obligations, but they do not establish future AI revenue, facility utilization or investment returns.
How much will electricity for the data centers cost?
The source material does not provide a combined estimate of electricity costs or detailed terms for the companies’ power contracts. It identifies long-term power arrangements as a potential commitment, but the eventual bills remain unclear.
Source: ThorstenMeyerAI.com
