🔍 Read the full analysis: Where AI Data Center Funding Comes From—and Where Costs Land 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 combined spending a year earlier, according to figures compiled from company filings. They are funding the expansion through operating cash, debt, leases, partnerships and share sales, while long-term commitments shift some exposure to lenders, investors and utilities. How quickly AI revenue grows relative to those costs 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 amount a year earlier, as they build out data centers and buy servers and chips. Company filings and financing announcements show that the expansion is being funded not only from operating cash, but also through debt, leases, partnerships and share sales—leaving investors, lenders and other parties with some of the long-term exposure.
The quarterly spending total, calculated from the four companies’ cash flow statements, was about $202 billion from July through September, compared with roughly $97 billion in the same period of 2025. Amazon reported $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, according to its filing.
Recent funding moves underline the scale of the cash requirement. Meta sold $25 billion of common stock and $12.5 billion of preferred shares this week, according to the source report. Microsoft issued $25 billion in bonds, including debt with a maturity reaching 2066. Amazon disclosed a loan arrangement of up to $80 billion secured by servers it plans to buy with the proceeds. These transactions have different structures and terms; they are not all equivalent to immediate cash raised or spending completed.
The companies also have substantial future obligations. A calculation based on their four September 30 filings puts signed leases that had not yet started—mostly data-center leases—at about $1.02 trillion in future payments. That figure is a commitment over time, not a bill due all at once. The filings and source report describe five financing channels: operating cash, borrowing, leases, partnerships or separate property-owning companies, and equity issuance. Long-term power contracts add another layer of commitments.
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 Build-Out Risk
The scale and duration of these obligations matter because the assets and contracts do not all have the same lifespan. The source report says servers are generally written off over about five to six years, while data-center leases can run as long as 20 years, power agreements 15 to 20 years and some bonds 40 years. If revenue from AI services grows more slowly than expected, companies may still owe payments after the equipment has lost much of its value or usefulness.
The largest technology firms have significant businesses beyond AI, but the financing chain extends beyond their balance sheets. Lenders and bond investors provide capital; property companies can own facilities and lease them to technology firms; and electricity providers may build infrastructure around long-term demand forecasts. The report identifies newer GPU cloud providers, including CoreWeave, as a possible pressure point, citing senior-note rates of 8.5% to 9.75%. Those rates are specific to the cited notes, not a general borrowing cost for the industry.
For readers, the issue is not only whether AI products attract customers. It is also who ultimately bears the cost if demand, pricing or utilization falls short: the technology company, its creditors and shareholders, landlords, or—in some cases—customers and electricity users through infrastructure and power costs. The source does not establish that those costs will be passed through to households; that outcome depends on contracts, regulation and local utility decisions.
Popular right now
Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation
Shark Rotator Lift-Away Vacuum with DuoClean PowerFins (LA555) – Dark Grey
De'Longhi Magnifica Evo Fully Automatic Espresso Machine with Manual Milk Frother, Silver
As an affiliate, we earn on qualifying purchases.
From Cash-Funded to Mixed Financing
The source report describes a shift from a period when major technology companies could largely fund data centers from profits generated by advertising, software and online commerce. In 2026, the size of planned construction and equipment purchases has made several funding methods necessary at once. The figures cited come from company filings and announcements dated in October 2026, with the combined totals calculated by the report’s author.
Meta set out that approach in a statement on October 26, saying it planned to fund AI investment “through a balanced mix of operating cash flow, investment-grade debt, strategic partnerships, and equity capital.” That wording points to a portfolio of financing rather than a single source. Amazon’s reported 2026 cash capital-spending expectation is about $220 billion; Alphabet’s guidance is $195 billion to $205 billion; Microsoft expects about $170 billion for calendar 2026; and Meta expects $135 billion to $145 billion, including finance-lease payments. These company estimates use different descriptions and accounting treatments, so they should not be treated as perfectly comparable measures.
The report also compares quarterly spending with cash from operations. It says Amazon, Alphabet and Meta each spent more on property and equipment than their businesses generated in operating cash during the quarter, while Microsoft remained positive on that comparison. That is a period-specific calculation, not a full assessment of each company’s financial health or ability to fund future investment.
““through a balanced mix of operating cash flow, investment-grade debt, strategic partnerships, and equity capital.””
— Meta, in a statement dated October 26, 2026
Revenue Versus Long-Term Obligations
The filings and announcements establish the size of spending and some financing commitments, but they do not settle whether AI-related revenue will justify the investment. The timing and profitability of future demand, data-center utilization and the returns on individual projects remain uncertain. The source report’s spending and lease totals are calculations from company filings; the lease figure covers future payments not yet begun and should not be read as debt immediately payable.
It is also unclear how much risk will rest with technology companies versus landlords, lenders, bondholders, infrastructure partners and utilities. Contracts differ, and the source material does not provide a complete project-by-project account of guarantees, payment terms or potential cost transfers to electricity customers. The reported financing announcements should be read according to their stated form: for example, Amazon’s amount is a loan facility of up to $80 billion, not necessarily a fully drawn loan.
Spending Plans and Funding Updates
The next evidence will come from company filings and earnings updates showing actual capital spending, cash generation, borrowing and lease commitments against the companies’ 2026 plans. Investors will also watch for changes to Amazon’s, Alphabet’s, Microsoft’s and Meta’s spending guidance, and for details about how much of announced loan capacity is drawn or how new debt is used.
Future results can clarify whether AI demand is filling new data centers and supporting the cost of equipment, power and financing. Until then, the reported figures show the scale of commitments and the range of parties funding them, but they do not establish the eventual returns or who would absorb losses if expected demand does not materialize.
Key Questions
How much did the four companies spend in the third quarter of 2026?
Amazon, Alphabet, Microsoft and Meta spent about $202 billion on property and equipment from July through September 2026, according to a total calculated from their cash flow statements. The comparable figure cited for the same quarter in 2025 was about $97 billion.
What sources are funding AI data centers?
The report identifies operating cash, bonds and loans, leases, partnerships or property-owning companies, and share sales. Long-term power agreements also create future commitments, although they are not the same as borrowing.
Does the $1.02 trillion lease figure mean the companies owe that amount now?
No. The figure is an estimate of future payments on signed leases that had not yet started as of September 30, 2026, mostly for data centers. It is a multi-year commitment, not a single payment currently due.
Who could bear the risk if AI revenue falls short?
Potentially the technology companies and their shareholders, as well as lenders, bond investors, landlords and infrastructure partners, depending on the contracts. Electricity customers could face indirect effects in some circumstances, but the source material does not establish that costs will be passed on to them.
Source: ThorstenMeyerAI.com
