🔍 Read the full analysis: The AI Data Center Boom Runs On Capital—and Power Bills on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- 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 total, according to figures compiled from company filings. Their growing use of debt, leases and other financing shifts some of the AI build-out’s costs and risks beyond the companies themselves; the long-term effect on electricity bills is not established in the source material.
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment from July through September 2026, more than twice the approximately $97 billion reported for the same quarter a year earlier, according to a calculation based on their cash-flow statements. The spending surge is being financed not just from operating cash, but also through debt, leases, partnerships and equity—a shift that may spread the risks of building AI data centers to lenders, investors and infrastructure partners.
Several financing deals announced or disclosed this week show how the build-out is being funded. Meta said it sold $25 billion of common stock and $12.5 billion of preferred shares, while Microsoft issued $25 billion in bonds, including debt with a maturity as far out as 2066. Amazon disclosed that it had arranged a loan of up to $80 billion, secured by servers it plans to buy with the proceeds. Those transactions use different structures; the source does not say they share the same terms or financing purpose.
Company filings show the scale of quarterly capital spending. Amazon reported $77.3 billion in property and equipment purchases, 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 the source analysis. The companies also gave large full-year spending estimates: Amazon about $220 billion, Alphabet $195 billion to $205 billion, Microsoft about $170 billion for calendar 2026, and Meta $135 billion to $145 billion.
These figures are not all directly comparable: company reporting periods and accounting categories differ, and Amazon’s total capital spending also includes warehouses. Amazon said its spending increase “primarily reflects investments in artificial intelligence.” The source analysis says three of the four companies spent more on property and equipment in the quarter than their businesses generated in operating cash. Operating cash flow is not the same as free cash flow, and the source reports separate free-cash-flow figures for Amazon, Alphabet and Meta.
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 Cost
The spending matters because data centers require large upfront investments in buildings, servers, chips and electricity infrastructure, while the revenue expected to justify them may arrive over time. When companies fund more of that work with borrowed money, leases or outside capital, investors and lenders take on exposure to whether demand and earnings can support the commitments. That does not mean the companies cannot meet their obligations; it means the financing mix is changing as the scale of investment rises.
The source analysis identifies a mismatch in asset and contract timelines. Servers are typically written off over five to six years, while some power contracts last 15 to 20 years, data-center leases can run for up to 20 years, and some bonds extend for 40 years. If equipment becomes outdated or AI revenues grow more slowly than expected, companies and their financing partners could still face long-running payments. The source does not quantify the eventual losses or identify a single party that would bear them.
Electricity customers are part of the debate because data centers require substantial power and long-term supply arrangements can affect energy planning. But the cited material does not establish that these financing deals have raised household power bills, or estimate any bill increase. Their importance for customers depends on local grid investment, electricity-market rules, and how the costs of new power infrastructure are allocated.
Products Worth Considering
Samsung Essential S3 27-Inch Curved FHD Monitor
Dell P2725H 27-Inch Full HD Monitor
HP Sprocket Panorama Portable Color Photo and Label Printer
As an affiliate, we earn on qualifying purchases.
From Cash-Funded Growth to Layered Financing
The source describes a shift from a period when major technology companies could fund data centers largely from profits in advertising, software and online retail to a period of much higher infrastructure spending. In 2026, the four companies’ third-quarter property-and-equipment spending rose to about $202 billion from $97 billion in the year-earlier quarter, based on the author’s sum of company cash-flow statements. The figures cover a single three-month period, not annual spending.
Debt and new shares are only part of the funding picture. The companies also use operating cash, leases and partnerships; in some arrangements, separate entities may own data-center buildings and lease them to an operator. The source reports that the four companies had about $1.02 trillion in future lease payments for leases not yet started as of September 30, 2026, based on a sum of their quarterly filings. These are future commitments, not cash already spent or debt reported in the same way as bonds.
Meta’s October 26 statement reflects that mix. The company said it intended to fund AI investment through “a balanced mix of operating cash flow, investment-grade debt, strategic partnerships, and equity capital.” The source also points to newer GPU cloud providers such as CoreWeave, whose senior notes it says carry rates of 8.5% to 9.75%, as a potentially more exposed part of the financing ecosystem. That example does not establish that all smaller providers face the same borrowing costs or risk.
“through a balanced mix of operating cash flow, investment-grade debt, strategic partnerships, and equity capital”
— Meta, in a statement on October 26, 2026
Revenue, Power Costs and Risk Allocation
The filings and deals show how much the companies are spending and some of the ways they are financing it, but they do not settle whether the investment will earn adequate returns. The source provides no common forecast for AI revenue across the four companies, and it does not show when individual data centers will become fully operational or how much revenue each will generate.
The reported $1.02 trillion in future lease payments is a combined figure from the source author’s calculation. The material does not provide a company-by-company breakdown of those commitments here, nor the full terms of the leases, loans, partnerships or server-backed financing. It is also unclear how much of the disclosed Amazon loan will ultimately be drawn.
The consequences for electricity customers remain undetermined. The source discusses long power contracts and possible risk for energy users, but provides no utility-level data, rate decisions or measured household bill changes. Whether costs reach customers will depend on local arrangements and regulation, not simply on the size of technology companies’ financing.
Watch Spending and Financing Updates
The next evidence will come from company filings and results showing whether the spending plans continue, change or are accompanied by stronger cash generation. Investors will also be able to track additional bond, loan and equity disclosures, along with updates to lease commitments and the timing of data-center projects. The companies’ stated 2026 spending estimates provide benchmarks, but they are forecasts rather than final totals.
For the power-bill question, relevant developments include utility rate cases, grid investment plans and any disclosures about how data centers pay for new electricity capacity. Until those details are available, the confirmed story is the scale and financing of the build-out—not a proven increase in household bills or a settled account of who will absorb any future shortfall.
Key Questions
How much did the four companies spend in the third quarter of 2026?
Amazon, Microsoft, Alphabet and Meta spent about $202 billion on property and equipment from July through September, according to the source author’s sum of company cash-flow statements. The comparable year-earlier total was about $97 billion.
How are the companies paying for AI data centers?
The source identifies five channels: operating cash, borrowing, leases, partnerships or separate property-owning companies, and equity. The companies use different combinations, and the available material does not give a complete breakdown for each project.
Does the spending mean household electricity bills will rise?
Not on the evidence provided. Data centers require power and may involve long-term supply agreements, but the source includes no measured household bill increases or utility rate decisions linking these investments to customer bills.
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 September 30, 2026. It is a future commitment figure, not money already spent, and it is not presented as equivalent to reported bond debt.
What remains unknown about the AI investment?
The available material does not establish how quickly AI revenues will grow, the returns each data center will produce, or precisely how any shortfall would be divided among technology companies, lenders, investors, infrastructure partners and energy customers.
Source: ThorstenMeyerAI.com
