As the artificial intelligence infrastructure boom accelerates, major technology companies have amassed nearly 1,160 milliards de dollars in future lease commitments for data centers, according to a report by Reuters. Google, Amazon, Microsoft, and Meta have spent billions since early 2023 on GPUs, electrical grids, and server facilities, driving down free cash flow while Wall Street scrutinizes their path to profitability.
Unrecorded Lease Commitments and Balance Sheet Impacts
The staggering infrastructure outlays are not fully captured on traditional balance sheets. According to Reuters, uncommenced leases—agreements for data center sites that are not yet operational—are generally not recorded as lease liabilities until the facilities become usable. Instead, these future commitments appear only in financial notes.
Together, the top tech groups currently recognize about 285 milliards de dollars in lease liabilities. However, their uncommenced leases total nearly four times that amount. These figures represent future payments spread across many years, unlike balance sheet debts calculated at present value. Microsoft leads the sector with 329 milliards de dollars in uncommenced leases, paired with 88 milliards de dollars in recognized lease liabilities. Meta follows closely with 279 milliards de dollars in uncommenced obligations, having added 68 milliards de dollars in new data center leases in July.
Oracle and the Financial Stakes of Long-Term AI Contracts
Oracle faces the most sensitive exposure relative to its size, carrying 260 milliards de dollars in uncommenced commitments—roughly seven times its 38 milliards de dollars in recognized lease liabilities, according to Reuters data. These contracts primarily target data center buildouts scheduled to launch between the 2027 and 2029 fiscal years, spanning terms of 15 to 19 years.
This dynamic creates a distinct financial vulnerability. Oracle acknowledged that its lease durations, renewal terms, and pricing structures do not always match the contracts it signs with end-users. If enterprise demand for artificial intelligence slows down, the company risks absorbing steep long-term liabilities without matching revenue.
Alphabet reports 85 milliards de dollars in uncommenced commitments, while Amazon records 137 milliards de dollars, though Amazon’s figures also encompass warehouses, offices, aircraft, and vehicles.
Wall Street Scrutiny and Cloud Revenue Growth
Financial markets are increasingly demanding proof of return on investment as capital expenditures soar. According to ZDNet France, cumulative capital spending among the major cloud providers could hit 745 milliards de dollars for the current year alone. This spending surge has already impacted liquidity; Alphabet posted a negative quarterly free cash flow for the first time since its initial public offering, while the combined free cash flow of the four major operators dropped to a ten-year low, as reported by ZDNet France.
Market reaction to these heavy outlays is sharp. Meta shares dropped nearly 8% following an earnings report where investors questioned the monetization timeline for its AI infrastructure, according to ZDNet France. Conversely, cloud divisions at Google, Amazon, and Microsoft report accelerating growth driven by enterprise adoption of AI models from providers like OpenAI and Anthropic. Meta CEO Mark Zuckerberg noted that the company is fielding external requests to rent out data center compute capacity at rates exceeding initial capital costs, pointing to potential new revenue models for hyperscalers.
Related reading