Amazon secures $17.5 billion loan to fund AI infrastructure spending

Amazon just added another layer of financing to the AI boom: a $17.5 billion borrowing facility that shows how even one of the most cash-rich companies in America is reaching for debt to keep up with the race to build cloud and AI infrastructure. The move points to the scale of the capex cycle now underway, and to the bet that today’s spending spree will translate into durable demand for compute, chips and data centers.

The facility is a senior unsecured delayed-draw term loan credit agreement, which means Amazon does not have to take down the full amount immediately. It can draw funds as needed before commitments expire on September 30, 2026, and each borrowing matures three years after it is drawn. Citibank led the lending group, with Bank of America Securities, JPMorgan Chase, HSBC and Wells Fargo among the lenders.

The financing lands against a dramatic jump in Amazon’s spending. The company said it expected to invest roughly $200 billion in capital expenditures in 2026, much of it tied to AI infrastructure. In the first quarter of 2026, Amazon reported net sales of $181.5 billion, up 17% from a year earlier, while AWS sales rose 28% to $37.6 billion. But trailing-12-month free cash flow fell to $1.2 billion from $25.9 billion a year earlier, driven primarily by a $59.3 billion increase in purchases of property and equipment, net of proceeds from sales and incentives. Amazon said first-quarter capital expenditures reached $44.2 billion.

Chief executive Andy Jassy has argued that the spending is a calculated response to a once-in-a-generation shift. He said the company is not investing about $200 billion in 2026 “on a hunch,” and that when technological changes are this large, “you want to bet big.” Amazon has said AWS’s AI revenue run rate was above $15 billion in the first quarter, while its custom chip business had reached an annual revenue run rate of more than $20 billion and was growing at triple-digit percentages year over year.

For investors, the message is bigger than one loan. Hyperscalers are increasingly using credit facilities, bonds and other structured borrowing to fund the physical buildout behind AI, turning software optimism into a capital-intensive industrial project. If AI demand keeps accelerating, the returns could justify the spending. If it cools, the industry will be left with a far larger bill.