Amazon's $17.5 Billion AI Infrastructure Bet

The terms are exceptionally favorable, a testament to Amazon's investment-grade credit profile:

• Commitment Window: The banks' obligation to lend expires on September 30, 2026, unless the entire amount is drawn before that date, creating a clear, short-term availability window .
• Repayment: Each individual draw under the facility matures three years from the date it is borrowed. This structure staggers repayment obligations to align with the timing of the capital deployments .
• Cost: The interest rate floats at the Secured Overnight Financing Rate (SOFR) plus a margin of just 0.625% to 0.875%, near-optimal pricing for large corporate borrowers in current market conditions .
• Covenants: The loan is "covenant-light," meaning it does not impose financial maintenance covenants that would restrict Amazon's operations if its leverage ratios were to shift .

The official use of proceeds is for "general corporate purposes," a deliberately broad description. However, the context is unambiguous. An Amazon spokesperson indicated to Bloomberg that the purposes include AI-related capital expenditures . The loan follows a trail of other historic debt moves by the company, including a record C$14 billion bond sale in Canada and Amazon's first-ever Eurobond sale in March 2026, which raised €14.5 billion (~$17 billion) .

The $200 Billion Plan: Doubling Down on Data Centers

The loan is a down payment on Amazon's breathtakingly large blueprint for 2026. In February, during its fourth-quarter earnings report, the company disclosed plans for approximately $200 billion in capital expenditures for the year, soaring past the roughly $130 billion it spent in 2025 and far exceeding the ~$150 billion that Wall Street analysts had forecast. CEO Andy Jassy insisted that the spending is overwhelmingly directed at Amazon Web Services (AWS) and is data-driven rather than a speculative land grab, famously stating it was "not a quixotic top-line grab."

The sheer scale of the physical buildout is immense. Amazon added 3.9 gigawatts of power capacity in the twelve months leading up to the announcement: twice what the company had in 2022. It plans to double total capacity again by the end of 2027, monetizing compute resources as fast as they come online. This infrastructure forms the backbone of an industry-wide sprint, with the five largest US hyperscalers—Amazon, Microsoft, Alphabet, Meta, and Oracle—collectively on track to spend $660–$690 billion on AI and data center capex in 2026, nearly double 2025 levels.

Big Tech's New Financial Blueprint: From Cash to Credit

The Amazon loan signals a fundamental shift in corporate finance for Silicon Valley. Historically, cash-rich giants like Google, Meta, and Microsoft funded their moonshots from swelling internal free cash flows. That model has been retired, replaced by an aggressive embrace of the credit markets to front-load AI infrastructure capacity faster than organic profits would allow. The five major hyperscalers issued roughly $121 billion in US corporate bonds in 2025 alone, more than four times their annual average from 2020 to 2024. Analysts project that figure could climb to between $130 billion and $150 billion from these companies in 2026 .

This trend is reshaping the bond market. Amazon and Alphabet together secured a combined $82 billion in a record-setting bond blitz in early 2026 . And in May 2026, Barclays issued a stark warning that the Big Tech borrowing wave is testing the very capacity of the investment-grade bond market, which it believes "won't be able to accommodate all of their financing needs" if spending continues at its current pace. JPMorgan projects the technology sector will need to issue up to $1.5 trillion in debt by 2030 to complete the AI infrastructure buildout, an amount that would significantly shift leverage ratios across the sector .

Return on Investment: The Skepticism Grows

The problem is not that Amazon can't borrow cheaply—it clearly can. The problem is that Wall Street's faith that this spending will generate a meaningful return is eroding. The announcement did not sit well with investors, and the reaction has been sharp and sustained:

• Market Punishment: Amazon's stock fell over 5% on the day of the February capex announcement, its largest single-day drop in six months. The sell-off didn't happen in a vacuum; it was part of a broader wave that erased roughly $1 trillion in market value from Big Tech stocks in early February 2026. The market was sending a clear signal: it is willing to fund a boom, but not a spending abyss.
• Cash Flow Crunch: The scale of the spending will have a direct and painful impact on Amazon's financial statements. Morgan Stanley analysts forecast that Amazon's free cash flow will fall into negative territory in 2026, possibly by nearly $17 billion. This pressure is a direct result of capital-intensive AI projects that may take years to mature into revenue-generating services .
• A Warning from Analysts: Jefferies' global equity strategist Christopher Wood has emerged as one of the most prominent skeptics, warning that hyperscaler spending is now consuming an ever-larger share of their cash flows with no clear timeline for commensurate returns. Wood and others have drawn parallels to the telecom infrastructure overbuild of the early 2000s, cautioning that if AI demand growth slows, a significant portion of this capital could become stranded assets. The core fear, articulated in multiple Reuters and New York Times reports, is that the investments may take years to yield returns, testing investor patience to the breaking point.

A High-Stakes Tension

The new $17.5 billion loan is not a sign of Amazon's weakness; it is a sign of its commitment to a vision. Amazon and its peers are betting that owning the global AI infrastructure stack will be the dominant competitive advantage of the next decade. The favorable terms of the loan show that banks are willing to fund that vision, at least for now. But the market's growing chorus of skepticism reflects a simple truth: with the entire industry sprinting toward a $700 billion annual capex threshold and little visibility on when the revenue will follow, the patience of shareholders is wearing thin. The race to build AI is on, but whether it leads to a new industrial era or a historic hangover is the trillion-dollar question that the loan itself does not answer.