Amazon plans first Swiss franc bond sale for AI

Amazon is preparing its first Swiss franc bond sale to help fund AI infrastructure and has hired BNP Paribas, Deutsche Bank, and JPMorgan to arrange six tranches from three to 25 years.

The move adds to a wider shift by big tech companies beyond US dollar debt as they raise money for data centers and other AI spending.

Alphabet raised about 3 billion Swiss francs in February in the largest Swiss franc bond sale by a company, and Amazon’s debut euro bond in March set a similar record in that market.

Amazon, Meta, Microsoft, and Alphabet are expected to spend as much as US$725 billion this year on AI data centers, equipment, and other capital expenditure.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

AI spending is squeezing cash flow

Amazon’s spending plans go well beyond the AI infrastructure amounts listed here, adding more strain to cash flow 1.
That strain is clear in the numbers. Operating cash flow reached about $130.7 billion, yet free cash flow was about $14.6 billion after equipment purchases jumped by more than $50.9 billion from a year earlier 2.
Amazon is also borrowing to help fund that gap. The company is targeting $37 billion to $42 billion equivalent through U.S. dollar and euro bonds, which would lift total debt above $100 billion 1.
Its move into Swiss franc bonds follows Alphabet, which raised almost $32 billion in less than 24 hours across U.S. dollar, sterling, and Swiss franc markets 3.

Tech borrowing for AI could reshape bond markets and chips

Amazon is part of a wider borrowing wave. Morgan Stanley estimates AI capital spending by hyperscalers, very large cloud companies such as Amazon, Microsoft, and Alphabet, will reach $2 trillion from 2025 to 2028, with more than $1 trillion funded by new debt 1.
The global bond market is large, yet some investors worry that repeated multi-billion-dollar sales from a small group of tech companies could stir volatility and push borrowing costs higher 4.
For Amazon, the debt helps fund AI infrastructure and Amazon Web Services (AWS), the company’s cloud division, including Trainium, a custom chip built to train AI models 2.
AWS wants more AI workloads to run on its own chips, which would cut reliance on Nvidia’s AI accelerators, the specialized processors used to train and run AI systems. Amazon estimates this lowers total cost of ownership, the full cost of buying and operating a system, by 30% to 40% while keeping margin that would otherwise go to Nvidia, whose product gross margins are about 60% to 70% 2.