In a move aligning with other major cloud service providers seeking new debt markets to fund artificial intelligence (AI) capital expenditures, Amazon.com has entered the fray. According to informed sources, the company is preparing its inaugural issuance of Swiss franc-denominated bonds. Amazon has mandated BNP Paribas, Deutsche Bank, and JPMorgan to manage the issuance, which will consist of six tranches with maturities ranging from 3 to 25 years.
Regarding Amazon's choice of the Swiss franc market, Apostolos Bantis, Managing Director of Fixed Income at Union Bancaire Privee UBP SA, noted, "The Swiss market continues to exhibit stable demand for high-rated issuers, coupled with strong investor appetite for well-known US corporate borrowers." He added, "Launching bonds across multiple tenors simultaneously allows the borrower to access different investor types, raise more capital, and keep pricing concessions low while the market window remains open."
AI Arms Race Fuels Debt Issuance Surge The explosive growth in computing power demand driven by AI has plunged tech giants into an "invest or fall behind" arms race. Data indicates that the combined capital expenditure for Amazon.com, Microsoft, Meta Platforms, Inc., and Alphabet this year has risen to $725 billion—primarily for AI data center equipment—surpassing the pre-earnings market expectation of $670 billion for these companies.
Against a backdrop of sustained AI demand growth and rising costs for chips and data centers, major tech firms have not signaled any investment cuts. Instead, they have further raised their capital expenditure forecasts. Amazon.com announced an annual capital expenditure plan reaching $200 billion, representing an increase of over 50% compared to its roughly $131 billion expenditure in 2025. Furthermore, Alphabet stated during its earnings call that it expects full-year 2026 capital expenditures to be between $180 billion and $190 billion, raising both the lower and upper bounds of the range by $5 billion. The company also anticipates capital expenditures will "increase significantly" in 2027.
Similarly, Microsoft indicated during its earnings call that it expects 2026 capital expenditures to reach $190 billion, a figure far exceeding the market consensus estimate of $117.5 billion. Meta Platforms, Inc. raised its full-year 2026 capital expenditure forecast to a range of $125 billion to $145 billion, increasing both ends of the range by $10 billion.
For these tech giants, massive investment plans are rapidly depleting cash reserves, making a turn to the debt markets for financing an inevitable choice. To raise enormous funds for AI infrastructure investment, these companies are seeking diversified financing beyond US dollar bonds. Goldman Sachs data shows that issuance of bonds denominated in euros, pounds sterling, and Swiss francs by hyperscale tech companies has risen significantly since 2024.
Karl Schamotta, Chief Market Strategist at Corpay, analyzed that competition in the US dollar bond market is intensifying, prompting companies to seek financing outside the US. This shift also meets global investors' demand for allocating to assets of US AI companies. It is reported that after Amazon.com issued $37 billion in bonds across 11 different tenors in the US bond market on March 10, it issued €14.5 billion (approximately $16.8 billion at the time) in bonds in the euro market the very next day. The company also completed its first US domestic bond issuance in three years last November, raising $15 billion.
Alphabet recently issued the largest euro-denominated bond in its history and launched its first Canadian dollar-denominated bond, raising nearly $17 billion in total, further expanding its non-US dollar financing scale. In fact, since last year, Alphabet has raised a cumulative $86 billion, with nearly half coming from non-US dollar currency financing. In February of this year, the company issued $20 billion in US dollar bonds, setting a record for its largest US dollar bond offering, with peak order book demand reaching $103 billion, far exceeding the fundraising target. Prior to that, the company also issued its first pound sterling and Swiss franc-denominated bonds, continuously diversifying its financing currencies.
Additionally, Oracle issued $25 billion in US dollar bonds in February. Meanwhile, Meta Platforms, Inc. filed a $30 billion debt issuance plan last October for AI infrastructure expansion.
Some data directly illustrates the scale of this AI-led debt financing. By the end of 2025 and early 2026, the total debt related to AI infrastructure had swelled to approximately $1.2 trillion. Within JPMorgan's US High-Grade Liquid Corporate Bond Index, the AI-related sector now accounts for 14%, officially surpassing the US banking sector to become the largest single weight in the investment-grade bond market.
Morgan Stanley estimates that these hyperscale tech companies could borrow up to $400 billion in total this year, significantly higher than the $165 billion in 2025. This is expected to drive global high-grade bond issuance to a record $2.25 trillion this year.
Return on Investment Remains a Concern; Bond Market Shows Signs of Fatigue While tech giants are issuing debt on a large scale, and bond investors generally show high confidence in their AI investment thesis based on subscription levels, some bond investors have raised concerns about the rapid pace of spending and uncertainties in the monetization cycle. Some investors point out that current AI commercialization remains limited, relying mainly on cloud service price hikes and enterprise subscriptions, which extends the payback period. If AI returns fall short of expectations, the maturity of large-scale debt could trigger refinancing pressure, impacting both bond and equity markets.
Concurrently, following such a massive wave of AI-related debt, investor fatigue is beginning to show. The most direct example is Meta Platforms, Inc.'s issuance of up to $25 billion in investment-grade bonds earlier this month, which attracted a peak order book of approximately $96 billion. While this figure is substantial, it represents a noticeable contraction compared to the $125 billion in demand attracted by the same issuer's $30 billion bond offering last October.
Other signs include an issuer related to SoftBank Group being forced to increase the offering yield due to insufficient demand to complete the financing. Investors are beginning to demand stronger protective covenants—including guarantees from Alphabet to cover data center rental payments in case of tenant default. Some investors are even outright rejecting certain deals. One investor stated they passed on a $14 billion bond for an Oracle data center in Michigan, partly because the bond contained call provisions unfavorable to creditors.
Robert Tipp, Global Head of Bonds at PGIM Fixed Income, stated, "Ultimately, these companies are selling a lot of debt, and they are going to have to pay up to borrow. The market, after corporate spreads tightened dramatically to historically low levels, is facing a wall of worry."
John Servidea, Global Co-Head of Investment Grade Debt Capital Markets at JPMorgan, said, "We are seeing what different investors value in these financings, how they assess risk versus return. We are seeing these trades being fairly well bid, but as supply increases, we expect terms and structures to continue to evolve."
More importantly, analysis suggests this trillion-dollar "AI borrowing spree" is transferring systemic risk across boundaries. One risk is the inversion of technological depreciation. The biggest hidden danger for bond investors currently is "computing obsolescence." Chip iteration is extremely rapid (e.g., from H100 to B200 to next-generation architectures). If existing GPU clusters become obsolete quickly, their asset lifespan could be far shorter than the 10-year or even 30-year debt they carry.
The other aspect is the transfer of risk to the credit market. In the past, Silicon Valley's technological gambles were primarily funded by venture capital and equity shareholders. Now, with the surge in debt scale, the cost of AI experimentation has substantively shifted onto the shoulders of pension funds, mutual funds, and fixed-income investors who purchase high-grade bonds.
Comments