Global AI big tech companies are experiencing significant revenue growth due to rising demand for AI, but their debt levels are also surging, primarily due to excessive infrastructure investments. Funding for AI investments is increasingly shifting from operating cash to bonds, new equity, and guarantees.
According to the IT industry on October 22, Meta's long-term debt rose from $59 billion at the end of March to $80.9 billion by the end of September, an increase of over $22 billion in just six months.
In its third-quarter earnings report, Meta announced a revenue increase of 26% to $64.6 billion, but capital expenditures reached $34.91 billion, accounting for 54% of its revenue. This suggests that the company spent about $2.1 billion more than it generated in cash flow.
To cover the shortfall, Meta has turned to external funding. Since November of last year, the company has raised $55 billion through bonds, including $30 billion in October and $25 billion in April. Notably, it did not repurchase any shares in the first quarter. As of June 30, Meta's net cash, excluding bonds, was approximately $6 billion, down from about $22.6 billion at the end of the previous year.
In its joint venture with Blue Owl to build the Hyperion data center, Meta is also carrying off-balance-sheet special purpose vehicle (SPV) debt amounting to approximately $27.3 billion. Additionally, the board approved the creation of non-voting Class C common stock, effective January 2027.
The financial situations of five major tech companies—Alphabet, Amazon, Meta, Microsoft, and Oracle—are similarly concerning. These companies issued approximately $220 billion in bonds by August this year, nearly 1.8 times the $121 billion issued in the previous year, which is roughly equivalent to 300 trillion won.
As of June 30, Amazon's long-term debt stood at $128.9 billion, while Alphabet's total debt exceeded $100 billion, including bonds with maturities of up to 100 years. S&P Global Ratings estimates that the capital expenditures of these five companies will reach about $750 billion this year, representing 38% of their combined revenue.
According to Barclays, AI-related bonds account for about 15% of U.S. investment-grade corporate bond issuance, and JPMorgan forecasts that global technology, media, and telecommunications (TMT) bond issuance will reach a record $540 billion this year.
Oracle faces a heavier burden, with long-term debt reaching $149 billion at the end of the recent fiscal year, an increase of $53 billion in one year. However, its cash reserves are only $31.3 billion, about one-fifth of its debt. The company's credit rating has been downgraded to BBB. OpenAI has also reportedly reduced its capital expenditures by nearly half due to investment burdens and is seeking an additional $30 billion in funding.
Anthropic's revenue surged to $16.3 billion in the first half of this year, up from $1.1 billion a year earlier, but its infrastructure spending increased nearly fourfold from $2.3 billion to about $9 billion. The company's net loss for the first half also rose to $80.7 billion, six times higher than the $13 billion reported in the same period last year, with $72 billion attributed to non-cash expenses related to convertible bond valuations.
The debt cycle extends to chip suppliers and model companies as well. According to Forbes, citing Nvidia's quarterly report, the company's commitments for data center leases and capacity guarantees amount to as much as $164.5 billion, with direct investments or commitments to AI cloud companies exceeding $7 billion.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.