3 weeks ago
Big Tech Borrowing Nearly $500 Billion for AI Boom
Big technology companies are borrowing a lot of money to build things for artificial intelligence, like giant computer centers.
Almost $500 billion has been borrowed this year.
When people lend money, they sometimes buy insurance in case the borrower cannot pay it back.
The cost of that insurance is going up.
That means the people lending money are a little more worried about getting it back.
The biggest worry is about a company called Oracle.
Other companies like Nvidia, Meta, and Broadcom are also on the list.
These companies want to build the best AI computers and are spending huge amounts of money to win the race.
Some people worry the money they spend might not earn back enough profit.
But experts also say these companies are still very big and strong, so they will probably keep paying their debts.
Goldman Sachs Research estimates that nearly $500 billion of AI-related debt has been issued so far in 2026.
Oracle's five-year credit default swap spread jumped 70 basis points in 2026 to around 215 basis points, the largest increase among major technology companies tracked.
Broadcom's CDS rose 48 basis points, while Meta's climbed 39 basis points to about 95 and Nvidia's rose about 32 basis points to roughly 82.
Costs of insuring Big Tech debt against default are climbing, signaling higher investor concern about credit risk, though default is not necessarily expected.
Nikkei Asia reported that Alphabet, Microsoft, Amazon, Meta and Oracle may have around $1.65 trillion in liabilities outside their public balance sheets.
- Who
- Major AI-linked technology companies, including Oracle, Broadcom, Meta, Nvidia, Amazon and Alphabet, along with their investors.
- What
- Nearly $500 billion in AI-related debt has been issued in 2026, and the cost of insuring that debt against default is rising sharply.
- Where
- Global credit markets; no specific geographic location was reported in the articles.
- When
- In 2026, with credit default swap data as of July 28.
- Why
- To finance the massive AI infrastructure buildout of data centres, computing equipment and power, amid uncertainty over whether the investments will generate adequate returns.
Skeptical View
Confident View
Is the AI debt risky?
Skeptical View
Rising leverage, enormous capital expenditure and uncertainty over AI returns mean the investments may not pay off before the costs of infrastructure are fully recovered.
Confident View
These companies have enormous cash reserves, strong balance sheets and substantial access to capital; the widening spreads do not mean default is expected.
Off-balance-sheet liabilities
Skeptical View
Nikkei Asia's investigation suggests tech giants may have about $1.65 trillion in liabilities outside their public balance sheets, raising concerns about hidden risk.
Confident View
Off-balance-sheet liabilities include contractual commitments and financing structures and do not necessarily represent conventional debt or an immediate repayment obligation.
Funding strategy: debt vs. shares
Skeptical View
Borrowing more raises credit risk and could pressure valuations if AI spending fails to generate adequate returns.
Confident View
Debt appropriately finances long-term infrastructure projects; issuing new shares instead could dilute existing shareholders.
Key facts
- AI-related debt issued in 2026
- Nearly $500 billion (Goldman Sachs Research estimate)
- Oracle five-year CDS spread
- About 215 basis points, up 70 bps in 2026
- Meta CDS spread
- About 95 basis points, highest since trading began in October 2025
- Nvidia CDS spread
- About 82 basis points, highest since trading began in November 2025
- Broadcom CDS increase
- 48 basis points in 2026
- Estimated off-balance-sheet liabilities
- About $1.65 trillion for Alphabet, Microsoft, Amazon, Meta and Oracle (Nikkei Asia)
- Data date
- July 28 (S&P Global Market Intelligence/LSEG data)







