Tech companies are issuing debt at a rapid pace this year, even as borrowing costs climb, according to a Bloomberg report. Nearly $500 billion in AI-related bonds has been sold in 2026 to date, with the proceeds aimed squarely at data centers, AI chips, and the infrastructure that ties them together. The pace is significant because it shifts the financing of the AI buildout into public debt markets, where bond investors price and track the costs and risks of that expansion as it happens.

Who Is Borrowing, and How Much

The largest borrowers are the so-called hyperscalers—industry shorthand for companies operating cloud computing at enormous scale: Amazon, Alphabet, Meta, Microsoft, and Oracle.

Collectively, the hyperscalers have issued approximately $200 billion or more in new debt in 2026. Amazon recently raised roughly $54 billion, Alphabet issued $31.5 billion, and Meta tapped the market for $30 billion. Oracle's issuance falls somewhere in the $18 billion to $30 billion range.

Beyond the cloud giants, Broadcom has also accessed the bond market, and SpaceX completed a $25 billion bond deal in June 2026. That SpaceX offering was oversubscribed, meaning investor orders exceeded the bonds available for sale.

Why the Timing Stands Out

The surge comes despite an increasingly expensive rate environment. Long-term Treasury yields have reached multi-decade highs, and corporate borrowing costs tend to move in tandem with them, since corporate bonds are typically priced at a spread over this benchmark. Every step up in the base rate raises the coupon companies must pay on newly issued debt.

Analysts at Bank of America raised their forecast for 2026 hyperscaler debt issuance from $140 billion to $175 billion following the latest wave of deals.

Projections for total AI-related financing this year vary widely, ranging from $300 billion to upwards of $570 billion. Goldman Sachs estimates that total AI debt could potentially exceed $1 trillion annually in the coming years.

The Revenue Story Behind the Borrowing

Underpinning the spending is an expected revenue ramp. OpenAI predicts annual revenues of $70 billion or more by year-end 2026. Because bondholders ultimately look to a borrower's revenue and cash flow to service its debt, projections like this serve as a reference point for whether the buildout's obligations can be supported.

Credit Markets Are Starting to Flinch

Credit investors are beginning to price in greater risk. Broadcom's five-year credit default swap (CDS)—a contract whose cost reflects what investors will pay to insure against a default—spiked to a record 136 basis points, or 1.36 percentage points, reflecting heightened volatility and increased perceived credit risk around one of the sector's key chip suppliers.

Market-implied five-year default rates have climbed for other borrowers as well. These figures represent the probability of default within five years that market pricing suggests investors currently assign to each borrower. Oracle's sits above 20%, while SpaceX's is around 16%.

What This Means for the Sector

If Goldman Sachs' projection of more than $1 trillion in annual AI debt comes to pass, traditional credit markets would be absorbing a volume of issuance that could test their capacity. Gauges such as CDS levels and market-implied default rates offer a running measure of how investors are weighing that debt load as it builds.

Key things to watch from here: whether Bank of America and other analysts revise their forecasts again, whether CDS levels for Broadcom, Oracle, and SpaceX keep climbing, and whether demand for new tech bond deals remains as strong as it was in June.