Breaking

Oracle's Credit-Default Swaps Reach Record High as AI Spending and Chinese Model Threat Rattle Investors

July 17, 2026 at 21:26 EDT

Finance
Infra & Chips
Foundation Models

A closely watched gauge of Oracle Corp.'s credit risk climbed to a fresh all-time high, underscoring mounting investor anxiety over the company's aggressive cash burn on artificial-intelligence infrastructure and the uncertain payoff of those bets as low-cost Chinese models advance. The five-year credit-default swap on Oracle rose roughly 10 basis points to 198.23 on July 17, edging past the previous peak of 198.18, according to ICE Data Services. A higher CDS level signals that traders are demanding more to insure against a potential default, a barometer of deteriorating market confidence.

July 17, 2026 · Oracle Credit Watch

Oracle's Debt-Fueled AI Bet Rattles the Market

A key gauge of Oracle's credit risk hit a record high as investors question whether a debt-financed data-center build-out will pay off — just as cheap Chinese AI models begin to undercut the economics.

198.23

5-yr CDS — fresh all-time high (bps)

~$130B

Total borrowings, after ~$43B raised in FY2026

−25%+

Oracle shares in 2026

The backlog exploded — and so did spending to fulfill it

Remaining performance obligations (contracted future revenue), year over year.

~$138B

Prior year

$638B

FY2026 (+363%)

+$85B in a single quarter, driven by giant AI contracts including a reported multibillion-dollar OpenAI deal. But fulfilling it demands sustained, capital-intensive spending.

Capex overshoot

$55.57B

FY2026 capital expenditure — above its own "$50B+" target, keeping free cash flow negative.

Rating pressure

BBB−

S&P cut Oracle to one notch above junk, citing heavy AI-related spending.

The new threat: near-frontier AI at a fraction of the cost

Z.ai's open-weight GLM-5.2 trails top proprietary models by ~1 point on benchmarks like FrontierSWE — but at roughly one-sixth the price.

6× cost

Proprietary
(Claude / GPT)

1× cost

GLM-5.2
(open weights)

If frontier-level capability becomes dramatically cheaper and freely available, the economics of long-dated AI infrastructure contracts look far less certain.

The bull case

A record $638B backlog signals enormous, contracted demand for AI capacity — evidence the build-out is meeting real customer need.

The bear case

Debt-funded, negative free cash flow plus cheaper rivals could squeeze margins before returns ever materialize.

The record CDS reading crystallizes one question: will the debt-funded race for AI capacity deliver durable, high-margin returns — or will cheap alternatives and financing costs win?

Until free cash flow turns positive, the market looks set to keep pricing in elevated risk.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.