Treasury yield drop may signal deeper AI credit risks

Geopolitical tensions between the US and Iran continue to underpin crude markets. WTI crude has surged past $100 a barrel this week, while Brent reached $114, its highest level since May.

Higher oil would typically increase inflation expectations, pushing Treasury yields upward. That pattern has not held this week, however.

Treasury yields have instead dropped steeply. The 10-year yield dipped back below 5% overnight and has stayed near that level. Falling yields normally provide a tailwind for equities.

Given how much surging yields have weighed on stocks in recent months, relief would be expected. Yet equities have taken a beating this week, with AI stocks hit hardest.

One explanation is that investors are growing concerned about growth once more. Given the severity of the sell-off in AI stocks, though, a different issue may be surfacing below the surface.

CoreWeave provides a useful starting point.

The company may not be a household name in artificial intelligence. Still, CoreWeave could represent the catalyst that triggered a broader reassessment of debt-financing risks across the sector.

Across major AI-related companies, the cost of insuring against defaults, tracked via changes in CDS spreads, has risen sharply over the past twelve months.

The underlying message is clear: investors are increasingly nervous about lending to companies riding the AI boom.

CoreWeave's CDS spread has widened by over 500 basis points over the period, marking it as the clearest outlier.

CDS functions as insurance against a company defaulting on its debt. A five-year CDS covers that period, and the spread is the cost of that protection. A spread of 100 basis points translates to 1% per year.

When CoreWeave's five-year CDS spread exceeded 1,000 basis points last week, it meant insuring $1 million of debt would cost roughly $100,000 each year. That is a significant premium to guard against a possible default.

The concerns are not limited to CoreWeave alone. Oracle's five-year CDS has moved towards 300 basis points, while Meta and Nvidia's have neared 100 basis points.

Oracle is being treated as if it carries junk credit, reflecting its aggressive use of debt for infrastructure. Meta's CDS level is close to its historic peak. The companies are not all in the same risk tier, however: Meta and Nvidia have far stronger financial positions and liquidity than CoreWeave.

Nevertheless, investors are clearly questioning a broader theme. The familiar question remains: who will finance the massive AI infrastructure build-out, and how long will it take for those investments to yield adequate returns?

CoreWeave is heavily dependent on debt for its data centres and computing power. Oracle is committing substantial sums to AI growth, and Meta is spending billions on infrastructure. Nvidia's fortunes are closely linked to that capital expenditure cycle.

The core worry is that rising financing costs or a slowdown in AI investment could ripple through companies that rely on each other for revenue, funding and expansion.

Where does this leave the bond market and Treasury yields?

Against this backdrop, investors are less comfortable holding corporate debt and equities perceived as riskier. The alternative that offers a safe return is government bonds. Demand for Treasuries pushes their prices up and yields down.

There is a catch here. Even as Treasury yields decline, corporate borrowing costs can climb if investors require a higher premium for credit risk. Safe assets get cheaper, while riskier ones get more expensive.

This helps explain why technology stocks are not rallying on the drop in yields. Investors may be more focused on financing risks and future profitability than the immediate advantages of cheaper government borrowing.

AI-related credit concerns are unlikely to be the sole driver of the Treasury rally. Still, they probably form a key component of the explanation for this week's market moves.

With inflation risks remaining and oil prices high, declining Treasury yields do not point toward improving economic conditions. They may instead signal that investors fear something else in the system will break first.