US Treasury Yields Collapse The Last Two Weeks. What’s Going On?

It’s not sinking inflation or falling oil. So what is it?

Yield Synopsis

30-Year: 5.30 percent down from 5.67 percent on October 7
10-Year: 4.89 percent down from 5.31 percent on October 5
2-Year: 4.67 percent down from 4.92 percent on September 28
3-Month: 4.17 percent down from 4.28 percent on October 15

The Fed

The Fed meets on October 28 with little chance of a hike
The Fed meets next on December 9 with a 75 percent chance of one quarte-point hike and an additional 7.5 percent chance that the Fed hikes by a half point.

That’s almost no change from a week ago, so this does not seem to be the catalyst.

Not Falling Inflation

Inflation has been hot. The CPI, PCE, and PPI have all been above consensus estimates.

The Cleveland Fed inflation Nowcast is for another hot PCE report month-over-over month of 0.47 percent for September with year-over-year PCE of 3.72 percent for October.

Something Consequential

View post on X

I don’t buy the recession theory yet. I do believe an AI-related credit bust is coming.

AI Bust

View post on X

Corporate Credit Spreads

View post on X

Nick Timiraos WSJ Columnist

View post on X

Tech Jitters

View post on X

Darker Side of the AI Boom

View post on X

AI Default Risk

Please consider Falling Treasury yields may be hiding a darker story about the AI boom

Let’s take a step back and check on the key market drivers this week. The US-Iran conflict is still raging on and geopolitical tensions remain high. So, that is underpinning oil prices once again with WTI crude breaching $100 and Brent crude hitting $114 – its highest since May.
Now, that sounds like a familiar backdrop for the bond market, doesn’t it? Higher oil prices mean stronger inflation pressures, which should normally push up Treasury yields. Well, that has not been the case this week.
Instead, Treasury yields have fallen sharply, with 10-year yields even slipping back below the 5% threshold in overnight trading and holding thereabouts since. But wait a second, shouldn’t falling yields then buy some relief for stocks?
You would certainly think so, especially when surging Treasury yields have been such a major bane for equities over the past two months. And yet stocks have been battered this week, with AI names in particular bearing the brunt of the pain. So, what gives?
It’s one thing to suggest that investors are perhaps starting to worry about growth conditions again. But considering how long that narrative has been brushed aside and the extent of the damage we are seeing in AI stocks, I think there is perhaps a rather different problem brewing beneath the surface.
Let’s jump straight into it and start with CoreWeave, shall we?
That’s not the most familiar name in the AI space. However, CoreWeave might well have been the stone that started an avalanche of worries about debt financing in the entire sector.
Just take a look at the chart [above], which shows how the cost of insuring against defaults among major AI-related companies has surged over the past year i.e. change in CDS spreads.
CoreWeave stands out like a sore thumb, with its CDS spread widening by more than 500 basis points over the observed period.
And while CoreWeave is clearly the standout, the chart also shows that the worries aren’t confined to just one company. Oracle’s five-year CDS has climbed towards 300 bps, while Meta and Nvidia have seen theirs approach 100 bps.
In the case of Oracle, it is being traded like junk credit due to aggressive debt-fueled infrastructure loans. Meanwhile, the five-year CDS level for Meta is near its historic peak so that helps to put things into better perspective. Having said that, I wouldn’t put all these companies in the same risk category. Meta and Nvidia clearly have much stronger financial positions and liquidity than CoreWeave for example.
However, it is clear that investors are beginning to question something bigger. And that brings us back to the same question we’ve been asking all year, that is who is going to pay for the enormous AI infrastructure boom, and how long before those investments generate sufficient returns?
CoreWeave relies heavily on debt to finance data centres and computing infrastructure. Meanwhile, Oracle is also committing massive sums to AI expansion and Meta is also pouring billions into infrastructure. As for Nvidia, their fortunes are very much closely tied to that spending cycle.
The major concern here is that if financing becomes more expensive or AI investment slows, the consequences could spread across companies that depend on one another for revenues, financing and growth.

Does anyone believe profit will support the money being spent, most with circular financing?

AI Race Puts $3.5 Trillion at Write-Down Risk

Bloomberg reports AI Race Puts $3.5 Trillion at Write-Down Risk, Gartner Warns

On October 18, 2026 I noted Data Centers Need $8.5 Trillion in Revenue Through 2030 to Cover Spending

How likely is that?
The AI Bubble Will Burst
But what inning is this? 7th, 8th, or 9th?
If you prefer to think in years, is it more like 1997 or March of 2000?

This is a theme I have been discussing for a long time.