MoneySpider Weekly — 25 October 2026
On Thursday the 22nd, the 10-year Treasury yield fell 17 basis points in a single day, from 5.13% to 4.96%, and over the week it went from 5.09% to 4.96%. The fall was in real yields (the return after inflation, down from 2.74% to 2.64%) while expected inflation stayed flat around 2.3%. It came with Brent crude near $116 a barrel and UK inflation rising to 3.5% from 3.1%. Investors were buying safety, not shrugging off prices.
What they were buying safety from widened at the edges. The spread on US junk bonds rose from 3.11% two weeks ago to 3.40% on the 22nd, the widest since at least March, with the riskiest CCC-rated borrowers at 13.40% and even investment-grade spreads drifting up, from 0.85% to 0.89%. In equities the Nasdaq fell 0.7% on the week, its second weekly loss, after a 2.6% fall on Thursday, while small companies rose 2.0%. Meta reported revenue up 26% but free cash flow of about −$2.1 billion after capital spending, and the cost of insuring CoreWeave’s debt passed 1,100 basis points. Revenue is growing; cash flow, at the company we can read, is not.
This week’s signal: Tightening Into Fracture — the crack widens. Confidence stays Medium, capped: the credit and Treasury figures are confirmed from primary sources, but the reading that stress is broadening around the AI-financing chain is ours, and jobs data (US jobless claims 194,000, UK unemployment at a one-year low) is not confirming it. Two falsification tests carry forward: evidence that the numbers reflect a one-off data-quality air pocket (we didn’t find one), and junk spreads falling back below about 2.8% (they are at 3.40%). The strongest alternative we can see is that AI spending turns into above-trend growth; if it does, revenue is the real story and credit spreads are noise around a productivity upswing. We don’t find the growth half of that case in this week’s evidence. The first official test is the US growth estimate for the third quarter on 29 October, after the Fed and the UK Budget on the 28th.