Yield stress, Rosenberg's rally and AI debt

2026-10-02 · What the week changed · What the week moved, Rosenberg's bond rally, tested, Hyperscaler debt and Nordic funding

Daily show11:14Download MP3About the show

How it is made. Aavistus Daily is written and voiced by our own systems, every morning, from data we gather ourselves. The voices are synthetic. Every fact is traced to a source on this page, and every dated call is settled in public on the calibration page. The presenters, Nora and Callum, are named characters, not people.

Aavistus is a new framework, still under development. We will make mistakes, and every dated call is settled here in public so you can see them; the aim is to keep calibrating until the record beats the average forecaster, not to sound right.

A week of duration pain, French bond stress and a mixed Nord Pool, with David Rosenberg's bond-rally case and the hyperscaler debt binge tested.

Aavistus briefs. Independent market intelligence — satellites, flows, filings. Register free to follow new briefs.

What the week moved

Three things are materially different on Friday from how the week opened on Monday.

The duration-pain picture tightened at the long end. Hannah and John on Brew Markets Daily put the Q3 move at about 85 basis points on the ten-year with an intraday touch of 5.34. The high-yield option-adjusted spread, which we flagged on Tuesday as a marker worth watching off historic tights, moved from 2.73 to 3.12 — 39 basis points of widening in a week when the S&P gave up less than half a per cent. The 10-year–2-year curve steepened from 26 to 41 basis points. This is term premium rebuilding into supply, not a credit accident building. The US labour market has not broken. Brew Markets framed the week as strong data meeting a souring consumer; we agree the labour side gives the Fed room, and push back on consumer softening as the main story — the curve is still pricing roughly five further 25-basis-point hikes in a year. Our standing view that the 10-year holds 5 to 5.5 per cent through year-end with a modest S&P de-rating is intact; what would make the read wrong is five consecutive daily closes above 5.4 with no equity break.

France deepened as a problem. Bloomberg's framing this morning is explicit, and the flow story is named Asian asset managers exiting French bonds on fiscal concerns. Our standing view is the European Central Bank keeps hiking — deposit facility rate to 3.25 per cent or higher by mid-2027, next decision 29 October — and if that is the path, French spreads over the Bund stay under pressure. The defensive read on European financials into year-end holds.

Nord Pool landed half the base case. Finland's weekly average fell 26 per cent to €58/MWh and southern Sweden's SE3 fell 23 per cent to €84, confirming the wind-returns call set out last Thursday. But southern Norway (NO1/NO2/NO5) and Denmark west went the other way, up 8 to 11 per cent on the week. The region is not one weather. Treating Nord Pool as one zone is the error, and the week showed it.

Rosenberg's bond rally, tested

David Rosenberg argued on MacroVoices this week that the recent surge in the US 10-year is mostly real rates rather than inflation expectations, and that November midterm gridlock plus Treasury maturity management will trigger a major bond rally. He is heavily positioned in two-year notes and scaling into longer duration.

On the decomposition we agree: the move is term premium, not inflation expectations. On direction we agree. Where we disagree is magnitude and timing. The labour market Rosenberg leans on for disinflation is softening at the edges but not breaking. CPI is still 3.4 per cent. The supply story, which he himself concedes, is a structural weight, not a cyclical one, and gridlock does not fix it. Three conditions would make his rally right: today's payrolls clearly below consensus with upward revisions to unemployment, Core Personal Consumption Expenditures below 2.5 per cent annualised in late October, and a dovish ECB into 29 October. Short of that trio, the range view holds.

Hyperscaler debt and Nordic funding

Emily Herbert, on the Financial Times podcast Unhedged with Katie Martin and Robert Armstrong, laid out the hyperscaler debt binge: heavy AI-related issuance year-to-date with the hyperscalers taking the lion's share, projections of much more by the end of the decade, a very large single euro transaction from Amazon and a 100-year sterling bond from Alphabet. Bessent and Warsh argue this crowds out Treasuries; Armstrong disagrees.

Our read: Armstrong is right narrowly — dedicated sovereign buyers insulate the government curve. Martin's cross-asset framing is the useful one. The 39-basis-point widening in high-yield OAS this week without an equity break is the fingerprint of supply pressing against demand at the mid-grade of the credit stack. The hyperscalers are investment grade and not in that index, but every billion they issue absorbs allocation that would otherwise be taken by names that are.

The Nordic consequence sits in bank wholesale funding programmes. Nordea and DNB can absorb a 30 basis-point widening on their offshore dollar programmes without the net-interest-margin story breaking. The read changes if a Nordic bank's Q3 release in October discloses a dollar funding cost increase above 50 basis points year-on-year with no reinvestment benefit — on more than one name, it says the credit channel is biting. The case for adding Nordic bank equity on a payroll-driven dip today is not there because the rate path is already in the price; the case for trimming into it is not there either, because the funding channel has not yet shown up in a disclosure.

Calls

Sources