The AI panic test and a BoJ hike signal

2026-10-21 · United States · US AI trade and the credit tape, UK Budget and the global long end, Bank of Japan hiking signal

Daily show10:21Download 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.

The US semis sell-off tested against the credit tape, the UK Budget as a lesson for long-end yields, and a Bank of Japan line that fires the hiking view.

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

The AI panic test

Jack Farley on Monetary Matters argued yesterday with his co-host Max that the AI panic in semis is overdone: hedge-fund positioning is negative, enterprise demand is intact, and sell-side estimates on names like Lam Research are sandbagged. Their clincher is the SpaceX compute figures they walk through on the show.

From here, the direction they describe is consistent with the standing view that this is a slow index-level de-rating rather than a credit break. The semi complex and the biggest AI-exposed names are leading the broader US index lower this week; the spread between the headline index and the AI-levered names is the de-rating in daylight.

Where I don't follow Farley is on CoreWeave. He and Max read the elevated yields on CoreWeave debt as a dislocation to lean into. Don Wilson on Odd Lots frames the same gap as a market-structure problem rather than a mispricing to buy: compute futures are not yet standardised, so neocloud providers at the lower end of the credit stack pay the full capital cost of unhedged hardware risk. Both readings sit inside a slow de-rating, not a break.

The convergence that would complete the de-rating is a US high-yield spread crossing four hundred basis points together with a top-four hyperscaler cutting 2027 capex guidance by more than ten per cent. The high-yield OAS is three point two nine per cent, up from three point one one a month ago — moving the right way for the standing view, not against it.

Where I'd push back on Farley is on the defensive rotation. He treats the move into healthcare, mature software and financials as mechanical short-covering with no fundamental catalyst. I'd leave more room than he does for earnings-driven moves inside that rotation. The read for a Nordic book is unchanged: short-dated euro and US government bonds over US equity.

The next FOMC. The US curve prices a one-year rate of five point zero seven per cent from a policy rate of three point eight seven five — close to four and three-quarter twenty-five-basis-point moves over twelve months. With the standing view that the Fed funds upper bound reaches at least four point seven five by mid-2027 from four per cent today, and CPI still running sticky, my read on the fourth of November is a twenty-five-basis-point hike.

The UK Budget and the global long end

Chris Giles on Unhedged with Katie Martin argued yesterday that the upcoming UK Budget has to be a boring budget because the gilt market cannot absorb fiscal slippage with ten-year gilts at five point four per cent, thirty-year gilts at five point nine, and issuance running at eight to nine per cent of GDP. UK growth has beaten expectations at one point eight per cent through August.

Even with growth stronger than feared, Rachel Reeves cannot credibly loosen into this curve. The standing view says the US thirty-year grinds higher by mid-2027 — it is at five point four two today, up thirteen basis points on the month. Where I'd reframe Giles is on the mechanism. He treats the UK long end as a borrowing-cost problem for the Chancellor. I'd read it as term premium rebuilding across all developed-market long ends simultaneously. The UK thirty-year at five point nine is on the same rope as the US thirty-year at five point four two and the euro-area ten-year AAA at three point four seven.

Jim Bianco on Thoughtful Money with Adam Taggart pushes that argument further. His case is that long rates ultimately track nominal GDP, and if AI delivers productivity gains nominal GDP could run at seven to eight per cent — pulling the US thirty-year toward seven or eight. The standing view already points toward a higher US thirty-year by mid-2027, so Bianco's number sits at the aggressive end of the same trajectory. His supporting point on persistent fiscal deficits anchoring the floor on long rates higher than consensus assumes is why I agree on direction. The qualifier is the horizon: Bianco's number is a two-to-four-year destination, not a next-year print. What matters for the UK Budget is that the gilt curve is already behaving as if Bianco's thesis is partly priced — which is why Reeves has no fiscal room.

Through the Nordic book. Euro-area HICP at three point eight in September with core at two point five sets the ECB on the twenty-ninth. My view is a hold at that meeting with the deposit facility at two and a half per cent, and the standing view that the cycle reaches at least three and a quarter per cent by mid-2027 is intact — that view is open and settles next week. EUR/SEK is eleven point one two, inside the standing range above eleven.

A Bank of Japan hiking signal

Kyodo led this morning with a line that the BoJ will decide at its next meeting whether the two-per-cent inflation target has been reached; Bloomberg carried an ex-official saying the BoJ is likely to keep hiking every quarter. The policy rate is one point two three per cent after the September move. The two-year JGB is one point nine one, the ten-year is two point nine three — both near the top of the range for this series. A central bank does not publicly commit to deciding whether a target has been reached unless it expects to say yes. The standing view that the BoJ reaches one point five per cent within nine months is on track.

The channel to a Nordic book is EUR/JPY. The pair is one hundred and seventy-seven point three four, and the standing view is that it drifts below one hundred and seventy-three by spring 2027 as the ECB-BoJ gap narrows. The gap is one point two seven percentage points today; both legs are moving the right way. A yen-long sleeve funded in euro is where the setup still points. What would change this read is a BoJ statement that reads clearly as a pause and a two-year JGB that gives back twenty basis points on the day.

Calls

Sources