The CoreWeave short case, tested
2026-10-25 · The investor's Sunday · Noble's CoreWeave short case, The week's rate decisions and prints
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.
George Noble says CoreWeave is a zero — we take the claim apart, ask what would prove him wrong, then look at the week's rate decisions.
Noble's CoreWeave short case, tested
This week George Noble took a strong, specific view on Monetary Matters with Jack Farley: CoreWeave's equity is going to zero. He named the mechanism — GPU-collateralised debt against a rapidly depreciating hardware base, with compute rental rates he thinks will collapse once supply normalises. The business model, he said, is WeWork: long debt, short asset life, no self-funding cash flow.
The business-model direction, the debt-pricing framing, and the moat case all hold up. The hardware depreciates faster than the paper matures; the duration mismatch is real. Compute rental rates have no monopoly protection — the hyperscalers' own supply plus credible Chinese open-weight competitors keep the ceiling on. Capability convergence between frontier Western proprietary and credible Chinese open-weight has narrowed sharply this year, enough to undercut the pricing power that justifies current frontier-layer multiples.
The disagreement is on timing of the broad-market consequence. On CoreWeave itself, Noble's case is defensible on its own terms. On the index-wide break, the standing view has been for weeks that this is a slow index-level de-rating, not a credit break, and the convergence that completes it is two legs firing together — high-yield spreads crossing four hundred basis points, and a top-four hyperscaler cutting 2027 capex guidance by more than ten per cent. US high-yield spreads sat at three point four per cent as of Thursday, up from three point two three on the previous reading: moving in the direction, still sixty basis points short of the level.
The read. The case for stepping into neo-cloud equity at this price is not there. The trigger that would change the read on the sector is either leg of the de-rating convergence firing. What would make the broad index view wrong is either leg firing without the index dragging with it.
The week ahead
Five events over the next twelve days.
Thursday 29 October — ECB. The hold at two and a half per cent on the deposit facility rate is the view set on the fifteenth and still intact. The Council does not move back-to-back without core-inflation confirmation; September headline inflation at three point eight per cent, up from three point two, is energy pass-through. The cycle out through at least three and a quarter per cent by mid-2027 holds. A hawkish surprise lifts the euro and pressures European equity multiples.
Friday 30 October — EU Q3 GDP flash. Tells you whether the hiking cycle still has growth cover.
The one-year rate one year ahead at four point nine four against a target range mid-point of three point eight seven five is the market pricing roughly four more quarter-point moves. The Fed and the ECB together bracket the AI de-rating question.
Thursday 5 November — Norges Bank. The view is a hold at four and a half per cent. Norges Bank publishes Monetary Policy Reports in March, June, September and December and tends to deliver rate moves at those meetings; interim meetings confirm the path. September HICP at three point four from three point two keeps them in cycle, not enough to force a move at an interim. EUR/NOK stays below eleven while Norges holds its premium.
Friday 6 November — US non-farm payrolls. The first US labour print the Fed sees after the fourth. The labour tape isn't breaking yet; the signal to watch is a payrolls print negative on the month.
This is a discussion, not investment advice.
This episode discusses investing in general. It is not investment advice, it does not describe anyone's positions, and nothing in it is a recommendation to buy or sell anything.
Calls
- Norges Bank holds the key policy rate at 4.5 per cent at the 5 November 2026 meeting. — resolves by 2026-11-05. How we will know: Norges Bank's own press release on 5 November 2026 announcing no change to the key policy rate, carried by Reuters.