1HeadlinesWhat does this mean?
Stage 2 (AI-specific credit stress), unchanged from the last report; earliest downgrade 16 November. AI economics: Elevated, worse.
The bad
- Microsoft terminated its agreement to use capacity at Nscale’s planned Loughton, Essex facility. The project was intended to provide approximately 50MW of capacity using 23,000 Nvidia chips; it lost its publicly announced anchor customer, potentially undermining Nscale’s planned IPO.1
- A wider investigation identified at least 47 major US projects stalled, withdrawn or blocked in 2026, while one analyst estimate suggested only about 35% of planned projects were currently feasible.2
- There was a separate regulatory warning around private-credit risk transparency.3
The good
- Broad markets remained relatively orderly, and there was no confirmed new forced liquidation or systemic funding freeze.4
Not what it looks like
- Microsoft’s withdrawal from the Nscale site looks like a hyperscaler cutting AI spending, but it appeared to be an execution failure rather than a decision to reduce AI expenditure: the grid connection may not be available until the 2030s.1
2ScorecardWhat does this mean?
| Pillar | Status | Trend | Why |
|---|---|---|---|
| AI equities | Elevated | Unchanged | The semiconductor index was 23.9% below its 52-week high at the 16 October close, the latest close available to this report, inside the Elevated band.5 No new equity evidence in this report; status held from the 17 October report. |
| Rates and macro | Elevated | Unchanged | US 10-year at 5.09%, inside the Elevated band.8 No new rates evidence in this report; status held from the 17 October report. |
| Broad credit (the firewall) | Calm | Unchanged | Investment-grade spread 85bp and high-yield 331bp, both inside Calm.134 Broad markets remained relatively orderly.4 |
| AI-specific credit | Stressed | Unchanged | No new AI credit prices in this report; status held from the 17 October report, which recorded CoreWeave’s bond yield briefly above 16% and its CDS near 1,200bp. |
| AI financing flow | Elevated | Unchanged | The Nscale Loughton project lost its publicly announced anchor customer, potentially undermining Nscale’s planned IPO.1 Investors are also demanding higher returns, so new money still arrives but on harder terms. |
| Leverage and liquidity | Elevated | Unchanged | No confirmed new forced liquidation and no systemic funding freeze.4 Status held from the 17 October report. |
| Private credit | Elevated | Unchanged | A separate regulatory warning around private-credit risk transparency, which is context rather than a market event.3 No new evidence of marks or redemptions; status held from the 17 October report. |
| AI economics | Elevated | Worse | AI infrastructure risk is translating into actual customer withdrawals and project disruption. Microsoft terminated its Nscale Loughton agreement, apparently because the grid connection may not be available until the 2030s, and at least 47 major US projects were stalled, withdrawn or blocked in 2026, with one analyst estimating only about 35% of planned projects currently feasible.12 |
3Change since last reportWhat does this mean?
Compared against the 17 October 2026 report, back-scored the same way. Scores for this report were back-scored on 4 November 2026 from its own text and from FRED and Yahoo Finance closing data, because it predates the scorecard format.
| Last report (17 Oct) | This report | Change | |
|---|---|---|---|
| Stage | 2 (evidence 16 Oct) | 2 (evidence 16 Oct) | Unchanged. No new AI credit or forced-selling evidence; the Stage 2 evidence is still dated 16 October. |
| AI economics | Watch | Elevated | Worse. Microsoft’s withdrawal from Nscale’s Loughton site and at least 47 stalled, withdrawn or blocked US projects show the pipeline proving less deliverable than its headline commitments.12 |
| AI equities | Elevated | Elevated | Unchanged. No new market data since the 17 October report; the same 16 October close applies.5 |
| Rates and macro | Elevated | Elevated | Unchanged. No new evidence. |
| Broad credit | Calm | Calm | Unchanged. No new market data; broad markets remained relatively orderly.4 |
| AI-specific credit | Stressed | Stressed | Unchanged. No new evidence in this report. |
| AI financing flow | Elevated | Elevated | Unchanged. Pressure built inside it: a data-centre project lost its anchor customer, potentially undermining Nscale’s planned IPO.1 |
| Leverage and liquidity | Elevated | Elevated | Unchanged. No confirmed new forced liquidation.4 |
| Private credit | Elevated | Elevated | Unchanged. A regulatory warning on private-credit risk transparency, but no new market evidence.3 |
Net: the AI economics axis and pillar moved from Watch to Elevated, because the evidence shifted from the price of AI money to the deliverability of AI projects. The market-priced pillars could not move, since no new close is available after 16 October, and the chain is unchanged.
4Transmission chainWhat does this mean?
- 1 AI equity correction Partial evidence
- 2 AI-specific credit stress Confirmed 16 Oct Earliest downgrade 16 Nov
- 3 Forced deleveraging Partial evidence
- 4 Firewall breached Partial evidence
- 5 Systemic liquidity event Not present
| Stage | Status |
|---|---|
| 1 — AI equity correction | Partial. At the 16 October close, the latest available to this report, the semiconductor index was 23.9% below its 52-week high, its third consecutive close more than 20% down, so the five-close tripwire is not yet met.5 |
| 2 — AI-specific credit stress | Confirmed. Entry evidence last confirmed on 16 October, when the 16 October report recorded data-centre bonds trading substantially below par alongside distressed CoreWeave debt. No new AI credit evidence in this report. Earliest downgrade 16 November. |
| 3 — Forced deleveraging | Partial. Stage 3 lapsed in the 11 October report. This report finds no confirmed new forced liquidation.4 |
| 4 — Firewall breached | Partial. High-yield 331bp and CCC 1,297bp at the 16 October close are wider but far from the tripwires; investment-grade 85bp against a 130bp tripwire.41413 Broad markets remained relatively orderly. |
| 5 — Systemic liquidity event | Not present. No systemic funding freeze.4 |
5JudgementWhat does this mean?
Meaningful new evidence: AI infrastructure risk is translating into actual customer withdrawals and project disruption. Microsoft terminated its agreement to use capacity at Nscale’s planned Loughton, Essex facility, a project of approximately 50MW using 23,000 Nvidia chips, which lost its publicly announced anchor customer and potentially undermined Nscale’s planned IPO.1 Crucially, this appeared to be an execution failure rather than a decision to reduce AI expenditure: the grid connection may not be available until the 2030s. It is a delivery problem, not a hyperscaler cutting capex.
The wider evidence points the same way. At least 47 major US projects were stalled, withdrawn or blocked in 2026, and one analyst estimate suggested only about 35% of planned projects were currently feasible.2 The investment cycle therefore faced two interacting problems: (1) investors demanding higher returns; and (2) delays increasing the capital required before projects can generate revenue. That combination is why the AI economics axis moves to Elevated even though the money is still being committed.
There was also a separate regulatory warning around private-credit risk transparency.3 Broad markets remained relatively orderly, and there was no confirmed new forced liquidation or systemic funding freeze.4 The significant change was that the AI infrastructure pipeline was proving less deliverable than its headline investment commitments implied.
6Key indicatorsWhat does this mean?
Market data as at the close on 16 October 2026, compared with the close used by the 17 October 2026 report (16 October 2026). Reported figures are carried forward from the report that last cited them until a newer one appears.
| Indicator | Latest | Previous | Change | Tripwire | Source |
|---|---|---|---|---|---|
| AI equities | |||||
| PHLX Semiconductor Index (SOX) | 11,137; 23.9% below its 52-week high (22 Jun) | 11,137 | ±0.0% | Stage 1 when more than 20% below its 52-week high for five closes | Confirmed, Yahoo Finance ^SOX5 |
| Nasdaq 100 | 29,188 | 29,188 | ±0.0% | Context | Confirmed, Yahoo Finance ^NDX6 |
| KOSPI | 5,668 | 5,668 | ±0.0% | Context | Confirmed, Yahoo Finance ^KS117 |
| Rates and macro | |||||
| US 10-year Treasury | 5.09% | 5.09% | ±0.00pp | Elevated 4.75–5.25%; Stressed above 5.25%, each for five closes | Confirmed, FRED DGS108 |
| US 30-year Treasury | 5.43% | 5.43% | ±0.00pp | Context | Confirmed, FRED DGS309 |
| Brent crude | $119.33 | $119.33 | ±0.00 | Context | Confirmed, FRED DCOILBRENTEU10 |
| US dollar index (broad) | 121.40 | 121.40 | ±0.0% | Context | Confirmed, FRED DTWEXBGS11 |
| Yen per US dollar | ¥158.00 | ¥158.00 | ±0.00 | Context | Confirmed, FRED DEXJPUS12 |
| Broad credit | |||||
| ICE BofA US investment-grade spread | 85bp | 85bp | ±0bp | Watch at 100bp; Stage 4 above 130bp for five days | Confirmed, FRED BAMLC0A0CM13 |
| ICE BofA US high-yield spread | 331bp | 331bp | ±0bp | Watch at 350bp; Stage 4 above 450bp for five days | Confirmed, FRED BAMLH0A0HYM24 |
| ICE BofA CCC & lower spread | 1,297bp | 1,297bp | ±0bp | Early warning only | Confirmed, FRED BAMLH0A3HYC14 |
| ICE BofA CCC & lower effective yield | 17.66% | 17.66% | ±0.00pp | Early warning only | Confirmed, FRED BAMLH0A3HYCEY15 |
| AI-specific credit | |||||
| AI investment-grade premium over the broad index | ~115bp vs 78bp broad IG (22 Sep report) | ~115bp vs 78bp broad IG | no new reading | n/a | Carried from the 22 September report |
| Oracle 5-year CDS | ~288bp (15 Oct report) | ~288bp | no new reading | n/a | Carried from the 15 October report |
| Meta 5-year CDS | almost 100bp (1 Oct report) | almost 100bp | no new reading | n/a | Carried from the 1 October report |
| CoreWeave 5-year CDS | ~1,200bp (17 Oct report) | ~1,200bp | no new reading | n/a | Carried from the 17 October report |
| SoftBank 5-year CDS | ~424bp (15 Oct report) | ~424bp | no new reading | n/a | Carried from the 15 October report |
| Named data-centre project debt | Hyperion, Sopaipilla and Project Excalibur bonds substantially below par (16 Oct report) | Hyperion, Sopaipilla and Project Excalibur bonds substantially below par | no new reading | n/a | Carried from the 16 October report |
Yahoo Finance ^SOX, daily closes to 16 Oct.
FRED DGS10, daily closes to 16 Oct.
FRED BAMLC0A0CM, daily closes to 16 Oct.
FRED BAMLH0A0HYM2, daily closes to 16 Oct.
Faint vertical lines mark each of the last 20 reports. Dashed red lines are the published thresholds.
What would change our view
Would make it worse (up toward Stage 3): Confirmed forced selling: reported margin calls, fund liquidations, prime brokers withdrawing lines, or prime-brokerage data showing synchronised deleveraging. A systemic funding freeze, or sustained widening of ordinary investment-grade and high-yield spreads, would point further up the chain.
Would make it better (down to Stage 1): A month with no new signs of AI-specific credit stress, with named AI debt back above 90 cents on the dollar; the earliest that can happen is 16 November.
Would move the AI-economics axis: More anchor customers withdrawing from, or hyperscalers cancelling, committed projects would move it further; grid connections and stalled US projects coming back on schedule would ease it.
7SourcesWhat does this mean?
- Reuters — Microsoft pulls out of UK data center deal with Nscale, Telegraph reports (2026-10-18) —
preserved-original-url - Financial Times — US data-centre project delays and execution constraints (2026-10-18) —
preserved-original-url - Reuters — BoE scrutinises Barclays over claim traders misrepresented private credit (2026-10-19) —
preserved-original-url - Federal Reserve Bank of St. Louis (FRED) — ICE BofA US High Yield Index Option-Adjusted Spread (continuous) —
official-data-series - Yahoo Finance — PHLX Semiconductor Index (^SOX) price history (continuous) —
market-data-series - Yahoo Finance — Nasdaq 100 (^NDX) price history (continuous) —
market-data-series - Yahoo Finance — KOSPI (^KS11) price history (continuous) —
market-data-series - Federal Reserve Bank of St. Louis (FRED) — 10-Year Treasury Constant Maturity Rate (continuous) —
official-data-series - Federal Reserve Bank of St. Louis (FRED) — 30-Year Treasury Constant Maturity Rate (continuous) —
official-data-series - Federal Reserve Bank of St. Louis (FRED) — Crude Oil Prices: Brent - Europe (continuous) —
official-data-series - Federal Reserve Bank of St. Louis (FRED) — Nominal Broad U.S. Dollar Index (continuous) —
official-data-series - Federal Reserve Bank of St. Louis (FRED) — Japanese Yen to U.S. Dollar Spot Exchange Rate (continuous) —
official-data-series - Federal Reserve Bank of St. Louis (FRED) — ICE BofA US Corporate Index Option-Adjusted Spread (continuous) —
official-data-series - Federal Reserve Bank of St. Louis (FRED) — ICE BofA CCC & Lower US High Yield Index Option-Adjusted Spread (continuous) —
official-data-series - Federal Reserve Bank of St. Louis (FRED) — ICE BofA CCC & Lower US High Yield Index Effective Yield (continuous) —
official-data-series
Not investment advice. Scoring, tripwires and the persistence rule are set out in How to read the AI Market Watch; terms are explained in the explainers.